* Fixes broken build for the SYCL CUDA backend caused by non-explicit gemm call in outprod (merged in with RWKV6 in
Optimize RWKV6 Operator Naming and Implement Multi-core CPU/ SYCL Acceleration #10133)
* Marks permuted MUL_MAT as unsupported to be able to run test-backend-ops
* Fixes asserts in norm to fix debug builds.
* tests: Fix memory bandwidth calculation for perf tests
Add a flops calculation for flash attention.
Add one GGML_OP_CPY perf test.
* vulkan: Optimize contiguous copies
Add a variant of the copy shader for when the tensors are contiguous. Avoid
the complex addressing calculations, and do four elements per invocation
to hide some other overhead.
Apply similar changes to the scale shader, since scale is always contiguous.
Add a "progress bar" for shader compiles.
Fixes#9582
Spawning too many concurrent copies of glslc leads to "Failed to create pipes"
errors on Linux. This change applies the same throttling we use for
multithreaded pipeline creation.
This change upstreams llamafile's cpu matrix
multiplication kernels for ppc64le using MMA
builtins for FP32 datatype.
This change results in a consistent 90%
improvement in input processing time, and 20%
to 80% improvement in output processing time,
across various batch sizes.
The patch is tested with Meta-Lllama-3-8B,
Mistral-7B, Llama-2-7B-chat-hf models on a
IBM POWER10 machine.
Signed-off-by: Amrita H S <amritahs@linux.vnet.ibm.com>
* metal : opt-in compile flag for BF16
ggml-ci
* ci : use BF16
ggml-ci
* swift : switch back to v12
* metal : has_float -> use_float
ggml-ci
* metal : fix BF16 check in MSL
ggml-ci
* ggml : add ggml_flash_attn_ext_get_prec
* metal : use F16 precision in FA kernels
ggml-ci
* metal : minor clean-up
* metal : compile-guard bf16 FA kernels
ggml-ci
* build : remove obsolete compile flag [no ci]
* metal : prevent int overflows [no ci]
* cuda : disable BF16 FA
ggml-ci
* metal : fix BF16 requirement for FA kernels
ggml-ci
* make : clean-up [no ci]
* rwkv6: rename to wkv6
* rwkv6: support avx2 avx512 armv8 armv9
* rwkv6: update cuda file name
* rwkv6: rename params
* wkv on sycl
* sycl: add some ops
* sycl: Enhance OP support judgment
* wkv6: drop armv9 and tranfer to GGML style
ggml-ci
* sync : ggml
* update the function to use appropriate types
* fix define error
* Update ggml/src/ggml-cpu.c
* add appropriate asserts
* move element-wise functions outside
* put the declaration outside the loop
* rewrite to be more inline with the common pattern for distributing threads
* use recommended way GGML_TENSOR_LOCALS
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
Co-authored-by: Diego Devesa <slarengh@gmail.com>
Co-authored-by: Plamen Minev <pacominev@gmail.com>
Co-authored-by: Yuri Khrustalev <ykhrustalev@users.noreply.github.com>
Co-authored-by: Meng, Hengyu <airdldl@163.com>
* ggml : add initial BF16 support
ggml-ci
* metal : add mul_mat_id BF16 support
ggml-ci
* metal : check for bfloat support on the Metal device
ggml-ci
* metal : better var names [no ci]
* metal : do not build bfloat kernels when not supported
ggml-ci
* metal : try to fix BF16 support check
ggml-ci
* metal : this should correctly check bfloat support
* metal : add quantized FA (vec) support
ggml-ci
* metal : add quantized FA (non-vec) support
* metal : fix support check
ggml-ci
* metal : clean-up
* metal : clean-up (cont)
* metal : fix shared memory calc + reduce smem + comments
* metal : float-correctness
* metal : minor [no ci]
* q6_k instruction reordering attempt
* better subtract method
* should be theoretically faster
small improvement with shuffle lut, likely because all loads are already done at that stage
* optimize bit fiddling
* handle -32 offset separately. bsums exists for a reason!
* use shift
* Update ggml-quants.c
* have to update ci macos version to 13 as 12 doesnt work now. 13 is still x86
* llama : fix buffer checks for mamba and rwk
* llama : fix missing worst case flag during reserve
* cuda : fix supports_op for norm
* disable sched SET_CAUSE
* ggml : fix gguf string leak when reading kv pairs fails
* ggml : avoid crashing with GGML_ABORT when the KV has an invalid type
* ggml : avoid crashing on failed memory allocations when loading a gguf file
* ggml: Add POOL2D OP for GPU ACC to the Vulkan.
- The MobileVLM model now supports inference acceleration through GPU by utilizing the Vulkan backend.
- A GGML_OP_POOL_2D shader has been added. (Pooling)
- The encoding performance of the CLIP model improved from 2.8s on the CPU to 0.7s on the GPU.
Signed-off-by: Changyeon Kim <cyzero.kim@samsung.com>
* [fix] Correct the incorrect order of the parameters.
fix casting to int.
Signed-off-by: Changyeon Kim <cyzero.kim@samsung.com>
---------
Signed-off-by: Changyeon Kim <cyzero.kim@samsung.com>
* Handle objs in Ruby code
* Add task to make Makefile
* Share commont constance in test suites
* Add model-related APIs
* Add Whisper::Model class
* Add tests for Whisper::Model
* Add missing LDFLAG -lstdc++
* Add tests for Whisper.log_set
* Add Whisper.set_log
* Define log level
* Add document on logging
* Add license section to README
* Add document on Whisper::Model
* Fix examples in README
* Add test for Model with GC
* Make dependency on Makefile more accurate
* Fix bug about Whisper::Model and GC
* passing samples_padded by ref to the threads.
* passing samples_padded by ref to the threads.
---------
Co-authored-by: Vinith Misra <physicsdemon@gmail.com>
* metal : support permuted matrix multiplicaions
ggml-ci
* cont : use nb01 directly for row steps
ggml-ci
* cont : add comments [no ci]
* metal : minor refactor
* metal : minor
* [CANN] Adapt to dynamically loadable backends mechanism
* Fix the Bug: inference running result is garbled in debug running model for LM models who's type is Q4_0 class
* Handle the review comments of this pull request
add intel amx isa detection
add vnni kernel for gemv cases
add vnni and amx kernel support for block_q8_0
code cleanup
fix packing B issue
enable openmp
fine tune amx kernel
switch to aten parallel pattern
add error message for nested parallelism
code cleanup
add f16 support in ggml-amx
add amx kernels for QK_K quant formats: Q4_K, Q5_K, Q6_K and IQ4_XS
update CMakeList
update README
fix some compilation warning
fix compiler warning when amx is not enabled
minor change
ggml-ci
move ggml_amx_init from ggml.c to ggml-amx/mmq.cpp
ggml-ci
update CMakeLists with -mamx-tile, -mamx-int8 and -mamx-bf16
ggml-ci
add amx as an ggml-backend
update header file, the old path for immintrin.h has changed to ggml-cpu-impl.h
minor change
update CMakeLists.txt
minor change
apply weight prepacking in set_tensor method in ggml-backend
fix compile error
ggml-ci
minor change
ggml-ci
update CMakeLists.txt
ggml-ci
add march dependency
minor change
ggml-ci
change ggml_backend_buffer_is_host to return false for amx backend
ggml-ci
fix supports_op
use device reg for AMX backend
ggml-ci
minor change
ggml-ci
minor change
fix rebase
set .buffer_from_host_ptr to be false for AMX backend
* fix: use `vm_allocate` to allocate CPU backend buffer on macOS
* fix: switch to `posix_memalign` to keep existing `free()` usages work
* feat: move `GGML_ALIGNED_MALLOC` to `ggml-backend-impl.h`, add support for `vm_allocate` on macOS
* style: formatting
* fix: move const outside of `#ifndef`
* style: formatting
* fix: unused var
* fix: transform `GGML_ALIGNED_MALLOC` and `GGML_ALIGNED_FREE` into functions and add them to `ggml-impl.h`
* fix: unused var
* fix: page align to `GGUF_DEFAULT_ALIGNMENT`
* fix: page align to `TENSOR_ALIGNMENT`
* fix: convert `TENSOR_ALIGNMENT` to a macro
* fix: increase page size to `32` on iOS
* fix: iOS page size
* fix: `hbw_posix_memalign` alignment
* Vectorize load instructions in dmmv f16 CUDA kernel
Replaces scalar with vector load instructions, which substantially
improves performance on NVIDIA HBM GPUs, e.g. gives a 1.27X overall
speedup for Meta-Llama-3-8B-Instruct-F16 BS1 inference evaluation on
H100 SXM 80GB HBM3. On GDDR GPUs, there is a slight (1.01X) speedup.
* addressed comment
* Update ggml/src/ggml-cuda/dmmv.cu
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
---------
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
* ggml : do not use BLAS with types without to_float
* ggml : return pointer from ggml_internal_get_type_traits to avoid unnecessary copies
* ggml : rename ggml_internal_get_type_traits -> ggml_get_type_traits
it's not really internal if everybody uses it
* docs : clarify building Android on Termux
* docs : update building Android on Termux
* docs : add cross-compiling for Android
* cmake : link dl explicitly for Android
* ggml : add metal backend registry / device
ggml-ci
* metal : fix names [no ci]
* metal : global registry and device instances
ggml-ci
* cont : alternative initialization of global objects
ggml-ci
* llama : adapt to backend changes
ggml-ci
* fixes
* metal : fix indent
* metal : fix build when MTLGPUFamilyApple3 is not available
ggml-ci
* fix merge
* metal : avoid unnecessary singleton accesses
ggml-ci
* metal : minor fix [no ci]
* metal : g_state -> g_ggml_ctx_dev_main [no ci]
* metal : avoid reference of device context in the backend context
ggml-ci
* metal : minor [no ci]
* metal : fix maxTransferRate check
* metal : remove transfer rate stuff
---------
Co-authored-by: slaren <slarengh@gmail.com>
* Single allocation of encode_async block with non-ARC capture in ggml-metal.m
* Moving Block_release to the deallocation code
* Release encode block when re-setting encoding buffer count if needed
* Update ggml/src/ggml-metal.m
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
This commit removes the buffer_id field from the leaf_alloc struct.
The motivation for is that this field is only written to and never
read/used as far as I can tell. Each tensor_alloc has a buffer_id field
and this is what caused me to look into this more closely, to
understand what the buffer_id in leaf_alloc was used for.
* Add test for built package existence
* Add more tests for Whisper::Params
* Add more Whisper::Params attributes
* Add tests for callbacks
* Add progress and abort callback features
* [skip ci] Add prompt usage in README
* Change prompt text in example
* Add Params#new_segment_callback= method
* Add tests for Params#new_segment_callback=
* Group tests for #transcribe
* Don't use static for thread-safety
* Set new_segment_callback only when necessary
* Remove redundant check
* [skip ci] Add Ruby version README
* Revert "Group tests for #transcribe"
This reverts commit 71b65b00cc.
* Revert "Add tests for Params#new_segment_callback="
This reverts commit 81e6df3bab.
* Add test for Context#full_n_segments
* Add Context#full_n_segments
* Add tests for lang API
* Add lang API
* Add tests for Context#full_lang_id API
* Add Context#full_lang_id
* Add abnormal test cases for lang
* Raise appropriate errors from lang APIs
* Add tests for Context#full_get_segment_t{0,1} API
* Add Context#full_get_segment_t{0,1}
* Add tests for Context#full_get_segment_speaker_turn_next API
* Add Context#full_get_segment_speaker_turn_next
* Add tests for Context#full_get_segment_text
* Add Context#full_get_setgment_text
* Add tests for Params#new_segment_callback=
* Run new segment callback
* Split tests to multiple files
* Use container struct for new segment callback
* Add tests for Params#new_segment_callback_user_data=
* Add Whisper::Params#new_user_callback_user_data=
* Add GC-related test for new segment callback
* Protect new segment callback related structs from GC
* Add meaningful test for build
* Rename: new_segment_callback_user_data -> new_segment_callback_container
* Add tests for Whisper::Segment
* Add Whisper::Segment and Whisper::Context#each_segment
* Extract c_ruby_whisper_callback_container_allocate()
* Add test for Whisper::Params#on_new_segment
* Add Whisper::Params#on_new_egment
* Assign symbol IDs to variables
* Make extsources.yaml simpler
* Update README
* Add document comments
* Add test for calling Whisper::Params#on_new_segment multiple times
* Add file dependencies to GitHub actions config and .gitignore
* Add more files to ext/.gitignore
* Improve Rakefile
* Remove intermediate files
* Remove unnecessary manipulations from extconf.rb
* Add README and LINCENSE to source files
* Manage ext source files using YAML file
* Use extsources.yaml to include files into gem package file
* Add git-managed source files to build dependency
* Add test task
* Download model for test if not exists
* Add test for build
* Ignore gem package directory
* Enable GitHub action for Ruby binding
* Fix model name
* Build lib file for test
* Use extension for each platform
* Use extension for each platform on testing
* Move built lib file rather than copy
* Add intermediate files to clean targets
* Fixed OpenVino init on state
* Removed an empty line
* Fixed typo
* Replaced tabs with spaces
---------
Co-authored-by: Sandro Hanea <sandrohanea@users.noreply.github.com>
* vulkan : do not use tensor->extra
This patch allows using the Vulkan backend with the RPC backend as
tensor->extra is no longer used.
Ref: #8536
* Adapt GGML_VULKAN_CHECK_RESULTS to extra removal (llama/2)
---------
Co-authored-by: 0cc4m <picard12@live.de>
When the device's warp size is less than 16,
it is possible for loadstride_a (mul_mm.comp:114)
and loadstride_b (mul_mm.comp:115) to be set to 0.
Because they are calculated as: the workgroup size,
multiplied by LOAD_VEC_* (which can be 1) and divided by 16.
And the workgroup size is set to be the same as the
warp/subgroup size.
The loadstride_* variables are used as increments in the
loops that populate the buffers used for the multiplication.
When they are 0 they cause an infinite loop.
But infinite loops without side-effects are UB and the
values of loadstride_* are known at compile time.
So, the compiler quietly optimizes all the loops away.
As a consequence, the buffers are not populated and
the multiplication result is just a matrix with all elements
set to 0.
We prevent the UB by making sure that the workgroup size
will never be less than 16, even if our device has a
smaller warp size (e.g. 8).
Signed-off-by: Salvatore Mesoraca <s.mesoraca16@gmail.com>
* ggml: Added run-time detection of neon, i8mm and sve
Adds run-time detection of the Arm instructions set features
neon, i8mm and sve for Linux and Apple build targets.
* ggml: Extend feature detection to include non aarch64 Arm arch
* ggml: Move definition of ggml_arm_arch_features to the global data section
* ggml : remove assert for AArch64 GEMV and GEMM Q4 kernels
* added fallback mechanism when the offline re-quantized model is not
optimized for the underlying target.
* fix for build errors
* remove prints from the low-level code
* Rebase to the latest upstream
a return before a barrier (that happens only in some threads in
a workgroup) leads to UB.
While the old code actually works on some devices,
it fails on some others (i.e. "smaller" GPUs).
BTW, I think it would be better to set specialization constants
when the graph is built, in that way the local workgroup
could be sized appropriately.
But it would take a lot of work.
Signed-off-by: Salvatore Mesoraca <s.mesoraca16@gmail.com>
* Remove possible leftover ffmpeg temp file from a previous failed conversion
* Revert "Remove possible leftover ffmpeg temp file from a previous failed conversion"
This reverts commit 00797403bd.
* Flag to force ffmpeg to overwrite output file if it exists
The script itself has a hashbang indicating that it is a shell script,
but the README indicates that it must be executed with `bash`.
I checked the script itself, and it seems to be valid POSIX shell. I can
confirm that it works with busybox sh.
Clarify the reference on the README, so it is clear that bash is not
actually a dependency for this script.
Make sure n_barrier and n_barrier_passed do not share the cache line to avoid cache line bouncing.
This optimization shows performance improvements even for n_threads <= 8 cases.
Resurect TSAN (Thread Sanitizer) check so that we can avoid doing expensive read-modify-write
in the normal case and just use thread-fence as originally intended.
* AVX512 version of ggml_gemm_q4_0_8x8_q8_0
* Remove zero vector parameter passing
* Rename functions and rearrange order of macros
* Edit commments
* style : minor adjustments
* Update x to start from 0
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* threadpool: skip polling for unused threads
Currently all threads do N polling rounds even if only 1 thread is active (n_threads_cur == 1).
This commit adds a check to skip the polling for unused threads (ith >= n_threads_cur).
n_threads_cur is now an atomic_int to explicitly tell thread sanitizer that it is written
from one thread and read from other threads (not a race conditions).
* threadpool: further simplify and improve ggml_barrier
Avoid using strict memory order while polling, yet make sure that all threads go through
full memory barrier (memory fence) on ggml_barrier entrace and exit.
* threads: add simple barrier test
This test does lots of small, parallel matmul ops where the barriers in between dominate the overhead.
* threadpool: improve thread sync for new-graphs
Using the same tricks as ggml_barrier. All the polling is done with relaxed memory order
to keep it efficient, once the new graph is detected we do full fence using read-modify-write
with strict memory order.
* threadpool: improve abort handling
Do not use threadpool->ec (exit code) to decide whether to exit the compute loop.
threadpool->ec is not atomic which makes thread-sanitizer rightfully unhappy about it.
Instead introduce atomic threadpool->abort flag used for this. This is consistent with
how we handle threadpool->stop or pause.
While at it add an explicit atomic_load for n_threads_cur for consistency.
* test-barrier: release threadpool before releasing the context
fixes use-after-free detected by gcc thread-sanitizer on x86-64
for some reason llvm sanitizer is not detecting this issue.
* Overlap cmdbuffer creation and cmdbuffer execution in Vulkan backend by submitting smaller cmdbuffers early.
* fix compile issues
* Fix issues where the last submit wasn't executed or handled properly.
* remove trailing whitespace
* Repair GGML_VULKAN_CHECK_RESULTS
* Increase submit counter only if actual work has been submitted and increase submit count to 100.
* Fix some nodes are not checked with GGML_VULKAN_CHECK_RESULTS enabled.
* add check malloc result on device
* update for review comments, check all malloc_device() result
---------
Co-authored-by: arthw <14088817+arthw@users.noreply.github.com>
* Improve Vulkan shader builds system
- Add dependency to vulkan-shaders-gen to rebuild shaders when changing the shader compilation utility.
- Add option to generate debug info for Vulkan shaders to provide shader source to Vulkan shader profiling tools
* remove not required self dependency
* ggml-quants : 1.625 bpw ternary packing for BitNet 1.58b
* ggml-quants : faster 1.625 bpw AVX2 vec_dot
Not using a lookup table anymore makes it match q4_0 speed.
* gguf-py : fix formatting
* llama : remove spaces on empty line
* ggml-quants : subtract 1 when back in epi8
This makes the 1.625 bpw type go faster than q4_0. Still not the fastest.
* ggml-quants : Q2_2 now faster than Q4_K on with AVX2
* ggml-quants : cleanup Q1_3 code formatting
* ggml-quants : ARM NEON vec_dot for q2_2 and q1_3
* ggml-quants : use ceiling division when quantizing q1_3
* convert-hf : simplify BitNet pre-quantization
This still results in the exact same tensor weights and scales,
but it reveals some weirdness in the current algorithm.
* convert-hf : allow converting the weird BitNet 1.3B
Its FFN size is 5460 which is not convenient.
The offending tensors are kept in F16,
which makes the final model 5.01 bpw.
* bitnet : replace 1.58b with b1.58, as in the paper
* ggml-quants : fix build failure on Windows
* ggml-quants : attempt to fix Arm 32-bit support
* ggml : add some informative comments in q1_3 vec_dot
* ggml : add TQ1_0 and TQ2_0 ternary quantization types
* ggml : even faster TQ2_0
* ggml : also faster TQ1_0
Same optimization as for TQ2_0 by offsetting the sum instead of the weights.
This makes TQ1_0 almost as fast as Q8_0 on AVX2.
* ggml : fix build issues in certain environments
* ggml : add NEON vec_dot implementation for TQ1_0 and TQ2_0
* ggml : avoid directly using vmlal_high_s8, for 32-bit ARM compat
The compiler seems smart enough to use the same instruction
even when using vget_high_s8 instead.
* ggml : remove q1_3 and q2_2
No more 1.625 bpw and 2.000 bpw,
now instead using 1.6875 bpw and 2.0625 bpw
with TQ1_0 and TQ2_0, respectively.
* llama : remove the separate scale tensors of BitNet b1.58
They won't be needed, since the remaining ternary quant types have
built-in scales.
* ggml-quants : rename fields of TQ1_0 and TQ2_0 structs for consistency
* ggml-quants : allow using vdotq_s32 in TQ2_0 vec_dot
Not yet tested on hardware which supports it,
might not work or might not even compile. But also it might.
It should make the performance better on recent ARM CPUs.
* ggml-quants : remove comment about possible format change of TQ2_0
Making it slightly more convenient for AVX512
but less convenient for everything else is not worth the trouble.
* gguf-py : Numpy (de)quantization for TQ1_0 and TQ2_0
* ggml-quants : use roundf instead of nearest_int for TQ1_0 and TQ2_0
This does not change anything for ternary models,
since their values should never end up being in halfway cases anyway.
* convert : allow direct conversion to TQ1_0 and TQ2_0
The token embeddings and output tensors are kept in F16
to allow quantizing them to Q4_K and Q6_K with llama-quantize.
* llama : handle fallback for TQ1_0 and TQ2_0 with Q4_0
Q4_0 is not completely symmetric (so not lossless for ternary models),
but it should be good enough.
* ggml-quants : allow using ARM dot product instructions for TQ1_0
* ggml-quants : deduplicate TQ1_0 and TQ2_0 __ARM_FEATURE_DOTPROD support
* ggml : remove unused ggml_mul special case
It would otherwise conflict with the more general
optimization coming with Mamba-2.
* ggml : handle TQ1_0 and TQ2_0 in dequantization-based operators
* test-backend-ops : add TQ1_0 and TQ2_0 comments for later
Not yet adding uncommented, because some backends like SYCL and Metal
do not properly handle unknown types in supports_op for GGML_OP_MUL_MAT.
(and Metal also doesn't handle it with GGML_OP_GET_ROWS)
Support for TQ1_0 and TQ2_0 for other backends than CPU
will be added in follow-up pull requests.
* Add AVX2 based implementations for quantize_q8_0_4x8, ggml_gemv_q4_0_8x8_q8_0 and ggml_gemm_q4_0_8x8_q8_0 functions
* Update code to fix issues occuring due to non alignment of elements to be processed as multiple of 16 in MSVC
* Update comments and indentation
* Make updates to reduce number of load instructions
* Introduce ggml_compute_threadpool
- OpenMP functional: check
- Vanilla ggml functional: Check
- ggml w/threadpool functional: Check
- OpenMP no regression: No glaring problems
- Vanilla ggml no regression: No glaring problems
- ggml w/threadpool no regression: No glaring problems
* Minor fixes
* fixed use after release bug
* fixed a harmless race condition
* Fix Android bulid issue
* fix more race conditions
* fix deadlock for cases where cgraph.n_nodes == 1
and fix --poll case
* threadpool: use cpu_get_num_math to set the default number of threadpool threads
This way we avoid using E-Cores and Hyperthreaded siblings.
* bench: create fresh threadpool for each test
For benchmarking it's better to start a fresh pool for each test with the exact number of threads
needed for that test. Having larger pools is suboptimal (causes more load, etc).
* atomics: always use stdatomics with clang and use relaxed memory order when polling in ggml_barrier
This also removes sched_yield() calls from ggml_barrier() to match OpenMP behavior.
* threadpool: make polling the default to match openmp behavior
All command line args now allow for setting poll to 0 (false).
* threadpool: do not wakeup threads in already paused threadpool
* fix potential race condition in check_for_work
* threadpool: do not create two threadpools if their params are identical
* threadpool: reduce pause/resume/wakeup overhead in common cases
We now start threadpool in paused state only if we have two.
The resume is now implicit (ie new work) which allows for reduced locking and context-switch overhead.
* threadpool: add support for hybrid polling
poll params (--poll, ...) now specify "polling level", i.e. how aggresively we poll before waiting on cond.var.
poll=0 means no polling, 1 means poll for 128K rounds then wait, 2 for 256K rounds, ...
The default value of 50 (ie 50x128K rounds) seems like a decent default across modern platforms.
We can tune this further as things evolve.
* threadpool: reduce the number of barrier required
New work is now indicated with an atomic counter that is incremented for
each new graph that needs to be computed.
This removes the need for extra barrier for clearing the "new_work" and
removes the special case for trivial graphs.
* threadpool: remove special-casing for disposable threadpools
With the efficient hybrid polling there is no need to make disposable pools any different.
This simplifies the overall logic and reduces branching.
Include n_threads in debug print for disposable threadpool.
Declare pause and stop flags as atomic_bool
This doesn't actually generate any memory barriers and simply informs
the thread sanitizer that these flags can be written & read by different
threads without locking.
* threadpool: do not clear barrier counters between graphs computes (fixes race with small graphs)
This fixes the race condition with very small graphs where the main thread happens to
start a new graph while the workers are just about to exit from barriers.
* threadpool: use relaxed order for chunk sync
Full memory barrier is an overkill for this since each thread works on different chunk
* threadpool: remove abort_callback from threadpool state
* threadpool: better naming for thread/cpumask releated functions
* threadpool: consistent use of int type for n_threads params
* threadpool: add support for ggml_threadpool_params_default/init
Also removes the need for explicit mask_specified param.
all-zero cpumask means use default (usually inherited) cpu affinity mask.
* threadpool: move typedef into ggml.h
* threadpool: fix apply_priority() function name
* threadpool: fix swift wrapper errors due to n_threads int type cleanup
* threadpool: enable --cpu-mask and other threadpool related options only if threadpool is enabled
* threadpool: replace checks for compute_thread ret code with proper status check
* threadpool: simplify threadpool init logic and fix main thread affinity application
Most of the init code is now exactly the same between threadpool and openmp.
* threadpool: update threadpool resume/pause function names
* threadpool: enable openmp by default for now
* threadpool: don't forget to free workers state when omp is enabled
* threadpool: avoid updating process priority on the platforms that do not require it
On Windows we need to change overall process priority class in order to set thread priorities,
but on Linux, Mac, etc we do not need to touch the overall process settings.
* threadpool: update calling thread prio and affinity only at start/resume
This avoids extra syscalls for each graph_compute()
* llama-bench: turn threadpool params into vectors, add output headers, etc
* llama-bench: add support for cool off between tests --delay
This helps for long running tests on platforms that are thermally limited (phones, laptops, etc).
--delay (disabled by default) introduces the sleep for N seconds before starting each test.
* threadpool: move process priority setting into the apps (bench and cli)
This avoids changing the overall process priority on Windows for the apps
that use ggml/llama.cpp directy.
* threadpool: move all pause/resume logic into ggml
* threadpool: futher api cleanup and prep for future refactoring
All threadpool related functions and structs use ggml_threadpool prefix.
* threadpool: minor indent fixes
* threadpool: improve setprioty error message
* Update examples/llama-bench/llama-bench.cpp
Co-authored-by: slaren <slarengh@gmail.com>
* threadpool: fix indent in set_threadpool call
* use int32_t for n_thread type in public llama.cpp API
* threadpool: use _new and _free instead of _create and _release
* fix two more public APIs to use int32_t for n_threads
* build: set _GNU_SOURCE for Adroid
---------
Co-authored-by: Max Krasnyansky <quic_maxk@quicinc.com>
Co-authored-by: fmz <quic_fzaghlou@quic.com>
Co-authored-by: Max Krasnyansky <max.krasnyansky@gmail.com>
Co-authored-by: slaren <slarengh@gmail.com>
the old code was trying to print a non-existent field (size)
and the struct as a whole (which doesn't have a operator<<
override defined).
Probably a typo happened during refactoring.
Signed-off-by: Salvatore Mesoraca <s.mesoraca16@gmail.com>
sin and cos failed test-backend-ops because they
tried to dereference a context pointer that is null
on dry runs.
This commit prevents that segfault.
Signed-off-by: Salvatore Mesoraca <s.mesoraca16@gmail.com>
test-backend-ops fails because ggml_cont aborts
when invoked passing an unsupported type.
This commit makes ggml_cont tests pass
Signed-off-by: Salvatore Mesoraca <s.mesoraca16@gmail.com>
* tests: add gradient checking to test-backend-ops
* remove old comment
* reorder includes
* adjust SIN/COS parameters
* add documentation, use supports_op if possible
* Fixed go cuda bindings building
* Added note to go bindings Readme to build using cuda support
* Added temperature bindings for Go
---------
Co-authored-by: Binozo <entwickler@binozoworks.de>
* Added libsdl2-dev for SDL samples
Building talk-llama seems to fail here as there is no sdl.h.
* Adding libsdl2-dev for sdl.h
* Adding libsdl2-dev for sdl.h
Update Go version to ^1.23, the actions/setup-go
to v5, actions/checkout to v4 and
github.com/stretchr/testify to v1.9.0.
Add test cases for the following model
struct methods:
- New
- Close
- NewContext
- IsMultilingual
- Languages
Add test cases for the following context
struct methods:
- SetLanguage
- IsMultilingual
- Language
- Process
* Fixed go cuda bindings building
* Added note to go bindings Readme to build using cuda support
---------
Co-authored-by: Binozo <entwickler@binozoworks.de>
Depending on the OS the lib dir can vary, on Fedora for instance it is
"${prefix}/lib64". Instead of hard-coding the directory name, let CMake fill
this variable for us.
* ggml_cont: fix issue with transposed tensors when one dimension is 1
when using multiple threads, it is not enough
to check for the tensors to be contiguous for
ggml_compute_forward_dup_same_cont to work correctly.
The tensors strides also need to match.
Signed-off-by: Salvatore Mesoraca <s.mesoraca16@gmail.com>
* Add ggml_cont tests
Signed-off-by: Salvatore Mesoraca <s.mesoraca16@gmail.com>
* Remove dead code
it isn't possible to reach this code because
all these functions are invoked by ggml_compute_forward_dup
if and only if src0->type != dst->type
Signed-off-by: Salvatore Mesoraca <s.mesoraca16@gmail.com>
* Make ggml_compute_forward_dup_same_cont work with contiguous tensors
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
Signed-off-by: Salvatore Mesoraca <s.mesoraca16@gmail.com>
---------
Signed-off-by: Salvatore Mesoraca <s.mesoraca16@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update README.md
Fix broken C-style API link
* Update whisper_processor.py
Update examples/python/whisper_processor.py to remove nonexistent flag "-np" from subprocess.Popen call.
* Add pywhispercpp to the Pybind11 Python wrapper list
abdeladim-s/pywhispercpp wasn't added to the list / was removed at some point (?)
It was referenced in issue #9, so I feel like it's worthy of being added as it's the first if not one of the first Python wrappers for whisper.cpp
* llama : advanced batch splits
This includes equal-sequence-length batch splits which are useful
to simplify recurrent model operators.
* llama : always make recurrent state slots contiguous
* ggml : simplify mamba operators
* llama : fix integer signedness mixing
* llama : logits_all has priority over batch->logits
Otherwise, the server embeddings tests failed.
This was likely an existing problem but was only detected here
because of an additional assertion.
* llama : apply suggestions
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* llama : fix t5 segfault
* llama : fix Mamba session save and restore
* llama : minor cosmetic changes
* llama : rename llama_reorder_outputs to llama_output_reorder
Also move it closer to llama_output_reserve.
* llama : fix pooled embeddings when using batches with equal_seqs
* minor : add struct members for clarity
ggml-ci
* llama : fix T5 segfault again
* llama : fix Mamba pooled embeddings with multiple sequences
Until the pooled embeddings are refactored to allow splitting
across ubatches for causal embeddings,
recurrent models can only process a single sequence per ubatch
when calculating pooled embeddings.
* llama : add llama_model_is_recurrent to simplify figuring that out
This will make it easier to more cleanly support RWKV-v6 and Mamba-2.
* llama : fix simple splits when the batch contains embeddings
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* fallback mmvq to mul_mat
* mmvq in cuda path
* Update ggml/src/ggml-sycl.cpp
Co-authored-by: Alberto Cabrera Pérez <alberto.cabrera@codeplay.com>
---------
Co-authored-by: Alberto Cabrera Pérez <alberto.cabrera@codeplay.com>
* ggml : move rope type enum to ggml.h
This commit moves the `llama_rope_type` enum from `llama.h` to
`ggml.h` and changes its name to `ggml_rope_type`.
The motivation for this change is to address the TODO in `llama.h` and
use the enum in ggml.
Note: This commit does not change the `mode` parameter to be of type
`enum ggml_rope_type`. The name `mode` and its usage suggest that it
might be more generic and possibly used as a bit field for multiple
flags. Further investigation/discussion may be needed to determine
if `mode` should be restricted to RoPE types.
* squash! ggml : move rope type enum to ggml.h
This commit removes GGML_ROPE_TYPE_NONE and GGML_ROPE_TYPE_GLM from
ggml.h, and back the llama_rope_type enum.
I've kept the assert for GGML_ROPE_TYPE_GLM as I'm not sure if it is
safe to remove it yet.
* squash! ggml : move rope type enum to ggml.h
This commit removes the enum ggml_rope_type from ggml.h and replaces it
with a define (GGML_ROPE_TYPE_NEOX). This define is used in the code to
check if the mode is set to GPT-NeoX. Also the enum llama_rope_type has
been updated to reflect this change.
* squash! ggml : move rope type enum to ggml.h
This commit contains a suggestion enable the GGML_ROPE_TYPE_NEOX
macro/define to be passed to the shader compiler.
* squash! ggml : move rope type enum to ggml.h
This commit fixes the editorconfig-checker warnings.
* squash! ggml : move rope type enum to ggml.h
Update comment for ggml_rope function.
* Revert "squash! ggml : move rope type enum to ggml.h"
This reverts commit 6261222bd0dc0efd51f0fb0435ad3f16a5b52fd6.
* squash! ggml : move rope type enum to ggml.h
Add GGML_ROPE_TYPE_NEOX to rope_common.comp.
* remove extra line
---------
Co-authored-by: slaren <slarengh@gmail.com>
* Optimize Vulkan backend for better CPU performance and less GPU synchronization overhead.
- Allocation overhead for the temporary std::vectors was easily detectable with a sampling profiler and simple to remove.
- ggml_vk_sync_buffer introduce a full pipeline sync which has a significant cost on the GPU side, sometimes larger than the actual kernel execution. Adding only barriers for shader read/writes and transfers seems to be sufficient looking at the code which either launches compute kernels or copies tensors.
* Fix small typo
---------
Co-authored-by: 0cc4m <picard12@live.de>
This change updates the -pc flag, so that a new xterm256 color scheme is
used. This color scheme is believed to be better for three reasons:
1. It should be friendlier to the colorblind. The scheme was designed by
Paul Tol (see: https://personal.sron.nl/~pault/). TensorBoard uses it
since 2017, so it's already popular in the machine learning community
2. It should appear to be the same colors as before to people who aren't
i.e. it's still a red-green spectrum like before but lightly modified
3. It is readable in both white and black background terminals. The neon
colors before were probably a bit too intense for white backgrounds.
* ggml: support forward pass broadcasting in ggml_sub
Signed-off-by: Salvatore Mesoraca <s.mesoraca16@gmail.com>
* Use assert instead of GGML_ASSERT in ggml_compute_forward_sub_f32
The check is already performed in ggml_sub_impl
Signed-off-by: Salvatore Mesoraca <s.mesoraca16@gmail.com>
---------
Signed-off-by: Salvatore Mesoraca <s.mesoraca16@gmail.com>
* ggml : reading the runtime sve config of the cpu
* change to one time init to prevent performance drop
* prefix variable to avoid possible conflicts
* revert xxhash fix and add brackets
---------
Co-authored-by: domke <673751-domke@users.noreply.gitlab.com>
* add truncate_bf16
* truncate intermediate fp32 if converting bf16 to bf16
* fix masking in __compute_fp32_to_bf16
* np.int16 no longer used
* missing cast and additional numpy 2.x fix
* ggml-impl : do not flush bf16 subnormals to zero
* ggml : add reference fp32 to bf16 conversion
The fast version is no longer equivalent for all platforms
because of the handling of subnormal values.
* gguf-py : remove flush to zero for bf16 subnormals
* gguf-py : remove float32 truncation to bf16
Rounding achieves the same thing in the cases where this was used.
* missed prototype update in merge
* merge cleanup
---------
Co-authored-by: Francis Couture-Harpin <git@compilade.net>
* Adding support for unified memory
* adding again the documentation about unified memory
* refactoring: Moved the unified memory code in the correct location.
* Fixed compilation error when using hipblas
* cleaning up the documentation
* Updating the documentation
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
* adding one more case where the PR should not be enabled
---------
Co-authored-by: matteo serva <matteo.serva@gmail.com>
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
In these codes, we want to retain the value that they previously held
when mask[i] is false. So we should use undisturbed. With the default
agnostic policy of rvv intrinsic, these values can be held or be
written with 1s.
Co-authored-by: carter.li <carter.li@starfivetech.com>
* Fix Vulkan repeat op
* Implement Vulkan concat op
* Delete old Vulkan shader generator
* Implement Vulkan im2col op
* Implement Vulkan unary gelu_quick op
* Implement Vulkan group_norm op
* Implement Vulkan timestep_embedding op
* Implement Vulkan upscale op
* Fix Vulkan vk_context tensor extra index issue
* Fix Vulkan matmul shader parameter bug
* Properly fix Vulkan matmul shader parameter bug
* Add Vulkan ADD f16 + f32 -> f16 operator support
* Implement Vulkan tanh op
* Fix Vulkan group count too large Validation error on non-Nvidia GPUs
* Throw error when too much memory is requested
* Fix another Vulkan group count too large Validation error on non-Nvidia GPUs
* Fix matmul MMQ condition
* Implement Vulkan pad op
* Fix Vulkan crash when tensor is used multiple times in a compute graph
* Add Vulkan CONCAT f16 + f16 -> f16 op
* Add Vulkan LEAKY_RELU op
This commit moves the comment for the c parameter from ggml_rope to
ggml_rope_ext. The comment is currently incorrect as ggml_rope does not
have a c parameter (freq_factors tensor).
Signed-off-by: Daniel Bevenius <daniel.bevenius@gmail.com>
`ggml_init` can fail if no unused context is found. In that case, a NULL-pointer deref will happen later in the code during a call to `ggml_set_on_alloc`.
This fixes it by bailing out if no context is found.
The check gating the use of `__builtin_amdgc_sdot4` specifically checks for gfx1030. This causes a severe perf regression for anything gfx103? that's not gfx1030 and not using `HSA_OVERRIDE_GFX_VERSION` (if you've built ROCm to support it). We already have a generic RDNA2 define, let's use it.
* [CANN] Add Ascend NPU backend
Ascend is a full-stack AI computing infrastructure for industry
applications and services based on Huawei Ascend processors and
software.
CANN (Compute Architecture of Neural Networks), developped by
Huawei, is a heterogeneous computing architecture for AI.
Co-authored-by: wangshuai09 <391746016@qq.com>
* delete trailing whitespaces
* Modify the code based on review comment
* Rename LLAMA_CANN to GGML_CANN
* Make ggml-common.h private
* add ggml_cann prefix for acl funcs
* Add logging for CANN backend
* Delete Trailing whitespace
---------
Co-authored-by: wangshuai09 <391746016@qq.com>
* lora: load to devide buft
* add patch tensor function
* correct tensor patch
* llama_lora_adapter_apply
* correct ggml_backend_tensor_copy
* add llm_build_mm
* fix auto merge
* update based on review comments
* add convert script
* no more transpose A
* add f16 convert
* add metadata check
* add sanity check
* fix ftype
* add requirements
* fix requirements
* fix outfile
* conversion: only allow selected models
* fix types
* cuda : do not use dmmv if the tensor does not have enough cols
* llama : lora fixes
* do not disable mmap with lora
Co-authored-by: slaren <slarengh@gmail.com>
* llm_build_lora_mm_id
* convert_lora : MoE LoRA conversion support
* convert_lora : prefer safetensors, similarly to convert_hf
* convert_hf : simplify modify_tensors for InternLM2
* convert_lora : lazy conversion
* llama : load and use alpha from LoRA adapters
* llama : use llm_build_lora_mm in most model graphs
* auto scale
* Revert "auto scale"
This reverts commit 42415a4874e0f963e4aca6796ea5dfb97cd17464.
* remove redundant params
* Apply suggestions from code review
Co-authored-by: slaren <slarengh@gmail.com>
* change kv metadata
* move add_type to __init__
* convert_hf : move add_type to main()
* convert_lora : use the GGUFWriter from Model instead of overwriting it
---------
Co-authored-by: slaren <slarengh@gmail.com>
Co-authored-by: Francis Couture-Harpin <git@compilade.net>
* cuda : suppress 'noreturn' warn in no_device_code
This commit adds a while(true) loop to the no_device_code function in
common.cuh. This is done to suppress the warning:
```console
/src/ggml-cuda/template-instances/../common.cuh:346:1: warning:
function declared 'noreturn' should not return [-Winvalid-noreturn]
346 | }
| ^
```
The motivation for this is to reduce the number of warnings when
compilng with GGML_HIPBLAS=ON.
Signed-off-by: Daniel Bevenius <daniel.bevenius@gmail.com>
* squash! cuda : suppress 'noreturn' warn in no_device_code
Update __trap macro instead of using a while loop to suppress the
warning.
Signed-off-by: Daniel Bevenius <daniel.bevenius@gmail.com>
---------
Signed-off-by: Daniel Bevenius <daniel.bevenius@gmail.com>
* Arm AArch64: optimized GEMV and GEMM kernels for q4_0_q8_0, and q8_0_q8_0 quantization
* Arm AArch64: add optimized GEMV and GEMM asm kernels for q4_0_q8_0 quantization and refactor code to address llama.cpp pr#5780 suggestions
* Arm AArch64: add optimized GEMV and GEMM asm kernels for q4_0_q8_0 quantization and refactor code to address llama.cpp pr#5780 suggestions
* Arm AArch64: add optimized GEMV and GEMM asm kernels for q4_0_q8_0 quantization and refactor code to address llama.cpp pr#5780 suggestions
* Arm AArch64: add optimized GEMV and GEMM asm kernels for q4_0_q8_0 quantization and refactor code to address llama.cpp pr#5780 suggestions
* Arm AArch64: add copyright claim only to ggml-aarch64.cpp and ggml-aarch64.h files
* Arm AArch64: minor code refactoring for rebase
* Arm AArch64: minor code refactoring for resolving a build issue with cmake
* Arm AArch64: minor code refactoring to split the Q4_0_AARC64 type into three separate types: Q4_0_4_4, Q4_0_4_8, and Q4_0_8_8
* Arm AArch64: minor code change for resolving a build issue with server-windows
* retrigger checks
* Arm AArch64: minor code changes for rebase
* Arm AArch64: minor changes to skip the pr#7433 vec_dot code for arm cpus with SVE VL not equal to 256 bits
* Arm AArch64: remove stale LLAMA_QKK_64 from CMakeLists.txt and delete build.zig
* Arm AArch64: add reference scalar gemm and gemv, and avoid dynamic memory allocations during quantization for Q4_0_4_4, Q4_0_4_8, and Q4_0_8_8
* Arm AArch64: add multithreaded quantization support for the new types: Q4_0_4_4, Q4_0_4_8, and Q4_0_8_8
* Arm AArch64: minor code refactoring
* Arm AArch64: simplify logic for calling gemm and gemv functions in ggml_compute_forward_mul_mat
* Arm AArch64: minimize changes in ggml_compute_forward_mul_mat
* Arm AArch64: minor code refactoring, and add reference scalar code to quantize routines for new quant types
* Arm AArch64: minor code refactoring
* Arm AArch64: minor code refactoring
* Arm AArch64: minor code refactoring
* rebase on the latest master commit 3fd62a6 and adapt to the new directory structure
* Arm AArch64: remove a redundant comment
* Arm AArch64: add pragma in ggml-aarch64.c to turn -Woverlength-strings warning off
* Arm AArch64: use __aarch64__ check to guard 64-bit neon kernels
* Arm AArch64: update docs/build.md README to include compile time flags for buiilding the Q4_0_4_4 quant type
Apply a loop tiling technique to the generic path, which provides
performance upside for ISAs with enough registers to take advantage
of it. Also helps the compiler optimize this path.
* Add support for float16 tensors in 1d pooling operations
* Add support for float16 input tensors in 2d pooling operations
* code cleanup
remove unnecessary casting during srow ptr initialization
---------
Co-authored-by: vanaka11 <vanaka1189@gmail.com>
This prevents invalid frees when destroying a partially initialized
vk_buffer_struct. For example, this could happen in ggml_vk_create_buffer
when running out of device memory.
Co-authored-by: Tony Wasserka <neobrain@users.noreply.github.com>
This commit removes an UNUSED macro call that is not needed as the
variable n0 is used in the code and will not produce a warning.
Signed-off-by: Daniel Bevenius <daniel.bevenius@gmail.com>
* conv transpose 1d passing test for 1d input and kernel
* working for different input and output channel counts, added test for variable stride
* initial draft appears to work with stride other than 1
* working with all old and new conv1d tests
* added a test for large tensors
* removed use cuda hardcoding
* restored test-conv-transpose.c
* removed unused arugments, and fixed bug where test failure would cause subsequent tests to fail
* fixed accumulator bug
* added test to test-backend-ops
* fixed mistake
* addressed review
* fixed includes
* removed blank lines
* style and warning fixes
* return failure when test fails
* fix supports_op
---------
Co-authored-by: slaren <slarengh@gmail.com>
* fix compile issues introduced by loongarch_asx
* restore quant changes to merge
* fix compile issues introduced by loongarch_asx
* further optimize by using vec_msum & vec_sum4s on ppc64le
* separate DPCT helpers outside
* replace global variables with context
* remove useless extra
* update mul_mat condition
* remove duplicate buft initialization
* remove duplicate extra and global work group size
* remove useless backend check
* remove duplicated extras
* use macro for group_size and remove cuda-related
* move BLAS to a separate backend
* rename GGML_USE_OPENBLAS to GGML_USE_BLAS
* alloc : reuse same buffer when the same buffer type if used multiple times
* set number of threads automatically for openblas and blis
* sched : print assignments when GGML_SCHED_DEBUG env variable is set
* sched : allow ops with weights on an incompatible buffer type
This will cause the weight to be copied to a backend that supports the
op, which is very costly. The weight should have been stored in a buffer
of a backend that can run the op, but llama.cpp cannot do this
automatically at the moment.
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update Vulkan RoPE implementation
* Return nullptr on alloc_buffer when allocation fails, instead of throwing an exception
Minor fixes
* Fix segfault when running out of VRAM
Co-authored-by: slaren <slarengh@gmail.com>
---------
Co-authored-by: slaren <slarengh@gmail.com>
* vulkan : reuse parent extra for views
* Fix validation error when multiple compute contexts are used in a graph
---------
Co-authored-by: 0cc4m <picard12@live.de>
Previously the code would have failed to cope in the case that the
number of nodes changes in an existing CUDA graph. This fixes the
issue by removing an unnecessary conditional.
This enforces a check that -fno-finite-math-only was set and that the operating
compiling mode is not in finite maths mode. This is because during rewriting of
silu and softmax for cpu #7154 there emerged an issue where the result that was
observed when >1 slot was nondeterministic as found by @JohannesGaessler.
@LostRuins narrowed the problem down to -ffinite-math-only which was theorised
to be due to SiLU, instead of flushing small values to 0, returns NaN or some
other garbage. @jart proposed a fix that @ggerganov then implemented in this fix
ref https://github.com/ggerganov/llama.cpp/pull/7154#issuecomment-2145661825
* llama : offload to RPC in addition to other backends
* - fix copy_tensor being called on the src buffer instead of the dst buffer
- always initialize views in the view_src buffer
- add RPC backend to Makefile build
- add endpoint to all RPC object names
* add rpc-server to Makefile
* Update llama.cpp
Co-authored-by: slaren <slarengh@gmail.com>
---------
Co-authored-by: slaren <slarengh@gmail.com>
* ggml: Added OpenMP for multi-threads processing
* ggml : Limit the number of threads used to avoid deadlock
* update shared state n_threads in parallel region
* clear numa affinity for main thread even with openmp
* enable openmp by default
* fix msvc build
* disable openmp on macos
* ci : disable openmp with thread sanitizer
* Update ggml.c
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
---------
Co-authored-by: slaren <slarengh@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
op_getrows_f32 is required since https://github.com/ggerganov/llama.cpp/pull/6122
for the Vulkan w/ Kompute backend to be functional.
As such, implement this op to make this backend functional again.
* ggml : fix loongson compile warnings
ggml-ci
* Fix loongarch quantize test fail.
Fix unexpected error introduced during rebase code.
* tests : disable json test due to lack of python on the CI node
ggml-ci
---------
Co-authored-by: junchao-loongson <zhaojunchao@loongson.cn>
* update HIP_UMA #7399
add use of hipMemAdviseSetCoarseGrain when LLAMA_HIP_UMA is enable.
- get x2 on prompte eval and x1.5 on token gen with rocm6.0 on ryzen 7940HX iGPU (780M/gfx1103)
* simplify code, more consistent style
---------
Co-authored-by: slaren <slarengh@gmail.com>
CUDA graphs require parameter updates to kernels associated with
GGML_OP_CPY nodes. Previously the implementation only checked for a
single CUDA kernel in such nodes, but this caused a bug in cases where
2 such kernels exist. This fixes the issue by using a vector to allow
multiple function pointers to be stored and checked against.
Fixes#7942
* cuda : fix rope pos data
ggml-ci
* ggml : drop mode & 1 == 1 support for ggml_rope
ggml-ci
* ggml : support freq_factors for f16 rope (CPU)
ggml-ci
* tests : add rope tests using frequency factors
ggml-ci
* add phi3 128k support in convert-hf-to-gguf
* add phi3 128k support in cuda
* address build warnings on llama.cpp
* adjust index value in cuda long rope freq factors
* add long rope support in ggml cpu backend
* make freq factors only depend on ctx size
* remove unused rope scaling type 'su' frin gguf converter
* fix flint warnings on convert-hf-to-gguf.py
* set to the short freq factor when context size is small than trained context size
* add one line of comments
* metal : support rope freq_factors
* ggml : update ggml_rope_ext API to support freq. factors
* backends : add dev messages to support rope freq. factors
* minor : style
* tests : update to use new rope API
* backends : fix pragma semicolons
* minor : cleanup
* llama : move rope factors from KV header to tensors
* llama : remove tmp assert
* cuda : fix compile warning
* convert : read/write n_head_kv
* llama : fix uninitialized tensors
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* android : use "ci-android" branch for CI
* ggml : disable SIMD exp and silu for 32-bit ARM
ggml-ci
* android : do not fetch, use add_subdirectory instead
* cmake : provide binary dir
This change upstreams llamafile's vectorized expf() functions. This lets
us compute softmax and silu more accurately than the short[65536] lookup
table that GGML previously used to make this operation go faster. We can
support aarch64 and sse2+ with the worst case rounding error of 2ulp. It
makes make -j8 tests && ./tests/test-backend-ops -o SOFT_MAX -b CPU perf
go 1.5x faster for SSE2+FMA, 1.9x faster for AVX2+FMA and 2.1x on AVX512
* logging: add proper checks for clang to avoid errors and warnings with VA_ARGS
* build: add CMake Presets and toolchian files for Windows ARM64
* matmul-int8: enable matmul-int8 with MSVC and fix Clang warnings
* ci: add support for optimized Windows ARM64 builds with MSVC and LLVM
* matmul-int8: fixed typos in q8_0_q8_0 matmuls
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* matmul-int8: remove unnecessary casts in q8_0_q8_0
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Just reordering some structs.
* Adding in the calls to mm_pause
* Passing around the state
* Renaming and moving a bunch of variables around.
* Extracting the logic to it's own function.
* Moving some variable definitions into the chunk function.
* Moving some variables around
* moving src1_cont inside
* Moving row_size
* adding the current_chunk
* Reorg the code.
* Formatting to match the orig patch
* starting to setup the chunking variables
* Starting the buildup of the loop
* The yield shouldn't be necessary.
* adding the looping structure based on the chunk configuration.
* Add in the re-chunking code.
* Making it much more likely to rechunk.
* disable resizing if numa is enabled.
* Updating comments with what we've learned.
* Fix formatting
* Couple more formatting fixes.
* More style fixes.
* Fix Warnings
* Going with unused because there's conditional logic that needs it.
* Update ggml.c
* Update ggml.c
---------
As discussed in PR #6766, CUDA graphs were being disabled in the presence of long prompts.
This fixes the issue by avoiding the consective update counter from incrementing unnecessarily
for tokens in which cuda graphs are disabled due to batch size > 1.
* initial commit with CPU implementation of upscale to shape and test, cuda implementation next
* experimental commit to see if dst shape is correct
* test version
* test
* removed unnecessary params
* refactor
* fixed tests
* ggml : metal impl + cleanup + sycl dev warnings
* patched ggml_upscale cuda op to handle non-contiguous tensors, added test for non-contiguous behavior
* metal : fix upsacle op to support nb00 + style
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* cuda : fix bounds check for src0 rows in MMVQ kernel
* Update ggml-cuda/mmvq.cu
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
---------
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
- search for ffmpeg libs/headers at cmake time
- added ffmpeg-transcode.cpp into libcommon if ffmpeg on
- hooked ffmpeg trancoding in common read_wav(...)
- passed test:
./main -m ggml-base.en.bin -f samples/jfk.mp3
* Extended talk-llama example to reject runs without required arguments.
Print warning and exit if models are not specified on the command line.
* Update examples/talk-llama/talk-llama.cpp
* Update examples/talk-llama/talk-llama.cpp
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
Add two options:
```
-tp, --temperature N [0.00 ] The sampling temperature, between 0 and 1
-tpi, --temperature-inc N [0.20 ] The increment of temperature, between 0 and 1
```
The sampling temperature, between 0 and 1. Higher values like 0.8 will
make the output more random, while lower values like 0.2 will make it
more focused and deterministic. If set to 0, the model will use log
probability to automatically increase the temperature until certain
thresholds are hit.
Signed-off-by: Daniel Ziegenberg <daniel@ziegenberg.at>
* optimize for ppc64le using VSX intrinsics
* 1. code clean up by removing comments about overflow concern.
2. fix typo in suffix of scaling.
* Continue to fix typo in suffix of scaling for QK_K <> 256
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Modify mat mat mul shader for mul_mat_id, modify mat vec mul shaders for single call batch operation
* Further work towards MoE, disabled for now
* Disable MoE code (not ready yet), fix a number of bugs in shaders and Vulkan code
* Add softmax with f16 mask and pos buffer support
* Disable mul_mat_id shaders for now
* Fix flake8
* Fix validation errors caused by empty buffers on larger batch sizes
* DRAFT: Introduction of CUDA Graphs to LLama.cpp
* FIx issues raised in comments
* Tidied to now only use CUDA runtime (not mixed with driver calls)
* disable for multi-gpu and batch size > 1
* Disable CUDA graphs for old GPU arch and with env var
* added missing CUDA_CHECKs
* Addressed comments
* further addressed comments
* limit to GGML_ALLOW_CUDA_GRAPHS defined in llama.cpp cmake
* Added more comprehensive graph node checking
* With mechanism to fall back if graph capture fails
* Revert "With mechanism to fall back if graph capture fails"
This reverts commit eb9f15fb6fcb81384f732c4601a5b25c016a5143.
* Fall back if graph capture fails and address other comments
* - renamed GGML_ALLOW_CUDA_GRAPHS to GGML_CUDA_USE_GRAPHS
- rename env variable to disable CUDA graphs to GGML_CUDA_DISABLE_GRAPHS
- updated Makefile build to enable CUDA graphs
- removed graph capture failure checking in ggml_cuda_error
using a global variable to track this is not thread safe, but I am also not safistied with checking an error by string
if this is necessary to workaround some issues with graph capture with eg. cuBLAS, we can pass the ggml_backend_cuda_context to the error checking macro and store the result in the context
- fixed several resource leaks
- fixed issue with zero node graphs
- changed fixed size arrays to vectors
- removed the count of number of evaluations before start capturing, and instead changed the capture mode to relaxed
- removed the check for multiple devices so that it is still possible to use a single device, instead checks for split buffers to disable cuda graphs with -sm row
- changed the op for checking batch size to GGML_OP_ADD, should be more reliable than GGML_OP_SOFT_MAX
- code style fixes
- things to look into
- VRAM usage of the cudaGraphExec_t, if it is significant we may need to make it optional
- possibility of using cudaStreamBeginCaptureToGraph to keep track of which ggml graph nodes correspond to which cuda graph nodes
* fix build without cuda graphs
* remove outdated comment
* replace minimum cc value with a constant
---------
Co-authored-by: slaren <slarengh@gmail.com>
* fix: use `malloc` instead of `posix_memalign` in `ggml-metal.m` to make it not crash Electron proccesses
* fix: typo
* fix: use `vm_allocate` instead of `posix_memalign`
* fix: don't call `newBufferWithBytesNoCopy` with `NULL` when `ggml_metal_host_malloc` returns `NULL`
* fix: use `vm_allocate` only on macOS
* Introduce bfloat16 support
Many models on Hugging Face (e.g. Mistral, TinyLLaMA) use bfloat16 as
their canonical floating point format.
┌sign
│
│ ┌exponent
│ │
│ │ ┌mantissa
│ │ │
│┌──┴───┐┌─┴───┐
0b0000000000000000 brain16
This encoding has the same number of exponent bits as float32. That
makes conversion relatively straightforward, even in the absence of
hardware support. For example, converting brain16 to binary32 means
simply shifting 16 bits to the left.
┌sign
│
│ ┌exponent
│ │
│ │ ┌mantissa
│ │ │
│┌──┴───┐┌─┴───────────────────┐
0b00000000000000000000000000000000 IEEE binary32
The issue is that converting bf16 to fp16 can result in information
loss. Only 13% of bf16 numbers can be precisely represented in fp16
which in practice ends up being 99.71% of Mistral 7b v0.2's weights
however there is currently no way other than fp32 to get the others
┌sign
│
│ ┌exponent
│ │
│ │ ┌mantissa
│ │ │
│┌─┴─┐┌─┴──────┐
0b0000000000000000 IEEE binary16
This change fixes that, by adding a bf16 data type to GGML. Support
for CPU inference has been implemented along with optimizations for
the AVX2, AVX512, and AVX512BF16 ISAs. Perplexity on Mistral 7b 0.2
improves somewhere around -0.0024 to -0.0046 compared to using fp16
* Remove GGML code that's not needed
* Minimize the GGML API surface area for BF16
* Remove bf16 luts
* Make the GGML header look nicer
* Fix documentation
* Apply ggerganov's fixes for test-backend-ops
* Add BF16 code for new ggml_validate_row_data() function
* ggml : add ggml_flash_attn_ext API
* ggml : fix GQA support in ggml_flash_attn_ext
* ggml : online attention (CPU)
* metal : initial implementation
* metal : f16 precision
* metal : reduce branches
* metal : specialize for head size
* wip : 8 rows per simd group
* wip : 4 rows per simd group
* wip : template for rows per warp
* metal : parallelize across KV size
* metal : parallel reduce across heads
* metal : efficient flash_attn_f16 implementation
* metal : avoid redundant loads of the attention
* metal : scale and mask in matrix form
* metal : fix comment
* llama : avoid ggml_cast, use F32 query
* metal : add parallel reduce version (disabled)
* metal : move output into local memory + optimize
- the result from each simdgroup now stays in the registers
- significantly reduced SRAM usage
- more efficient skipping of -INF blocks
- avoid simdgroup barrier in hot loop
- add comments
* metal : add tests, fix scaling, support C > 32
* metal : improve precision
* ggml : fix f16 mad
* metal : minor
* metal : support Q > 8
* tests : add ATTN tests
* metal : disable buffer allocation logs
* tests : more
* metal : faster inner loop for C == 32
* metal : fix array initialization
* tests : ifdef
* ggml : switch to padded F16 mask for ggml_soft_max, ggml_flash_attn_ext
* ggml : fix ggml_soft_max mask requirement
* cuda : fix soft_max to use correct mask size
* cuda : add flash_attn kernel (wip)
* metal : optimize softmax for C > 32
* metal : optimize softmax
* tests : minor fix
* cuda : avoid zeroing fragments
* tests : update dims
* cuda : fix __hisinf() result check
* cuda : avoid warp_reduce for smax
* cuda : use int instead of int64_t
Noticeably improves performance (thanks to Johannes)
* cuda : make loops use the same loop values
Thanks Johannes again for the tip
* cuda : unroll some of the loops
* cuda : avoid __hisinf branches
* cuda : use half2 in softmax
* cuda : switch to 1 warp for bs > 16
* cuda : speed-up reduce part of the kernel
* cuda : unroll Q*K^T loop
* cuda : fix -INF block check
* cuda : simplify softmax
* cuda : fix matrix names
* cuda : minor
* llama : adapt to F16 KQ_pos
* llama : adapt new models to F16 KQ_mask
* ggml : fix F16 store (ARM NEON)
* llama : fix type of KQ_mask and KQ_pos
* ggml : fix CPU soft_max
* tests : add hs=256
* cuda : fix build
* metal : improve perf via smaller int registers
* cuda : adapt soft_max to F16 mask and pos
* CUDA: faster FlashAttention, kernel for bs == 1
* 16 cols for Phi-2
* no vec for hs, no hs==256 ncols==32 for Volta
* adjust kernel selection logic
* 4 warps, 256 stride for all D
* no ncols == 64
* Multiple parallel blocks for batch size 1
* fix compile warnings
* fix excessive KQ_b loads
* fix cmake build
* fix KV cache padding, NaN from INFINITY (llama/6438)
* llama : flash_attn cparam + fix defrag
* server: support flash_attn param
* server: bench: enable flash_attn param
* CUDA: refactor host code, dyn. par. blocks
* fix flash_attn_vec_f16 race condition
* flush softmax exp below threshold to 0
* store temp KQ in registers
* Calculate KQ as FP32 if KQV has GGML_PREC_F32
* Add __hgt2_mask implementation for CUDA 11
* fix KQ FP32 precision fpr parallel_blocks > 1
* llama-bench : add -fa,--flash-attn arg
* metal : add BS=1 kernel for flash attention (llama/6508)
* metal : add BS=1 kernel for flash attention (wip)
* metal : support more than 1 warps
* metal : opts
* metal : opt
* metal : switch to parallel reduce
* metal : reduce registers
* metal : simplify
* metal : initial FA vec kernel
* metal : use F32 attention accumulators
* batched-bench : add fattn arg
* llama : simplify llama_build_kv_store
ggml-ci
* llama : adapt build_olmo to changes
* ggml : fix arm fp16 store on windows
* metal : clean-up
* metal : clean-up kernel code
* metal : minor
* tests : remove benchmarks
ggml-ci
* ggml : fix avx512 const correctness
ggml-ci
* ggml : fix soft_max with bias on CPU
ggml-ci
* common : print --flash-attn in help
* ggml : fix num dimensions in ggml_flash_attn_ext
* llama : force disable flash attention for incompatible models
* ggml : ggml_soft_max support F16/F32 mask/pos
ggml-ci
* cuda : uint -> uint32_t
* cuda : "constexpr dim3" -> "const dim3"
ggml-ci
* cuda : try to fix __hgt2_mask
ggml-ci
* ggml : add TODO's for F16/F32 mask/pos support in other backends
* llama : replace bool need_kq_pos with use_alibi
* llama : prep ALiBi support for BERT models
ggml-ci
* llama : fix n_batch requirements
ggml-ci
* cont
* server : add help for --flash-attn arg
* llama : disable FA for AMD
* tests : remove TMP_ATTN_BENCH
ggml-ci
* llama : support save/load state with FA enabled
ggml-ci
* ci : add CUDA save-load-state tests
ggml-ci
* llama : llama_kv_cache_clear zeroes data + fix save-load seq
ggml-ci
* llama : fix copy-paste errors, add TODO
* llama : disallow incompatible states
* llama : update llama_state_get_size after v_trans field
* metal : remove tmp log
* llama : add static reminder for llama_state_get_size
* metal : fix max nsg
ggml-ci
* ci : fix arg order
ggml-ci
---------
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
Co-authored-by: Pierrick HYMBERT <pierrick.hymbert@gmail.com>
* add basic tensor data validation function
* add --check-tensors command line argument
tensor validation is disabled by default and can be enabled by adding
`--check-tensors` to the command line arguments.
quantize always validates tensors.
* always use calloc
clamp n_kv on failure to read a kv
* ggml : alternative ctx->header.n_kv update
---------
Co-authored-by: slaren <slarengh@gmail.com>
* llamafile : improve sgemm.cpp
- Re-enable by default
- Fix issue described in #6716
- Make code more abstract, elegant, and maintainable
- Faster handling of weirdly shaped `m` an `n` edge cases
* Address review comments
* Help clang produce fma instructions
* Address review comments
Latest gcc complains here:
/home/airlied/devel/llama.cpp/ggml-alloc.c: In function ‘ggml_gallocr_new_n’:
/home/airlied/devel/llama.cpp/ggml-alloc.c:374:59: warning: ‘calloc’ sizes specified with ‘sizeof’ in the earlier argument and not in the later argument [-Wcalloc-transposed-args]
374 | ggml_gallocr_t galloc = (ggml_gallocr_t)calloc(sizeof(struct ggml_gallocr), 1);
| ^~~~~~
/home/airlied/devel/llama.cpp/ggml-alloc.c:374:59: note: earlier argument should specify number of elements, later size of each element
and a bunch more.
calloc is specified to take nmemb first then size, so realign the code.
In a couple of places there was a * x, 1 so I fixed those to use calloc properly.
* ggml : group all experts in a single ggml_mul_mat_id
cuda : improve mmid row copy
* cuda : fix bin bcast with non-cont src0
* test-backend-ops : only run all mul mat tests for base types
* llama : disable moe offloading with SYCL
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
This change upstreams llamafile's cpu matrix multiplication kernels
which improve image and prompt evaluation speed. For starters, Q4_0
and Q8_0 weights should go ~40% faster on CPU. The biggest benefits
are with data types like f16 / f32, which process prompts 2x faster
thus making them faster than quantized data types for prompt evals.
This change also introduces bona fide AVX512 support since tinyBLAS
is able to exploit the larger register file. For example, on my CPU
llama.cpp llava-cli processes an image prompt at 305 tokens/second,
using the Q4_K and Q4_0 types, which has always been faster than if
we used f16 LLaVA weights, which at HEAD go 188 tokens/second. With
this change, f16 LLaVA performance leap frogs to 464 tokens/second.
On Intel Core i9-14900K this change improves F16 prompt perf by 5x.
For example, using llama.cpp at HEAD with Mistral 7b f16 to process
a 215 token prompt will go 13 tok/sec. This change has fixes making
it go 52 tok/sec. It's mostly thanks to my vectorized outer product
kernels but also because I added support for correctly counting the
number of cores on Alderlake, so the default thread count discounts
Intel's new efficiency cores. Only Linux right now can count cores.
This work was sponsored by Mozilla who's given permission to change
the license of this code from Apache 2.0 to MIT. To read more about
what's improved, and how it works, see: https://justine.lol/matmul/
* support qwen2moe
* fix-review
* metal : support unary ops for nelements % 4 != 0
* metal : require contiguousness for float4 unary kernels
* metal : require contiguousness for float4 unary kernels (cont)
* fix-review
* names : for brevity "SHARED_EXP" -> "SHEXP"
* llama : reuse build_moe_ffn()
* llama : add model type name
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* disable mmap to fix memcpy crash, add missed cmd in guide, fix softmax
* refactor to disable mmap for SYCL backend
* fix compile error in other os
* refactor the solution, use host buf to fix it, instead of disable mmap
* keep to support mmap()
* use host buff to reduce malloc times
* revert to malloc/free solution, for threaad safe
* Remove split metadata when quantize model shards
* Find metadata key by enum
* Correct loop range for gguf_remove_key and code format
* Free kv memory
---------
Co-authored-by: z5269887 <z5269887@unsw.edu.au>
* cmake : make WHISPER_NO_AVX512=ON disable all subsets of AVX-512
Previously it happened only for MSVC, but it makes sense to have the
same behavior for other compilers too.
* make : reorder x86 ISA extensions in chronological order
And update compiler flags at the end to ease modifying conditions.
* make : support WHISPER_NO_AVX512=1 for disabling all AVX-512 subsets.
That way you do not have to override each AVX-512 subset setting
individually if it has been turned on during autodetection.
fft_out needs to be twice the frame_size, not the frame_step. It is resized in fft() anyway, but this change prevents an unnecessary reallocation.
n_fft must match the mel filter size, so it is best not to calculate it from the framesize.
We only need to get the magnitudes for half the spectrum since the other half is a mirror and not used in the mel filter loop later.
* make : add AVX512 detection to Makefile and CMakeLists.txt
* make : autodetect more AVX512 instruction subsets
* cmake : do not default to AVX512, must be enabled explicitly
* cmake : enable a set of AVX512 subsets, when AVX512 is turned on
* make : consolidate AVX512 subsets, add AVX512 VBMI
* cmake : revert to NO AVX512 setting, add settings for AVX512 VNNI and VBMI
* make : re-introduce AVX512VNNI back
* cmake : remove superfluous comment line
* Add Command R Plus GGUF
* Add Command R Plus GGUF
* Loading works up to LayerNorm2D
* Export new tensors in 1D so they are not quantized.
* Fix embedding layer based on Noeda's example
* Whitespace
* Add line
* Fix unexpected tokens on MPS. Re-add F16 fix. ((Noeda)
* dranger003: Fix block index overflow in CUDA dequantizing.
* Reverted blocked multiplication code as it still has issues and could affect other Llama arches
* export norms as f32
* fix overflow issues during quant and other cleanup
* Type convention
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* dranger003: Fix more int overflow during quant.
---------
Co-authored-by: S <seast@Ss-Mac-Studio.local>
Co-authored-by: S <s@example.com>
Co-authored-by: slaren <slarengh@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* The "main" example now allows a response-file as the sole parameter.
A response-file is a text file with command-line parameters, one per line.
Prefix the name of the response-file with "@" to identify it as such.
It's used under MS Windows to work around command-line length limits.
It may be useful under other platforms to simplify character-escaping.
* minor : style
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Allow a regular expression to describe tokens to suppress.
Example: --suppress-tokens-re "[,\.]|[ ]?[0-9]+" will suppress commas, periods, and numeric tokens.
Technique inspired by https://github.com/openai/whisper/discussions/1041
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Blind change to fix Java test.
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* ggml : update mul_mat_id to use the same tensor for all the experts
* update cuda
* minor
* update metal
* update test-backend-ops
* fix cuda
* Update ggml-metal.m
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* update convert.py
* update convert-hf-to-gguf.py
* update convert.py for mixtral hf models
* Update convert-hf-to-gguf.py
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* cuda : support non-pow-2 number of experts
* allow quantize to work for split and merged experts models in the same way
* cleanup + disable mmap automatically with split tensors models
* update imatrix
* test-backend-ops : test qwen argsort
* update grok model loading
* llama : add merged experts tensors to the grok tensor map
* minor
* gguf : bump version
* fix quantizing of merged experts
* convert-hf-to-gguf.py : update grok (untested)
* make linter happy
* cuda/argsort : use shared memory instead of pool memory
* convert : fix grok tensor names
* metal : add support for non-pow-2 argsort
* llama : more loader cleanup, better error checking
* cuda : fix warning
* llama : still use mmap for loading old models, but copy the data to a host buffer
* add review note
* llama : remove ffn tensor counting + add sanity check
ggml-ci
* convert : fix handling of n_experts == None
ggml-ci
* imatrix : fix ncall counters
* llama : produce error if imatrix size does not match
* quantize : terminate on errors + trace logs
ggml-ci
* metal : pad shared memory to 16 bytes
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Fix Vulkan no kv offload incoherence
* Add k-quant mul mat mat shaders
* Rework working buffer allocation, reduces vram use noticeably
Clean up cpu assist code, replaced with ggml-backend offload function
* Default to all dedicated GPUs
* Add fallback for integrated GPUs if no dedicated GPUs are found
* Add debug info which device is allocating memory
* Fix Intel dequant issue
Fix validation issue
* Fix Vulkan GGML_OP_GET_ROWS implementation
* Clean up merge artifacts
* Remove Vulkan warning
* make : use pkg-config for finding CFLAGS & LDFLAGS needed by OpenBLAS
That way building on *nix like environments (including MSYS2 on Windows)
with WHISPER_OPENBLAS=1 works out of the box.
Fix handling of WHISPER_OPENBLAS, so that empty value or 0 won't be
misinterpreted by make as enabled. Mind that it's not intended to
detect CMake false constants (OFF NO FALSE N). make is not CMake.
By default OpenBLAS with 64-bit interface is used, but that can be
changed with `WHISPER_OPENBLAS_INTERFACE64=0` if 32-bit one is desired.
If OpenBLAS headers and library are respectively in include/ and lib/
subdirectories of given path, then you can specify it, e.g.
`OPENBLAS_PATH=/usr/local/openblas`, and this will take precedence over
any pkg-config file.
If there is no pkg-config file (.pc) for OpenBLAS and OPENBLAS_PATH is
empty, then headers are assumed to be in /usr/include/openblas and
library as assumed to be called 'openblas64' (or 'openblas' if
`WHISPER_OPENBLAS_INTERFACE64=0`). If different headers location should
be used, then it can be done, e.g.
`WHISPER_BLAS_CFLAGS=-I/usr/local/include/openblas`.
If different library should be used, it can be specified, e.g.
`WHISPER_BLAS_LIB=openblasp64` (pthreads version as seen on Fedora), or
you can provide LDFLAGS needed to link with OpenBLAS directly:
`WHISPER_BLAS_LDFLAGS="-L/usr/local/lib/openblas -lopenblas64"`.
Current solution is flexible enough to handle most cases out there
without needlessly hardcoding possible OpenBLAS installation details.
* cmake : fix how pkg-config is used for finding include dirs and libraries needed by OpenBLAS
That way building on *nix like environments (including MSYS2 on Windows)
with -DWHISPER_OPENBLAS=ON should work out of the box as long as you
have CMake 3.25 or newer.
Make OPENBLAS_PATH environment variable supported not only on Windows.
It sets OpenBLAS include dir to ${OPENBLAS_PATH}/include and library to
${WHISPER_BLAS_LIB} (name without prefixes and suffixes) in
${OPENBLAS_PATH}/lib and avoids further package finding.
By default OpenBLAS with 64-bit interface is used (equivalent to setting
`-DWHISPER_BLAS_LIB=openblas64`), but that can be changed with
`-DWHISPER_OPENBLAS_INTERFACE64=OFF` (equivalent to setting
`-DWHISPER_BLAS_LIB=openblas`) if 32-bit one is desired.
Turn on BLA_STATIC for FindBLAS only when WHISPER_STATIC is enabled.
BLA_STATIC may not work as expected for pkg-config based operation.
Get rid of supporting BLAS_HOME environment variable. If OPENBLAS_PATH
is insufficient in your case, there is no pkg-config file to rely on,
then you can manually specify include dir, e.g.
`-DBLAS_INCLUDE_DIRS=/usr/local/include/openblas`, and library, e.g.
`-DBLAS_LIBRARIES=/usr/local/lib/libopenblas.so`.
* make / cmake : use OpenBLAS with 32-bit interface by default.
OpenBLAS w/o INTERFACE64=1 vel USE_64BITINT=1 seems to be more common.
* cmake : hardcode "lib" prefix for OpenBLAS lib filename (even on Windows)
* cmake : hardcode OpenBLAS library name when building in MSVC (Windows)
Most *nix like environments (including MSYS2 on Windows) have OpenBLAS
packages that allow coexistence of OpenBLAS builds with 32-bit and
64-bit interface (w/o and w/ OPENBLAS_USE64BITINT defined) and they
differ by not having or having "64" suffix in their library filenames.
That's not the case for OpenBLAS prebuilt libraries for Windows.
* Implemented command-style grammar in the main example.
Mostly just copied the relevant parts from the command example.
* main : code style
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* whisper.cpp: impl dtw algo
* WIP: producing and placing DTW timestamps on tokens
* Fix compile and assertion errors. Attempt to DTW timestamp with single_segment=false.
* Fix mistake causing incorrect alignment of dtw timestamps
* implement N_TOP_MOST and CUSTOM alignment heads setting
* whisper: fix typo on alignment heads enum
* Fix issues related to changes in whisper.cpp
* Fixed excessive memory use when using DTW timestamps. Other minor fixes to DTW timestamping function
* decoder: save cross QKs only if requested
* Calling median filter with ggml_map_custom1
* Reimpl aheads n_top_most and custom. Sanity checks on chosen aheads
* Copying cross QKs from decoder backend correctly
* dtw: cleanup
* Fix incorrect n_frames passed to dtw when near end of audio
* Fix aheads_masks_init for backend != CPU
* whisper : minor style
* main : add dtw (wip)
* whisper: fix invalid memory access in aheads_masks_init
* main : add dtw (cont)
* whisper : minor
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* llama : add pipeline parallelism support for batch processing with multiple CUDA GPUs
ggml-ci
* server : add -ub, --ubatch-size parameter
* fix server embedding test
* llama : fix Mamba inference for pipeline parallelism
Tested to work correctly with both `main` and `parallel` examples.
* llama : limit max batch size to n_batch
* add LLAMA_SCHED_MAX_COPIES to configure the number of input copies for pipeline parallelism
default increase to 4 (from 2)
changing this value may improve performance for some systems, but increases memory usage
* fix hip build
* fix sycl build (disable cpy_tensor_async)
* fix hip build
* llama : limit n_batch and n_ubatch to n_ctx during context creation
* llama : fix norm backend
* batched-bench : sync after decode
* swiftui : sync after decode
* ggml : allow ggml_get_rows to use multiple threads if they are available
* check n_ubatch >= n_tokens with non-casual attention
* llama : do not limit n_batch to n_ctx with non-casual attn
* server : construct batch with size of llama_n_batch
* ggml_backend_cpu_graph_compute : fix return value when alloc fails
* llama : better n_batch and n_ubatch comment
* fix merge
* small fix
* reduce default n_batch to 2048
---------
Co-authored-by: Francis Couture-Harpin <git@compilade.net>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* iq1_s: we can do even better
Spent one of the 4 scale bits on a signs of a 0.125 shift.
I.e., quants are now -1 + delta, delta, 1 + delta, where delta
is +/- 0.125.
CUDA works, same performance as before.
PPL(LLaMA-v2-7B) is now 11.85!
* iq1_s: make scalar and AVX2 work with the new version
* iq1_s: make Neon work with new version.
~10% drop in performance, so will need some more work.
* iq1_s: make Metal work with new version
* iq1_s: very slightly faster dequantize on Metal
* iq1_s: fix dequantize on the CPU
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* windows arm ci
* fix `error C2078: too many initializers` with ggml_vld1q_u32 macro for MSVC ARM64
* fix `warning C4146: unary minus operator applied to unsigned type, result still unsigned`
* fix `error C2065: '__fp16': undeclared identifier`
* Trying blocvks of 16 for IQ1_S - seems slightly better
* iq1s_blocks16: Adjust scale fudge factor to 1.125
* iq1s_blocks16: going to blocks of 32
with 2048 lattice points, so same bpw.
This is even better than blocks of 16.
Should I try blocks of 64? But to keep the same
bpw, when I go to 4096 lattice points, I need to
remove blocks alltogether and just have superblocks of
256 weights.
* iq1s_blocks16: Use 2*<x^2> as sigma2 in weight adjustment
* iq1s_blocks16: scalar and AVX2 dot products
* iq1s_blocks16: CUDA dot product
* iq1s_blocks16: Metal works, Neon does not
Metal works but TG is dog slow (35 t/s). PP is OKish (493 t/s).
Not seeing the bug in the Neon implementation for now.
* iq1s_blocks16: fixed Neon
* iq1s_blocks16: very slightly faster TG on Metal
Still pathetic at 37 t/s
* iq1s_blocks16: speedup Metal by packing codebook into uint32_t's
* Formatting
* iq1s_blocks16: uint32_t codebook is also better in CUDA
TG-128 is now 204 t/s up from 194 t/s.
PP-512 is 5890 t/s, so significantly better than other quants
* iq1s_blocks16: slightly faster Neon dot product
* iq1s_blocks16: faster AVX2 dot product
* iq1s_blocks16: adjust to ggml-common.h
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* mamba : begin working on support for Mamba SSM
* mamba : begin figuring out how to (ab)use the kv cache for Mamba
* mamba : recurrent inference almost works, but incoherent
* mamba : recurrent inference WORKS!!!
* convert : optionally use d_conv and d_state from config.json for Mamba
* mamba : refactor recurrent conv, resulting in 20% perf increase
It's still slower than I'd like, but I did not really optimize `ggml_exp` yet.
I also refactored `ggml_exp` to work with tensors with more than 2 dimensions.
* ggml : parallelize ggml_exp
This results in 8% faster token generation for Mamba-130M.
* mamba : simplify the conv step with a self-overlapping view
Turns out the conv_state can be made smaller by one column.
Note that this breaks existing GGUFs of Mamba,
because the key_value_length field is tied to the conv_state size.
Convolution with a self-overlapping view is cool!
And it's much simpler than what I initially thought would be necessary
to make the convolution step work with more than 1 token at a time.
Next step is to make the SSM step work on batches of tokens too,
and thus I need to figure out a way to make a parallel selective scan
which will keep the ssm_state small and won't make it bigger
by a factor of (n_layer * batch_size).
* llama : fix Mamba KV self size wrongly displaying as f16 instead of f32
Relatedly, I also tried to see if other types than f32 worked for the states,
but they don't, because of the operators used.
It's probably better anyway to keep lots of precision there,
since the states are small anyway.
* mamba : fix self-overlapping view depth stride
* mamba : handle batches of more than 1 token
This means running Mamba no longer crashes when using the default settings!
And probably also slightly faster prompt processing.
Both batched and non-batched processing yield the same output.
Previously, the state was not cleared when starting a sequence.
Next step is to make the KV cache API work as expected for Mamba models.
* ggml: add ggml_ssm_scan to help with parallel selective scan
If the selective scan was implemented without a custom operator,
there would be waaay too many nodes in the graph. For example,
for Mamba-130M, with a batch size of 512 (the default),
a naive selective scan could add at least 24*512=12288 nodes,
which is more than LLAMA_MAX_NODES (8192),
and that's only for the smallest Mamba model.
So it's much cleaner with a custom operator.
Not sure about the name, though.
* ggml : in ggml_ssm_scan, merge multiple rows in the same vec operation
This will help with performance on CPU if ggml_vec_mul_f32
and ggml_vec_add_f32 are ever optimized with SIMD.
* mamba : very basic quantization support
Mostly works, but there is currently no difference
between the variants of a k-quant (e.g. Q4_K_S and Q4_K_M are the same).
Most of the SSM-specific weights can be kept in f32 without affecting
the size that much, since they are relatively small.
(the linear projection weights are responsible for most of Mamba's size)
Too much quantization seems to make the state degrade quite fast, and
the model begins to output gibberish.
It seems to affect bigger models to a lesser extent than small models,
but I'm not sure by how much.
Experimentation will be needed to figure out which weights are more important
for the _M (and _L?) variants of k-quants for Mamba.
* convert : fix wrong name for layer norm weight of offical Mamba models
I was using Q-bert/Mamba-* models before, which have a slighlty different
naming scheme for the weights.
(they start with "model.layers" instead of "backbone.layers")
* mamba : fuse more steps of the SSM scan in the ggml_ssm_scan operator
This increases performance on CPU by around 30% for prompt processing,
and by around 20% for text generation.
However, it also makes the ggml_exp and ggml_soft_plus operators unused.
Whether or not they should be kept will be decided later.
* convert : for Mamba, also consider the "MambaLMHeadModel" arch name
It's the name of the class of the official implementation,
though they don't use it (yet) in the "architectures" field of config.json
* mamba : fix vocab size problems with official models
The perplexity was waaaay to high for models with a non-round vocab size.
Not sure why, but it needed to be fixed in the metadata.
Note that this breaks existing GGUF-converted Mamba models,
but **only if** the vocab size was not already rounded.
* ggml : remove ggml_exp and ggml_soft_plus
They did not exist anyway outside of this branch,
and since ggml_ssm_scan fused operations together, they are unused.
It's always possible to bring them back if needed.
* mamba : remove some useless comments
No code change.
* convert : fix flake8 linter errors
* mamba : apply suggestions from code review
* mamba : remove unecessary branch for row-wise ssm_state and C multiplication
It was previously done to avoid permuting when only one token is processed
at a time (like when generating text), but permuting is cheap,
and dynamically changing the compute graph is not future-proof.
* ggml : in ggml_ssm_scan, use more appropriate asserts
* ggml : rename the destination pointer in ggml_compute_forward_ssm_scan_f32
* mamba : multiple sequences, but one at a time
This is a step towards making this Mamba implementation usable
with the server example (the way the system prompt is kept when clearing
the client slots will need to be changed before this can work, though).
The KV cache size for this kind of model is tied to the maximum number
of sequences kept at any single time.
For now, this number is obtained from n_parallel (plus one,
to have an extra sequence to dedicate to the system prompt),
but there might be a better way to do this which won't also
make the main example use 2 cells even if only 1 is really used.
(for this specific case, --parallel 0 helps)
Simultaneous sequence processing will probably require changes to
ggml_ssm_scan, and possibly a new operator for the conv step.
* mamba : support llama_kv_cache_seq_cp
This (mis)uses the logic around K shifts, because tokens in a state
can't be shifted anyway, and because inp_K_shift has the right shape and type.
Using ggml_get_rows is a nice way to do copies, but copy chains can't work.
Fortunately, copy chains don't really seem to be used in the examples.
Each KV cell is dedicated to the sequence ID corresponding to its own index.
* mamba : use a state mask
It's cleaner than the previous heuristic of
checking for the pos of the first token in the batch.
inp_KQ_mask could not be re-used for this, because it has the wrong shape
and because it seems more suited to the next step of
simultaneous sequence processing (helping with the problem of
remembering which token belongs to which sequence(s)/state(s)).
* llama : replace the usage of n_ctx with kv_self.size in many places
* mamba : use n_tokens directly instead of n_tok
* mamba : in comments, properly refer to KV cells instead of slots
* mamba : reduce memory usage of ggml_ssm_scan
From 290.37 MiB to 140.68 MiB of CPU compute buffer size
with Mamba 3B with a batch size of 512.
The result tensor of ggml_ssm_scan was previously a big part
of the CPU compute buffer size. To make it smaller,
it does not contain the intermediate ssm states anymore.
Both y and the last ssm state are combined in the result tensor,
because it seems only a single tensor can be returned by an operator
with the way the graph is built.
* mamba : simultaneous sequence processing
A batch can now contain tokens from multiple sequences.
This is necessary for at least the parallel example, the server example,
and the HellaSwag test in the perplexity example.
However, for this to be useful, uses of llama_kv_cache_seq_rm/cp
will need to be changed to work on whole sequences.
* ggml : add ggml_ssm_conv as a new operator for the conv step of Mamba
This operator makes it possible to use and update the correct states
for each token of the batch in the same way as ggml_ssm_scan.
Other solutions which use existing operators would need loops which would
add too many nodes to the graph (at least the ones I thought of).
Using this operator further reduces the size of the CPU compute buffer
from 140.68 MiB to 103.20 MiB with Mamba 3B with a batch size of 512.
And (at least on CPU), it's a bit faster than before.
Note that "ggml_ssm_conv" is probably not the most appropriate name,
and it could be changed if a better one is found.
* llama : add inp_s_seq as a new input tensor
The most convenient implementation to select the correct state (for Mamba)
for each token is to directly get the correct index from a tensor.
This is why inp_s_seq is storing int32_t and not floats.
The other, less convenient way to select the correct state would be
to have inp_KQ_mask contain 1.0f for each state used by a token
and 0.0f otherwise. This complicates quickly fetching the first used
state of a token, and is also less efficient because a whole row
of the mask would always need to be read for each token.
Using indexes makes it easy to stop searching when there are
no more sequences for a token, and the first sequence assigned
is always very quickly available (it's the first element of each row).
* mamba : support llama_kv_cache_seq_cp copy chains
* mamba : support shifting and dividing the kv cache pos
* mamba : make the server and parallel examples work with whole sequences
A seq_id is dedicated to the system prompt in both cases.
* llama : make llama_kv_cache_seq_rm return whether it succeeded or not
* mamba : dedicate an input tensor for state copy indices
This is cleaner and makes it easier to adapt when/if token positions
(and by extension, inp_K_shift) are no longer integers.
* mamba : adapt perplexity, batched, and batched-bench examples
* perplexity : limit the max number of sequences
This adapts to what the loaded model can provide.
* llama : add llama_n_max_seq to get the upper limit for seq_ids
Used by the perplexity example.
* batched : pass n_parallel to the model's context params
This should have been there already, but it wasn't.
* batched-bench : reserve sequences to support Mamba
* batched-bench : fix tokens being put in wrong sequences
Generation quality isn't what's measured in there anyway,
but at least using the correct sequences avoids using non-consecutive
token positions.
* mamba : stop abusing attention metadata
This breaks existing converted-to-GGUF Mamba models,
but will allow supporting mixed architectures like MambaFormer
without needing to break Mamba models.
This will also allow changing the size of Mamba's states
without having to reconvert models in the future.
(e.g. using something else than d_conv - 1 columns for the conv_states
will not require breaking existing converted Mamba models again)
* gguf-py : add new KV metadata key-value pairs for Mamba
* llama : add new metadata key-value pairs for Mamba
* llama : guard against divisions by zero when n_head is 0
* mamba : rename "unlimited" KV cache property to "recurrent"
* mamba : more correctly update the "used" field of the KV cache
* ggml : in ggml_ssm_scan, use a threshold for soft_plus
This is how the official Mamba implementation does it,
and it's also what torch.nn.Softplus does.
* convert : for Mamba, fallback to internal NeoX tokenizer
The resulting models are exactly the same
as if the tokenizer.json and tokenizer_config.json of GPT-NeoX were there.
* mamba : support state saving and restoring
* ggml : implicitly pass src tensors through dst for Mamba-related ops
* mamba : clarify some comments
* server : fix cache_tokens not getting correctly resized
Otherwise, when the "we have to evaluate at least 1 token" special case
was triggered, an extra token was kept in cache_tokens even if it was
removed from the KV cache.
For Mamba, this caused useless prompt reprocessing when the previous
request triggered the above case.
* convert-hf : support new metadata keys for Mamba
For the models available at
https://huggingface.co/collections/state-spaces/transformers-compatible-mamba-65e7b40ab87e5297e45ae406
* mamba : rename metadata to be more similar to transformers library
This breaks existing converted-to-GGUF models,
but the metadata names are more "standard".
* mamba : support mamba-*-hf models
These models share their token_embd.weight with their output.weight
* mamba : add missing spaces
This is purely a formatting change.
* convert-hf : omit output.weight when identical with token_embd.weight
Only for Mamba for now, but it might be relevant for other models eventually.
Most Mamba models actually share these two tensors, albeit implicitly.
* readme : add Mamba to supported models, and add recent API changes
* mamba : move state_seq and state_mask views outside layer loop
A few tensors were also missing `struct` in front of `ggml_tensor`.
As of #1486, whisper.cpp uses a unified KV cache with KQ masking.
As a result, depending on their location in the batch,
identical sequences in a batch can have slightly different outputs
due to floating point rounding errors during reduction.
See the discussion in #1941 for more details.
The beam search code used "has identical sum of log probabilities"
as a shorthand for "is an identical token sequence". However, per above,
identical tokens do not necessarily result in identical probabilities.
Instead, explicitly compare on sequences.
This is linear in cost when they are identical,
but the lengths are always small and the comparisons are cheap.
This increases diversity during beam search.
This improves output quality for some short samples I've been working
with, at no detectable performance cost.
I haven't checked against larger corpuses.
Fixes#1941
All else being otherwise equal, this encourages the beam candidate
selection to re-use the same decoder, which slightly
reduces the cache size.
I wouldn't expect it to make much of a performance difference,
but it helps when debug printing the cache and beam.
Added as part of understanding #1941.
* fix mul_mat fault in cpy_f32_f16
* rm unused function
* add wait() for memcpy
* restore ci/run.sh, rename struct defination, fix bug in ggml_sycl_op_mul_mat_sycl
* fix format issue
* llama : fix segfault from unknown model arch name (llama/5820)
* llama : fix segfault from unknown model arch name
* llama : make all LLM maps const
This also requires using `std::map::at` instead of its `operator[]`
which does not exist for const maps.
* llama : name LLM_ARCH_UNKNOWN to "(unknown)"
This avoids errors from `std::map::at` when
getting the general name of the model architecture.
Using "(unknown)" instead of an empty string as per suggestion
https://github.com/ggerganov/llama.cpp/pull/5820#issuecomment-1973735284
* llama : remove redundant inner const for LLM_TENSOR_NAMES
The extra const won't do anything here as const maps
return const references to values.
Co-authored-by: Jared Van Bortel <cebtenzzre@gmail.com>
* llama : remove redundant nullptr check in llm_arch_from_string
Since LLM_ARCH_NAMES is a const map, no spurious elements
with a NULL name are inserted anymore, so this check is dead code.
---------
Co-authored-by: Jared Van Bortel <cebtenzzre@gmail.com>
* llama : refactor internal quantization functions (llama/5830)
* scripts : add pod-llama.sh
* ggml : IQ3_S improvements (llama/5829)
* iq3_s: somewhat faster AVX2 dot product
On Ryzen a 7950X TG-128 increases to 16 t/s from 15.5 t/s using
16 threads. For 8 threads it is 13.85 t/s vs 11.75 t/s.
PP-512 increases to 28.5 t/s from 23.8 t/s.
* iq3_s: somewhat faster ARM_NEON dot product
Still dog slow - 10.7 t/s up from 9.9 t/s.
* iq3_s: another small ARM_NEON improvement
10.7 -> 11.0 t/s. Using vmulq_s8 is faster than the xor - sub trick
that works best on AVX2.
* iq3_s: minor improvement on Metal
49.4 t/s -> 50.3 t/s
* iq3_s: PPL improvement
E.g., for a context of 4096 LLaMA-v2-7B goes to 5.1340 from 5.1653.
* iq3_s: use new grid everywhere
* Fix ARM_NEON
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* convert-hf : make model class definitions self-contained (llama/5825)
* convert : automatically fall back to HfVocab if tokenizer.model doesn't exist (llama/5821)
* ggml : fix IQ3_S AVX implementation (llama/5834)
ggml-ci
* llama : add abort_callback to interrupt computation (llama/5409)
* using abort_callback from ggml to stop llama computation
* format fix
* a brief explaining comment
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* server: tests: passkey challenge / self-extend with context shift demo (llama/5832)
* server: tests: add models endpoint scenario
* server: /v1/models add some metadata
* server: tests: add debug field in context before scenario
* server: tests: download model from HF, add batch size
* server: tests: add passkey test
* server: tests: add group attention params
* server: do not truncate prompt tokens if self-extend through group attention is enabled
* server: logs: do not truncate log values
* server: tests - passkey - first good working value of nga
* server: tests: fix server timeout
* server: tests: fix passkey, add doc, fix regex content matching, fix timeout
* server: tests: fix regex content matching
* server: tests: schedule slow tests on master
* server: metrics: fix when no prompt processed
* server: tests: self-extend add llama-2-7B and Mixtral-8x7B-v0.1
* server: tests: increase timeout for completion
* server: tests: keep only the PHI-2 test
* server: tests: passkey add a negative test
* flake.lock: Update (llama/5842)
Flake lock file updates:
• Updated input 'flake-parts':
'github:hercules-ci/flake-parts/b253292d9c0a5ead9bc98c4e9a26c6312e27d69f' (2024-02-01)
→ 'github:hercules-ci/flake-parts/f7b3c975cf067e56e7cda6cb098ebe3fb4d74ca2' (2024-03-01)
• Updated input 'flake-parts/nixpkgs-lib':
'github:NixOS/nixpkgs/97b17f32362e475016f942bbdfda4a4a72a8a652?dir=lib' (2024-01-29)
→ 'github:NixOS/nixpkgs/1536926ef5621b09bba54035ae2bb6d806d72ac8?dir=lib' (2024-02-29)
• Updated input 'nixpkgs':
'github:NixOS/nixpkgs/cbc4211f0afffe6dfd2478a62615dd5175a13f9a' (2024-02-23)
→ 'github:NixOS/nixpkgs/1536926ef5621b09bba54035ae2bb6d806d72ac8' (2024-02-29)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
* server : init http requests thread pool with --parallel if set (llama/5836)
* ci : schedule slow server tests only on Release or on demand (llama/5839)
* llama : fix llama_copy_state_data with fragmented KV cache (llama/5840)
The row size of the saved states was based on kv_self.head while
it should be based on llama_kv_cache_cell_max.
Existing session files should still work.
* llama : fix llama_kv_cache_cell_max inability to return 1
I've also changed its return type to uint32_t,
because this function is always used to set the value of uint32_t variables,
and because the index already has this type.
* llama : fix state size calculation
Some bytes in the state were unaccounted for in llama_get_state_size.
Since the logits reserve so much space, it did not cause problems.
* gguf-dump : support i-quants (llama/5841)
Co-authored-by: Black_Fox <radekliska@gmail.com>
* llama : allow for user specified embedding pooling type (llama/5849)
* allow for user specified pooling type
* llama : use enum types over int
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* readme : add API changes section
* cuda : fix data race in soft max (llama/5853)
* main : support special tokens as reverse/anti prompt (llama/5847)
* Support special tokens as reverse/anti prompt.
* Tokenize antiprompts only once.
* main : minor
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* common : use LLAMA_DEFAULT_SEED (llama/5855)
* add some new ops, fix some operators and add batch operations to certain operators. (ggml/747)
* cuda: fix group_norm
* cuda: add batch inference support for ggml_pad/ggml_upscale
* add ggml_arrange
* add ggml_timestep_embedding
* update ggml_arange/ggml_timestep_embedding tests
* cuda: fix im2col
* add ggml_arange/ggml_timestep_embbeding support for metal backend
* fix some bugs
* fix some bugs
* Update ggml.h
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update ggml-cuda.cu
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update ggml-metal.m
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update ggml-metal.m
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update ggml-metal.metal
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* modify according to the review comments
* ggml : fix compile warnings + code style
* ggml : normalize compute_forward calls + fix seg fault in debug
* minor
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
Co-authored-by: slaren <slarengh@gmail.com>
* sync : ggml
* add alias for chat template (llama/5858)
* speculative : implement stochastic speculative sampling (llama/5625)
* (WIP) Implement stochastic speculative decoding
* sample from residual distribution on draft accept failure
* fix#5657: force greedy sampling with probs when temp is 0
* remove p_accept parameter
* fix style
* remove unused variables
* add srand() in speculative.cpp
* replace use of rand() with mt19937 sampling
* fixes based on review (@JohannesGaessler)
* fix r random generation
* randomly select next sequence to verify + fix bug in memory freeing
* fix bug in active_seqs sync
* fix uniform int distribution initialization
* remove warnings from comparison between int and size_t
* check grammar in `llama_sample_probability_distribution_impl`
* remove malloc code by utilizing vectors
* add PR link to README
* cmake : handle cases where git index is not found in .git (llama/5844)
* Update CMakeLists.txt
* Update CMakeLists.txt
* ggml : introduce ggml_status (ggml/750)
* using enum as an exit code instead of macros
* update return type from enum to unsigned int
* indentation fix
* compound update
ggml_compute_exit_code -> ggml_status
changed ggml_status from a bit-field type to simple codes
ggml_status to string cast
* ggml_status to string cast
* GGML_CALL was removed
Co-authored-by: slaren <slarengh@gmail.com>
---------
Co-authored-by: slaren <slarengh@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* sync : ggml
ggml-ci
* ggml : fix unknown status (llama/0)
* flake : fix
* llama : fix embeddings (llama/5796)
* llama : fix embeddings
ggml-ci
* llama : do not use KV cache for non-causal models
ggml-ci
* embeddings : fix llama_batch_init arg
* llama : add pooling switch
* llama : distinguish token vs sequence embeddings
ggml-ci
* llama : assert pooling tensor
* llama : simplify causal mask condition
ggml-ci
* llama : assert input batch with pooling enabled
* readme : update API changes list
* nix: static build (llama/5814)
* fix speculative decoding build on windows (llama/5874)
* rebase and rm tailing space
---------
Co-authored-by: LiangtaoJin <liang-tao.jin@intel.com>
Co-authored-by: compilade <113953597+compilade@users.noreply.github.com>
Co-authored-by: Jared Van Bortel <cebtenzzre@gmail.com>
Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
Co-authored-by: Kawrakow <48489457+ikawrakow@users.noreply.github.com>
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Co-authored-by: Jared Van Bortel <jared@nomic.ai>
Co-authored-by: Michael Podvitskiy <podvitskiymichael@gmail.com>
Co-authored-by: Pierrick Hymbert <pierrick.hymbert@gmail.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Nindaleth <Nindaleth@users.noreply.github.com>
Co-authored-by: Black_Fox <radekliska@gmail.com>
Co-authored-by: Douglas Hanley <thesecretaryofwar@gmail.com>
Co-authored-by: slaren <slarengh@gmail.com>
Co-authored-by: DAN™ <dranger003@gmail.com>
Co-authored-by: leejet <leejet714@gmail.com>
Co-authored-by: Minsoo Cheong <54794500+mscheong01@users.noreply.github.com>
Co-authored-by: Dane Madsen <dane_madsen@hotmail.com>
Co-authored-by: hutli <6594598+hutli@users.noreply.github.com>
Co-authored-by: Jeffrey Quesnelle <emozilla@nousresearch.com>
* using enum as an exit code instead of macros
* update return type from enum to unsigned int
* indentation fix
* compound update
ggml_compute_exit_code -> ggml_status
changed ggml_status from a bit-field type to simple codes
ggml_status to string cast
* ggml_status to string cast
* GGML_CALL was removed
Co-authored-by: slaren <slarengh@gmail.com>
---------
Co-authored-by: slaren <slarengh@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* iq3_s: somewhat faster AVX2 dot product
On Ryzen a 7950X TG-128 increases to 16 t/s from 15.5 t/s using
16 threads. For 8 threads it is 13.85 t/s vs 11.75 t/s.
PP-512 increases to 28.5 t/s from 23.8 t/s.
* iq3_s: somewhat faster ARM_NEON dot product
Still dog slow - 10.7 t/s up from 9.9 t/s.
* iq3_s: another small ARM_NEON improvement
10.7 -> 11.0 t/s. Using vmulq_s8 is faster than the xor - sub trick
that works best on AVX2.
* iq3_s: minor improvement on Metal
49.4 t/s -> 50.3 t/s
* iq3_s: PPL improvement
E.g., for a context of 4096 LLaMA-v2-7B goes to 5.1340 from 5.1653.
* iq3_s: use new grid everywhere
* Fix ARM_NEON
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* suport multiple cards: split-mode - layer|row
* rm warning
* rebase with master, support tow new OPs, close feature for -sm=row, fix for unit test
* update news
* fix merge error
* update according to review comments
* WIP: make i-quants work for QK_K = 64
* iq2_xs: attempt to fix AVX dot product for QK_K = 64
Tests pass, but I get gibberish.
* QK_K = 64 tests pass on ARM_NEON and Metal
Sadly, that does not mean it actually works.
* Make CUDA compile with QK_K = 64
Tests don't pass, plus we get misaligned access
* Q2_K: fixed bug in imatrix quantization for QK_K = 64
* iq1_s: turn off SIMD implementation for QK_K = 64 (it does not work)
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* Try IQ4_NL with blocks of 64 - does not look good
* iq4_xs: go to super-blocks of 256 and 6-bit scales for blocks of 32
* iq4_xs: CUDA works - 133.2 t/s
* iq4_xs: AVX2 dot product
* iq4_xs: ARM_NEON dot product
* iq4_nl: Metal implementation
As usual, Metal / Apple Silicon don't like my quants.
* iq3_xs: minor fix
* iq4_xs: shrink by using IQ3_S for attn_k and attn_q
* iq4_xs: revert using IQ3_S for attn_k and attn_v
PPL vs size is good, but CPU performance suffers: on M2 Max
TG-128 drops to 21.7 t/s from 28.8, and on a Ryzen-7950X
to 14.5 t/s from 15.8 t/s. On CUDA we have 135 t/s when
using IQ3_S vs 133 t/s with pure IQ4_XS.
* Fix CI
* iq4_xs: Added forgotten check for 256 divisibility
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* Adding IQ2_S and IQ2_M as a single cumulative commit
* Update examples/quantize/quantize.cpp
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* [ggml-quants] Provide ggml_vqtbl1q_u8 for 64bit compatibility
vqtbl1q_u8 is not part of arm v7 neon library
* [android-example] Remove abi filter after arm v7a fix
* [github-workflows] Do not skip Android armeabi-v7a build
* add magika inference example
* ggml : fix unaligned accesses in custom ops
* ggml : fix FP32 GELU for values that exceed the FP16 range
* use ggml_pool_1d
* add README
* Update README.md
* pad inputs if the files are too small
* cleanup
ggml-ci
* iq4_nl: squash commits for easier rebase
* Basics (quantize, dequantize)
* CUDA dequantize and dot product
* Slightly faster CUDA dot product (120 t/s)
* Switch to 6-bit scales
* Scalar dot product
* AVX2 dot product
* ARM_NEON dot product
* Works on metal, but still slow
* Slightly better Metal dot product
* Another small Metal improvement
* Metal dot product is getting there
* Faster CUDA dot product
* Add 1/8 ffn_down layers as Q5_K when no imatrix has been provided
* Report the actual bpw
* Add _xs mix that is 4.05 bpw for non-MoE models
* Remove IQ4_XS for now, slightly adjust kvalues_iq4nl
* AVX2 dot product uses Q8_0 instead of Q8_K
* Add to test-backend-ops
* Minor fix
* Also use use Q5_K for attn_output in MoE models
* Fixes after merging latest master
* Switching to blocks of 32
* AVX2 for blocks of 32
* Scaler dot product for blocks of 32
* ARM_NEON dot product for blocks of 32
* Metal kernels for blocks of 32
* Slightly faster Metal kernels
* Resurrecting iq3_xs
After all the experimentation, nothing was better than this.
* Minor PPL improvement via a block scale fudge factor
* Minor improvement via 3 neighbours
* iq3_xs: working scalar and AVX2 dot products
* iq3_xs: ARM_NEON dot product - works but extremely slow (10 t/s)
* iq3_xs: working Metal implementation
* Adding IQ3_M - IQ3_XS mix with mostly Q4_K
* iiq3_xs: a 3.4375 bpw variant
* iq3_xs: make CUDA work for new version
* iq3_xs: make scalar and AVX2 work for new version
* iq3_s: make ARM_NEON work with new version
* iq3_xs: make new version work on metal
Performance is very similar to Q3_K_S
* iq3_xs: tiny Metal speed improvement
* iq3_xs: tiny Metal speed improvement
* Fix stupid warning
* Q3_K_XS now uses a mix of IQ3_XS and IQ3_XXS
* iq3_xs: rename to iq3_s
* iq3_s: make tests pass
* Move Q3_K_XS mix to 3.25 bpw
* Attempt to fix failing tests
* Another attempt to fix the Windows builds
* Attempt to fix ROCm
* ROCm again
* iq3_s: partial fix for QK_K = 64
* iq3_s: make it work on metal for QK_K = 64
Pleasent surprise: the coding was super-block size independent,
so all it took was to delete some QK_K == 256 guards.
* Will this fix ROCm?
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* Introduce backend GUIDs
Initial proposed implementation of backend GUIDs
(Discussed in https://github.com/ggerganov/ggml/pull/741)
Hardcoded CPU backend GUID (for now)
Change ggml_backend_is_cpu logic to use GUID
* Remove redundant functions
Remove redundant functions `ggml_backend_i::get_name` and `ggml_backend_guid` which are not desired for future expansion
* Add spaces to match style
Co-authored-by: slaren <slarengh@gmail.com>
* Fix brace style to match
Co-authored-by: slaren <slarengh@gmail.com>
* Add void to () in function signature
Co-authored-by: slaren <slarengh@gmail.com>
* Add back ggml_backend_guid and make CPU_GUID a local static in ggml_backend_cpu_guid
* add guids to all backends
ggml-ci
---------
Co-authored-by: slaren <slarengh@gmail.com>
* talk-llama: pass file instead of arg
it is too hard to quote text in a portable way
* talk-llama: pass heard_ok as a file
* talk-llama: let eleven-labs.py accept options
Options: -v voice, -s savefile, -p (--play)
* talk-llama: check installed commands in "speak"
Pass "-q" to eleven-labs.py to skip checking whether elevenlabs is installed
* talk-llama: pass voice_id again
in order to sync talk with talk-llama
* talk: sync with talk-llama
Passing text_to_speak as a file is safer and more portable
cf. https://stackoverflow.com/a/59036879/45375
* talk and talk-llama: get all installed voices in speak.ps1
* talk and talk-llama: get voices from api
* talk and talk-llama: add more options to eleven-labs.py
and remove DEFAULT_VOICE because it is deprecated (https://www.reddit.com/r/ElevenLabs/comments/1830abt/what_happened_to_bella/)
```
usage: eleven-labs.py [-q] [-l] [-h] [-n NAME | -v NUMBER] [-f KEY=VAL] [-s FILE | -p] [TEXTFILE]
options:
-q, --quick skip checking the required library
action:
TEXTFILE read the text file (default: stdin)
-l, --list show the list of voices and exit
-h, --help show this help and exit
voice selection:
-n NAME, --name NAME get a voice object by name (default: Arnold)
-v NUMBER, --voice NUMBER
get a voice object by number (see --list)
-f KEY=VAL, --filter KEY=VAL
filter voices by labels (default: "use case=narration")
this option can be used multiple times
filtering will be disabled if the first -f has no "=" (e.g. -f "any")
output:
-s FILE, --save FILE save the TTS to a file (default: audio.mp3)
-p, --play play the TTS with ffplay
```
* examples: add speak_with_file()
as suggested in the review
* talk and talk-llama: ignore to_speak.txt
* refactored compute forward to not pass in the src tensors each time
* fix merge issues with flags
* missed one place in the last commit to fix the is_param / flags issue
* minor spacing fix
* fixed some variable assignments so all tests locally are passing
* new change after merge fix
---------
Co-authored-by: siddharthvader <siddharth@coinlist.co>
Fix issue: Conversion from Whisper to OpenVino failed #1870
convert-whisper-to-openvino.py stopped working with OpenVINO version 2023.0.0-10926-b4452d56304-releases/2023/0 .
Error was: TypeError: load(): incompatible function arguments. The following argument types are supported:
1. (self: openvino._pyopenvino.FrontEnd, path: object) -> ov::frontend::InputModel
Tested successfully with a large-v3 conversion.
Co-authored-by: Stefan Grundmann <grundmanns@sandiego.gov>
In commit dda4b0e of PR #1872, I've introduced a check for the
existence of files before loading the model. However, I haven't
considered the case where whisper.cpp might read from stdin as well,
and in such cases, the checks should ignore the "-" argument as it
does not represent a regular file.
Additionally, this commit removes the usage of 'stat()' in favor of
the recently introduced function 'is_file_exist()' in common.cpp from
PR #1871.
Apologies for the bug introduced in the previous PR and any
inconvenience it may have caused.
* #ifdef out some code NUMA blocks for Android due to lack of support
* added in some __ANDROID__ if def gates around numa code and forced GLIBC prior to 2.29 to use a syscall for getcpu instead of the wrapper
* Changed gates on numa platform specific stuff to __gnu_linux__ to skip any platforms without glibc
* harmonizing #if defined blocks for numa code to __gnu_linux__ since that's the only model that's being followed anyways
---------
Co-authored-by: root <root@nenya.lothlorien.ca>
* iq1_s: WIP basics
* iq1_s: CUDA is working
* iq1_s: scalar CPU dot product
* iq1_s: WIP AVX2 dot product - something is not right
* Fix tests
* Fix shadow warnings
* Fix after merge with latest master
* iq1_s: AVX2 finally works
* iq1_s: ARM_NEON dot product. Works, but not very fast
* iq1_s: better grid
* iq1_s: use IQ2_XXS for attn_output
At a cost of 0.04 extra bpw this gives a big improvement in PPL.
* iq1_s: Metal basics
Dequantize works, but not dot product
* iq1_s: Metal works, but quite slow
As usual, Apple Silicon does not like the code I write.
* iq1_s: Tests
* iq1_s: slightly faster dot product
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* Added numa options to allow finer grained control as well as plumbing for a new mirror mode that will require numa.h
* Reverted Makefile
* Fixed include
* Removed sched.h from ggml.h, moved ggml_get_numa_affinity into ggml.c, removed trailing whitespace and fixed up a few inconsistent variables
* removed trailing whitespace
* Added numa options to allow finer grained control as well as plumbing for a new mirror mode that will require numa.h
* Reverting Makefile
* Fixed a number of issues with the move from BOOL to ggml_numa_strategies. Added a note about mirror mode note being implemented yet
* Removing MIRROR_MODE code for this PR
* Removing last bit of MIRROR_MODE code for this PR
* Removing unneeded branch in server.cpp example and moving get_numa_affinity and making it static
* Fixed lingering init_llama_backend() bool calls in tests and examples
* Remote enum llama_numa_strategies
* Revert bad merge with dynatemp flags
* add missing enum ggml_numa_strategies declaration and revert sync problem with master
* add missing enum ggml_numa_strategies declaration
* fixed ggml_init_numa variable
* Update ggml.h
Co-authored-by: Jared Van Bortel <cebtenzzre@gmail.com>
* Update READMEs with info about numa flags, change INTERLEAVE strategy name to DISTRIBUTE everywhere, implement the improved distribution strategy from @rankaiyx, fix a spelling mistake and un-merge some bad merges
* split numa init out from llama_backend_init and created llama_numa_init. Updated all code paths and samples
* Fix up some boolean vs enum comparisons
* Added #ifdefs for non-Linux OS that don't have cpu_set_t datatype
* Update ggml.h
Align enum values
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update ggml.c
Remove whitespace
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update ggml.c
align paremeters
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update examples/server/server.cpp
remove whitespace and align brace
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update common/common.cpp
Remove whitespace and align brace
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* unified ggml_numa_strategy enum and fixed text alignment in server.cpp example
* Update ggml.c
simplified return for platforms without NUMA support
Co-authored-by: Jared Van Bortel <cebtenzzre@gmail.com>
* removed redundant else from cli argument processing of --numa
* whitespace
---------
Co-authored-by: root <root@nenya.lothlorien.ca>
Co-authored-by: Jared Van Bortel <cebtenzzre@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
Co-authored-by: Jared Van Bortel <jared@nomic.ai>
* Early return for zero size calls to get_tensor.
Signed-off-by: Adam Treat <treat.adam@gmail.com>
* Update ggml-kompute.cpp
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update ggml-kompute.cpp
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Add an early return to the get/set tensor when the size is null.
Signed-off-by: Adam Treat <treat.adam@gmail.com>
* Early return after the assertions.
Signed-off-by: Adam Treat <treat.adam@gmail.com>
* Since we do the early return in the generic backend now no reason to do so here as well.
Signed-off-by: Adam Treat <treat.adam@gmail.com>
---------
Signed-off-by: Adam Treat <treat.adam@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
Until the most recent commit (3d42463), the main.cpp sample file does
not check whether the input files exist or not. Consequently, the
model is loaded first before reporting whether there was a failure or
not when processing a file. In environments with HDD, this can take
about 50 seconds or more, depending on the loaded model.
This commit addresses this issue by checking in advance whether the
input files exist or not.
* ggml-alloc : allocate all leafs as if they were inputs
* ensure static leafs are allocated
* gpt-2-backend : remove unnecesary ggml_new_tensor
* update other gpt-2 examples to remove ggml_new_tensor calls in the graph
* vulkan: refactor guess_matmul_pipeline for vendor
Refactor ggml_vk_guess_matmul_pipeline to simplify adding per-vendor
conditionals.
Signed-off-by: Sergio Lopez <slp@redhat.com>
* vulkan: only use M-sized matmul on Apple GPUs
L-sized and S-sized matmuls are broken on Apple GPUs, force using
M-size with this vendor.
Signed-off-by: Sergio Lopez <slp@redhat.com>
---------
Signed-off-by: Sergio Lopez <slp@redhat.com>
* ggml: aarch64: implement smmla kernel for q8_0_q8_0 quantized gemm
armv8.2-a and above supports MMLA instructions that have higher
throughput than DOT. this commit adds mmla kernel for
q8_0_q8_0 gemm. The feature is enabled if the platform supports
"__ARM_FEATURE_MATMUL_INT8"
On AWS Graviton3 processors this kernel resulted up to 1.5x
improvement for prompt evaluation throughput compared to the
default sdot kernel.
* ggml: aarch64: implement smmla kernel for q4_0_q8_0 quantized gemm
armv8.2-a and above supports MMLA instructions that have higher
throughput than DOT. this commit adds mmla kernel for
q4_0_q8_0 gemm. The feature is enabled if the platform supports
"__ARM_FEATURE_MATMUL_INT8"
On AWS Graviton3 processors this kernel resulted up to 1.5x
improvement for prompt evaluation throughput compared to the
default sdot kernel.
* ggml: aarch64: implement smmla kernel for q4_1_q8_1 quantized gemm
armv8.2-a and above supports MMLA instructions that have higher
throughput than DOT. this commit adds mmla kernel for
q4_1_q8_1 gemm. The feature is enabled if the platform supports
"__ARM_FEATURE_MATMUL_INT8"
On AWS Graviton3 processors this kernel resulted up to 1.5x
improvement for prompt evaluation throughput compared to the
default sdot kernel.
* ggml: update unit tests for the new vec_dot interface
* llama.cpp: add MATMUL_INT8 capability to system_info
* added audio_ctx argument to main and server examples
* Better default value
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* better default value (again)
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* ggml : embed Metal library source (ggml-metal.metal) into binary
enable by setting WHISPER_EMBED_METAL_LIBRARY
* rename the build option
* rename the preprocessor directive
* generate Metal library embedding assembly on-fly during build process
* Initial Vulkan multi-gpu implementation
Move most global variables into backend context
* Add names to backend device functions
* Add further missing cleanup code
* Reduce code duplication in tensor split layer assignment
* generalize LLAMA_SPLIT_LAYER for all backends, do not expose device count and memory in llama.h
* Only do device info print in the beginning and initialize one backend for cpu assist
Add missing cleanup code
* Rework backend memory management to make sure devices and buffers get properly allocated and freed
* Rename cpu assist free function
---------
Co-authored-by: slaren <slarengh@gmail.com>
* Make use of ggml-quants.h possible in C++ code
* One cannot possibly be defining static_assert in a C++ compilation
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* Avoid duplicating function calls when using MIN/MAX macros.
Since these copy "a" and "b" they ask the compiler to evaluate one of them twice. The compiler doesn't have a problem with removing the duplication in something like MAX(0, x + 2), but in some cases we're calling functions, and those calls just happen twice.
By explicitly evaluating at the expression we get smaller and faster code without duplicate calls. See ggml_rope_yarn_corr_dims in Compiler Explorer:
https://godbolt.org/z/Ee4KMrvKh
Code behaves exactly the same.
* Update ggml.c
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
We get slightly better PPL, and we cut quantization time in
nearly half.
The trick is to 1st quantize without forcing points onto the E8-lattice.
We can then use a narrower search range around the block scale that we
got that way.
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* New Feature:
1. Sum_Rows:
fix cuda kernel overflow
fix block shape error when nrows too big
2. Im2Col:
Support Batch in cuda
Support f32 to f32 both in cpu && cuda
3. DepthWiseConv:
Support by Im2Col && MulMat
4. Pool_2d:
Supoort avg pooling in cuda
5. HardSigmoid:
Imp in cuda
6. HardSwish:
Imp in cuda
* fix tabs instead of spaces
* code clean
* CUDA POOL2D
* ADD POOL2D test case in test-backend-ops.cpp
* code clean
* fix pool2d_kernel
nits
* fix bug in pool2d kernel
* fix avg pooling, count_include_pad
nits
* test-backend-ops : add more pool_2d tests
* cuda : fix warnings and formatting
* ggml : check types in release builds too in pool_2d
* test-backend-ops : remove f16 pool_2d tests
* cuda : more style fixes
* Add assert in ggml_cuda_op_pool2d
* pool2d float padding fallback
* test-backend-ops : add dst_type to im2col
---------
Co-authored-by: slaren <slarengh@gmail.com>
Whisper plugin in Obsidian requires an API key which is
then sent as an authorization header.
However, the presence of an authorization header requires
a CORS Preflight, so both the OPTIONS method and
the Access-Control-Allow-Headers: authorization must be
handled.
* iq3_xxs: quantize/dequantize
RMSE seems a bit high-ish at about half-way between q2_K and
q3_K, so need to check more.
* iq3_xxs: CUDA dequantize works
* iq2_xxs: tuning quantization
* iq3_xxs: starting to look better
PPL on wiki.test.raw
LLaMA-v1-7B: 6.4218
LLaMA-v2-7B: 6.3560
Mistral-7B : 6.0717
This is better than Q3_K_XS, with a 5% reduction in quantized model
size.
* iq3_xxs: CUDA dot product
We have
PP-512: 5891 t/s
TG-128: 143.9 t/s
* iq3_xxs: scalar and AVX2 dot products
* iq3_xxs: ARM_NEON and Metal
Metal performance is decent, ARM_NEON is pathetic
* iq3_xxs: slightly better grid points
* Faster iq3_xxs and iq2_xs dot products on CUDA
* iq3_xxs: add some quant mix
* iq3_xxs: fix failing quantization test
Dot product still fails. Is this real?
* iq3_xxs: hopefully fix ROCm
* iq3_xxs: failing tests
This time the dot product accuracy did find an actual bug
in the AVX2 implementation.
* Add IQ3_XXS to test-backend-ops
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* Fix for a null pointer dereference if a metal GGML buffer fails to be allocated
* Freeing the allocated buffers rather than the pointer in ggml-alloc.c
* Fixed the fix of the fix
* added cuda float16->float32 upcasting to ggml_cuda_cpy
* added ability to copy 4d tensors with the cuda backend
* added tests for float16_>float32 upcast and 4d tensor cuda copys
* added 4d copy test for float32->float16 copy
* applied patch suggested by @iamlemec
* simplify cpy tests
---------
Co-authored-by: slaren <slarengh@gmail.com>
* server: include additional fields in the verbose_json response as OpenAI does
* server: show request examples on home page
* server: todo note for compression_ratio and no_speech_prob
* server: add simple demo form to the homepage
I just upgraded the R wrapper at https://github.com/bnosac/audio.whisper to use whisper.cpp 1.5.4
I'm working on Windows and noticed while doing that that it did not pick up the relevant CFLAGS/CXXFLAGS as my system showed
```
I whisper.cpp build info:
I UNAME_S: MSYS_NT-10.0-19045
I UNAME_P: unknown
I UNAME_M: x86_64
```
Many thanks for all the tremendous hard work on maintaining whisper.cpp!
* Vulkan loader code
* Fix matmul kernel, continue implementation
* Continue implementation
* Vulkan memory management
* Vulkan development
* Matmul call
* Add aligned malloc and free for VMA
* Continue implementation
* First matmul success
* GEMM Kernel optimization
* 1D Blocktiling
* 2D Blocktiling
* Write coalescing
* Continue vulkan implementation and optimization
* First FP16 attempt, disabled for now
* Code abstraction, FP16 implementation, fix kernel, add FP16 to FP32 kernel
* Enable device extensions properly, restore fp16 matmul op
* Fix mulmat_f16
* Output FP32 in fp16 matmul shader
* Fix f16_to_f32 kernel
* dequant_q4_0 kernel
* Add VMA library
* Avoid requesting dedicated memory, VMA can decide that by itself
* Add bounds checking to matmul kernels, improve implementation, fix command buffers not freed properly
* add cmake commands
* Add 2d write operation, profiling code
* Fix 2d write
* Fix queue selection for AMD RADV
* Fix trailing whitespace in vk_mem_alloc.h
* Add WIP warp tile mat mul shaders
* Disable glslc optimization
* Disable glslc optimization for CMake
* Optimize warptile matmul shader, replace blocktile with it
* Add split-k optimization for small matrix multiplication
Use semaphores for synchronization instead of fences or waitidle
Rework async write/read for synchronization
* Fix validation errors, improve compatibility with AMD GPUs
* Rework command buffer handling
* Variable matmul kernel using specialization constants
* Fix synchronization on AMD, add barriers for buffer ownership transfer, add debug flag and prints
* Reuse semaphores
* Handle stage flags during command buffer submission properly
* Increase matmul test runs for consistent results
* Fix F32 matmul
* Add vectorized loading and zeropadding for matrix multiplication
* Use pinned memory for f16 preprocessing
* Don't force aligned matmul
* Don't free before queue done
* Replace VMA library with native Vulkan buffer management
* Basic offloading support with mul_f32 and dmmv for q4_0
* Run glslc commands in parallel
* Unroll loops in dmmv shader
* Reduce usage of waitIdle
* Reuse pinned allocation for f16 conversion
* Handle devices with only a single queue
* Fix trailing whitespace in CMakeLists.txt
* Allow parallel execution of kernels, parallelize third and fourth dimension calls
* Add fallback for devices only supporting one DescriptorSet per DescriptorPool
* Move to graph function similar to CUDA implementation
* Use F16 kernel for most things, replace q_f32 with mul_mat_q_f16 function
* Add F32 dmmv shaders
* Batch submissions
* Add .spv to gitignore
* Split off matrix vector multiplication for separate optimization
* Use single command buffer for matrix vector multiplication ops
* Reduce overhead of mul_f32 calls by using a single command buffer
* Add submission batching to mul_f32
* Fix tests
* Add missing barrier
* Add further missing barrier
* Add further ops
* Replace vk::QueueFamilyIgnored with VK_QUEUE_FAMILY_IGNORED to support more Vulkan header versions
* Remove unnecessary cblas link
* Fix descriptor set pre-allocation assert
* Add runtime shader compilation, start transferring shaders to this approach
* Transfer remaining shaders to header and compile on runtime
* Fix fp32 fallback if device doesn't support fp16, add force disable env var GGML_VULKAN_DISABLE_F16
* Add support for q4_1, q5_0, q5_1 and q8_0
* Remove unnecessary scalar layout extension
* Parse graph early to pre-record command buffers
* Add q6_k support
* Add multi-submit for command buffers
* Fix q6_k dequant shader for AMD
* Fix q6_k for GPUs without fp16 support
* Simplify q6_k fp16 fix
* Minor fixes
* Fix wg_denom of m-mulmat shaders
* Add Python-based Vulkan shader generator
* Replace shaderc dependency with precompiled shaders
Fix python script to generate shaders
* Clean up code
* Fix shader generator script Windows compatibility
Co-authored-by: Concedo <39025047+LostRuins@users.noreply.github.com>
* Close file before deletion
* Fix vulkan shader fp32 name
* Add q2_k and q3_k support
Add validation check to compare shader results to cpu results
* Add q4_k support
* Add q5_k support
* Bake SPIR-V bytecode into the library instead of loading shaders from file
* Switch to signal semaphores for flexibility
Prepare broadcasting support for mul mat
* Finish broadcasting mul mat support for GQA
* Clean up unused functions
Add repeat op
* Add further ops, not yet enabled. Improve semaphore code
* Reduce number of used semaphores by utilizing timelines more properly
* Remove queue information
* Reuse timeline semaphores, allow parallel operation with binary semaphores to work around nvidia driver limitations
* Add Vulkan to llama-bench
* Remove cblas dependency
* Fix matmul k-split bug
* Fix q4_k dmmv K_QUANTS_PER_ITERATION 1 shader
* Add RMS Norm shader, rework op_f32 shader setup, fix matmul bug
* Fix issues with float16 overflows in shaders
* Fix issues with older Vulkan headers on Ubuntu 22.04
* Allow multi-op partial offloading by parsing the graph to preallocate enough between-op buffers
* Implement further ops, rework op_f32 calls, fix bugs
* Finish full offloading support, add last remaining ops, fix bugs, remove redundant code
* Upload generated file ggml-vulkan-shaders.hpp, remove redundant shaders
* Merge upstream changes, fix conflicts, adapt soft_max op
* Fix Python and shader header format
* Free model gpu buffers on exit
* Use single queue per device to simplify code
* Add matmul shader support for running multiple calculations in parallel
* Switch from semaphore-synchronized multiple command buffers per op to single command buffer for multiple ops, whole graph if possible
* Fix missing event cast
* Replace uint64_t(-1) with UINT64_MAX, rename function for clarity
* Fix warning about empty C function parameters
* Fix compiler warnings
* Properly implement Vulkan backend buffer handling
* Fix oversized host staging buffers
* Simplify barrier synchronization calls
* Fix gcc warnings
* Implement max_size for backend buffer types to limit the size of a single allocation
* Use min of maxMemoryAllocationSize and maxBufferSize for device max allocation size
* refactor multi buf
* Disable unsupported ops to fix tests
* Check for maintenance4 support before using it
* Handle devices with only a single queue
* Fix single queue logic
* propagate buffer usage in multi buffers
* Implement rope_neox op
* Cleanup header and other files
* Simplify gpu_extras by removing events and putting staging memcpys into contexts
* Move queue into context
Add not-yet-enabled async backend ops
* Simplify context use, optimize matmul shader for warp size 64 (AMD GCN), fix split_k matmul shader optimization
* Add get_max_size to SYCL backend.
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* llama : fix trailing whitespace
---------
Co-authored-by: Henri Vasserman <henv@hot.ee>
Co-authored-by: Concedo <39025047+LostRuins@users.noreply.github.com>
Co-authored-by: slaren <slarengh@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* first update for migration
* update init_cublas
* add debug functio, commit all help code
* step 1
* step 2
* step3 add fp16, slower 31->28
* add GGML_LIST_DEVICE function
* step 5 format device and print
* step6, enhance error check, remove CUDA macro, enhance device id to fix none-zero id issue
* support main device is non-zero
* step7 add debug for code path, rm log
* step 8, rename all macro & func from cuda by sycl
* fix error of select non-zero device, format device list
* ren ggml-sycl.hpp -> ggml-sycl.h
* clear CMAKE to rm unused lib and options
* correct queue: rm dtct:get_queue
* add print tensor function to debug
* fix error: wrong result in 658746bb26702e50f2c59c0e4ada8e9da6010481
* summary dpct definition in one header file to replace folder:dpct
* refactor device log
* mv dpct definition from folder dpct to ggml-sycl.h
* update readme, refactor build script
* fix build with sycl
* set nthread=1 when sycl, increase performance
* add run script, comment debug code
* add ls-sycl-device tool
* add ls-sycl-device, rm unused files
* rm rear space
* dos2unix
* Update README_sycl.md
* fix return type
* remove sycl version from include path
* restore rm code to fix hang issue
* add syc and link for sycl readme
* rm original sycl code before refactor
* fix code err
* add know issue for pvc hang issue
* enable SYCL_F16 support
* align pr4766
* check for sycl blas, better performance
* cleanup 1
* remove extra endif
* add build&run script, clean CMakefile, update guide by review comments
* rename macro to intel hardware
* editor config format
* format fixes
* format fixes
* editor format fix
* Remove unused headers
* skip build sycl tool for other code path
* replace tab by space
* fix blas matmul function
* fix mac build
* restore hip dependency
* fix conflict
* ren as review comments
* mv internal function to .cpp file
* export funciton print_sycl_devices(), mv class dpct definition to source file
* update CI/action for sycl code, fix CI error of repeat/dup
* fix action ID format issue
* rm unused strategy
* enable llama_f16 in ci
* fix conflict
* fix build break on MacOS, due to CI of MacOS depend on external ggml, instead of internal ggml
* fix ci cases for unsupported data type
* revert unrelated changed in cuda cmake
remove useless nommq
fix typo of GGML_USE_CLBLAS_SYCL
* revert hip cmake changes
* fix indent
* add prefix in func name
* revert no mmq
* rm cpu blas duplicate
* fix no_new_line
* fix src1->type==F16 bug.
* pass batch offset for F16 src1
* fix batch error
* fix wrong code
* revert sycl checking in test-sampling
* pass void as arguments of ggml_backend_sycl_print_sycl_devices
* remove extra blank line in test-sampling
* revert setting n_threads in sycl
* implement std::isinf for icpx with fast math.
* Update ci/run.sh
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update examples/sycl/run-llama2.sh
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update examples/sycl/run-llama2.sh
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update CMakeLists.txt
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update CMakeLists.txt
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update CMakeLists.txt
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Update CMakeLists.txt
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* add copyright and MIT license declare
* update the cmd example
---------
Co-authored-by: jianyuzh <jianyu.zhang@intel.com>
Co-authored-by: luoyu-intel <yu.luo@intel.com>
Co-authored-by: Meng, Hengyu <hengyu.meng@intel.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
updated the n_task calculation to use max number of
threads possible. This has improved the prompt eval
performance by around 5% for DOT kernels and by
around 10% for MMLA kernels on AWS Graviton3.
* make GGML_TASK_INIT phase can be run in multithread
* multithreaded dequantize in mul_mat when using blas library
* minor fixes
* update outdated comment
* fix coding style
* simplify code
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* MobileVLM native implementation
* delete depthwise_conv_2d and permute_cpy relative code, replace the two by the existed functions, and opt ldp definition, support LLAMA_PERF option for CMake
* move android script to example/llava directory
* Fix the editor config checks
---------
Co-authored-by: Chenxiaotao03 <chenxiaotao03@meituan.com>
Benchmarks are failing because JNI expects a jstring and the benchmarks
are missing a return statement (i.e., returning null). The functions
actually build a jstring but don't return it, so this seems to have been
an oversight.
This patch returns the jstring and now the benchmarks run successfully.
Fixes#1783.
* examples/server: implement "verbose_json" format with token details.
This is intended to mirror the format of openai's Python
whisper.transcribe() return values.
* server: don't write WAV to a temporary file if not converting
* server: use std::lock_guard instead of manual lock/unlock
* ggml : add IQ2 to test-backend-ops + refactoring
ggml-ci
* cuda : update supports_op for IQ2
ggml-ci
* ci : enable LLAMA_CUBLAS=1 for CUDA nodes
ggml-ci
* cuda : fix out-of-bounds-access in `mul_mat_vec_q`
ggml-ci
* tests : avoid creating RNGs for each Q tensor
ggml-ci
* tests : avoid creating RNGs for each tensor
ggml-ci
* backend : add eval callback
ggml-ci
* backend : group nodes in a single compute when user don't need them
* backend : clean-up the implementation
ggml-ci
* simple : do not perform tensor data copy if not needed
* simple : fix
* imatrix : offload to GPU support
* imatrix : fix ggml_mul_mat_id hanlding
ggml-ci
* ci : add imatrix test
ggml-ci
* ci : rearrange output
ggml-ci
* backend : add eval callback
ggml-ci
* backend : group nodes in a single compute when user don't need them
* backend : clean-up the implementation
ggml-ci
* simple : do not perform tensor data copy if not needed
* simple : fix
* simple : no need for ggml_is_contiguous + fix bool parse
* llama : fix callback placement in llama_context_params
* backend : avoid double-ask callback calls
* simple : restore examples, imatrix will serve as a demo
* metal: Log `recommendedMaxWorkingSetSize` on iOS 16+
* Only log on iOS and macOS, ignoring tvOS and other platforms
* Check for Xcode version before using recommendedMaxWorkingSetSize
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
This change makes it possible to build ggml-cuda.cu and ggml-metal.m as
independent dynamic shared objects, that may be conditionally linked at
runtime in a multiplatform binary. It introduces a GGML_CALL annotation
that documents which functions have a cyclic call relationship, between
the application code and GPU modules.
This change does nothing, unless the build defines -DGGML_MULTIPLATFORM
which causes back-references and function pointers to conform to MS ABI
which is supported by NVCC, ROCm, XCode, GCC and Clang across platforms
* make : fix server example building on MSYS2 environments (Windows)
It was not working since commit eff3570f78
when server was introduced.
* cmake : simplify server example lib deps on Windows
server uses httplib::Server, not httplib::SSLServer, so there is no need
to mention cryptographic libraries in target_link_libraries.
Winsock (ws2_32) suffices here.
Also use plain library names like we use in other places.
* metal : detect more GPU families
* metal : refactor kernel loading
* metal : set kernel family requirements
* metal : fix kernel init + fix compile options
* metal : take into account simdgroup reduction support
* metal : print only skipped kernels
* metal : fix check for simdgroup reduction support
* metal : check for Metal 3
* metal : free allocations
* metal : normalize encoder:setComputePipelineStatus calls
ggml-ci
* metal : fix Metal3 family check
ggml-ci
* metal : check for simdgroup matrix mul. feature
ggml-ci
* llama : ggml-backend integration
* ggml-backend : add names to buffers
* fix unmap after loading
* batched-bench : add tensor_split param
* llama : check for null tensor_split
* ggml-backend : increase GGML_MAX_BACKENDS
* improve graph splitting, partial fix for --no-kv-offload
* cuda : add ggml-backend split buffer support
* cuda : do not create buffer types for devices that don't exist (fixes usage without CUDA devices available)
* ggml : fix null backend dereference (llama/4807)
* ggml : fix null backend dereference
* ggml : also check ggml_backend_is_cpu
* test-backend-ops : check buffer allocation failures
* llama : add cparam (split_mode) and command line argument (--split-mode, -sm) to configure the split mode (none, layer or row)
* ggml : fix mul_mat_id work size
* llama : rewrite session kv load/set without graphs
* minor
* llama : only initialize used backends, free backends on context free
* llama : abort ctx if cuda backend init fails
* llama : rewrite lora with ggml-backend and compute on CPU
ggml-ci
* llama : only map to a backend buffer the region of the file mapping containing the tensors used in the buffer
* opencl : add ggml-backend buffer type
* cuda : only use batched_cublas with batched mat muls (fixes fp16 tg perf)
* llama : on Metal, by default offload the full model
ggml-ci
* metal : page align the data ptr (llama/4854)
* Apply suggestions from code review
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
* cuda : fix split buffer free
* address review comments
* llama-bench : add split-mode parameter
* fix whitespace
* opencl : fix double initialization
* server : add --split-mode parameter
* use async copy and compute to improve multi-gpu performance
ggml-ci
* use async memcpys to copy the graph outputs to the CPU
* fix opencl
* use a host buffer for the cpu compute buffer for faster copies to the gpu
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
* iq2_xs: basics
* iq2_xs: this should have been in the basics
* iq2_xs: CUDA and scalar CPU works
* iq2_xs: WIP Metal
* iq2_xs: Metal now works
* iq2_xs: working, but dog slow, ARM_NEON dot product
* iq2_xs: better ARM_NEON dot product
We are now at 19.5 t/s for TG-128 and 61 t/s for PP-512 when
running on the CPU.
* iq2_xs: AVX2 dot product - 19.5 t/s
* iq2_xs: faster AVX2 dit product
21.4 t/s for TG-128, 59.2 t/s for PP-512.
The latter is 2x compared to the previous version.
* iq2_xs: had forgotten to delete iq2-data.h
* Add llama enum for IQ2_XS
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
NULL can be an integer constant expression with the value zero, in this case the behavior would be undefined because of an incorrect type being passed to the variable arguments.
* iq2_xxs: basics
* iq2_xxs: scalar and AVX2 dot products
Needed to change Q8_K to have quants in the -127...127 range,
else the IQ2_XXS AVX implementation becomes very awkward.
The alternative would have been to use Q8_0 instead. Perhaps
I'll change later, for now this is what we have.
* iq2_xxs: ARM_NEON dot product
Somehow strangely slow (112 ms/token).
* iq2_xxs: WIP Metal
Dequantize works, something is still wrong with the
dot product.
* iq2_xxs: Metal dot product now works
We have
PP-512 = 475 t/s
TG-128 = 47.3 t/s
Not the greatest performance, but not complete garbage either.
* iq2_xxs: slighty faster dot product
TG-128 is now 48.4 t/s
* iq2_xxs: slighty faster dot product
TG-128 is now 50.9 t/s
* iq2_xxs: even faster Metal dot product
TG-128 is now 54.1 t/s.
Strangely enough, putting the signs lookup table
into shared memory has a bigger impact than the
grid values being in shared memory.
* iq2_xxs: dequantize CUDA kernel - fix conflict with master
* iq2_xxs: quantized CUDA dot product (MMVQ)
We get TG-128 = 153.1 t/s
* iq2_xxs: slightly faster CUDA dot product
TG-128 is now at 155.1 t/s.
* iq2_xxs: add to llama ftype enum
* iq2_xxs: fix MoE on Metal
* Fix missing MMQ ops when on hipBLAS
I had put the ggml_supports_mmq call at the wrong place.
* Fix bug in qequantize_row_iq2_xxs
The 0.25f factor was missing.
Great detective work by @ggerganov!
* Fixing tests
* PR suggestion
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Add commented-out command to optionally use Piper (https://github.com/rhasspy/piper) as text-to-speech solution for the talk-llama example. Piper voices sound almost like real people which is a big improvement (e.g.) from something like espeak.
High-performance inference of [OpenAI's Whisper](https://github.com/openai/whisper) automatic speech recognition (ASR) model:
- Plain C/C++ implementation without dependencies
- Apple Silicon first-class citizen - optimized via ARM NEON, Accelerate framework, Metal and [Core ML](https://github.com/ggerganov/whisper.cpp#core-ml-support)
- Apple Silicon first-class citizen - optimized via ARM NEON, Accelerate framework, Metal and [Core ML](#core-ml-support)
- AVX intrinsics support for x86 architectures
- VSX intrinsics support for POWER architectures
- Mixed F16 / F32 precision
- [4-bit and 5-bit integer quantization support](https://github.com/ggerganov/whisper.cpp#quantization)
- [Integer quantization support](#quantization)
- Zero memory allocations at runtime
- [Vulkan support](#vulkan-gpu-support)
- Support for CPU-only inference
- [Efficient GPU support for NVIDIA](https://github.com/ggerganov/whisper.cpp#nvidia-gpu-support-via-cublas)
- [Partial OpenCL GPU support via CLBlast](https://github.com/ggerganov/whisper.cpp#opencl-gpu-support-via-clblast)
- [x] Windows ([MSVC](https://github.com/ggerganov/whisper.cpp/blob/master/.github/workflows/build.yml#L117-L144) and [MinGW](https://github.com/ggerganov/whisper.cpp/issues/168)]
The entire high-level implementation of the model is contained in [whisper.h](whisper.h) and [whisper.cpp](whisper.cpp).
The rest of the code is part of the [ggml](https://github.com/ggerganov/ggml) machine learning library.
The entire high-level implementation of the model is contained in [whisper.h](include/whisper.h) and [whisper.cpp](src/whisper.cpp).
The rest of the code is part of the [`ggml`](https://github.com/ggerganov/ggml) machine learning library.
Having such a lightweight implementation of the model allows to easily integrate it in different platforms and applications.
As an example, here is a video of running the model on an iPhone 13 device - fully offline, on-device: [whisper.objc](examples/whisper.objc)
@ -55,33 +57,39 @@ Or you can even run it straight in the browser: [talk.wasm](examples/talk.wasm)
## Implementation details
- The core tensor operations are implemented in C ([ggml.h](ggml.h) / [ggml.c](ggml.c))
- The transformer model and the high-level C-style API are implemented in C++ ([whisper.h](whisper.h) / [whisper.cpp](whisper.cpp))
- The core tensor operations are implemented in C ([ggml.h](ggml/include/ggml.h) / [ggml.c](ggml/src/ggml.c))
- The transformer model and the high-level C-style API are implemented in C++ ([whisper.h](include/whisper.h) / [whisper.cpp](src/whisper.cpp))
- Sample usage is demonstrated in [main.cpp](examples/main)
- Sample real-time audio transcription from the microphone is demonstrated in [stream.cpp](examples/stream)
- Various other examples are available in the [examples](examples) folder
The tensor operators are optimized heavily for Apple silicon CPUs. Depending on the computation size, Arm Neon SIMD
intrinsics or CBLAS Accelerate framework routines are used. The latter are especially effective for bigger sizes since
the Accelerate framework utilizes the special-purpose AMX coprocessor available in modern Apple products.
The tensor operators are optimized heavily for Apple silicon CPUs. Depending on the computation size, Arm Neon SIMD intrinsics or CBLAS Accelerate framework routines are used. The latter are especially effective for bigger sizes since the Accelerate framework utilizes the special-purpose AMX coprocessor available in modern Apple products.
## Quick start
First clone the repository.
Then, download one of the Whisper models converted in [ggml format](models). For example:
# run the examples as usual, specifying the quantized model file
@ -278,7 +287,8 @@ speed-up - more than x3 faster compared with CPU-only execution. Here are the in
- To ensure `coremltools` operates correctly, please confirm that [Xcode](https://developer.apple.com/xcode/) is installed and execute `xcode-select --install` to install the command-line tools.
- Python 3.10 is recommended.
- [OPTIONAL] It is recommended to utilize a Python version management system, such as [Miniconda](https://docs.conda.io/en/latest/miniconda.html) for this step:
- MacOS Sonoma (version 14) or newer is recommended, as older versions of MacOS might experience issues with transcription hallucination.
- [OPTIONAL] It is recommended to utilize a Python version management system, such as [Miniconda](https://docs.conda.io/en/latest/miniconda.html) for this step:
- To create an environment, use: `conda create -n py310-whisper python=3.10 -y`
- To activate the environment, use: `conda activate py310-whisper`
@ -304,8 +314,8 @@ speed-up - more than x3 faster compared with CPU-only execution. Here are the in
This will produce ggml-base.en-encoder-openvino.xml/.bin IR model files. It's recommended to relocate these to the same folder as ggml models, as that
This will produce ggml-base.en-encoder-openvino.xml/.bin IR model files. It's recommended to relocate these to the same folder as `ggml` models, as that
is the default location that the OpenVINO extension will search at runtime.
- Build `whisper.cpp` with OpenVINO support:
@ -366,24 +378,28 @@ This can result in significant speedup in encoder performance. Here are the inst
After downloading & extracting package onto your development system, set up required environment by sourcing setupvars script. For example:
@ -407,36 +423,23 @@ For more information about the Core ML implementation please refer to PR [#1037]
With NVIDIA cards the processing of the models is done efficiently on the GPU via cuBLAS and custom CUDA kernels.
First, make sure you have installed `cuda`: https://developer.nvidia.com/cuda-downloads
Now build `whisper.cpp` with cuBLAS support:
Now build `whisper.cpp` with CUDA support:
```
make clean
WHISPER_CUBLAS=1 make -j
GGML_CUDA=1 make -j
```
## OpenCL GPU support via CLBlast
For cards and integrated GPUs that support OpenCL, the Encoder processing can be largely offloaded to the GPU through CLBlast. This is especially useful for users with AMD APUs or low end devices for up to ~2x speedup.
First, make sure you have installed `CLBlast` for your OS or Distribution: https://github.com/CNugteren/CLBlast
Now build `whisper.cpp` with CLBlast support:
## Vulkan GPU support
Cross-vendor solution which allows you to accelerate workload on your GPU.
First, make sure your graphics card driver provides support for Vulkan API.
Now build `whisper.cpp` with Vulkan support:
```
Makefile:
cd whisper.cpp
make clean
WHISPER_CLBLAST=1 make -j
CMake:
cd whisper.cpp
cmake -B build -DWHISPER_CLBLAST=ON
cmake --build build -j --config Release
make GGML_VULKAN=1 -j
```
Run all the examples as usual.
## BLAS CPU support via OpenBLAS
Encoder processing can be accelerated on the CPU via OpenBLAS.
@ -446,16 +449,66 @@ Now build `whisper.cpp` with OpenBLAS support:
```
make clean
WHISPER_OPENBLAS=1 make -j
GGML_OPENBLAS=1 make -j
```
## BLAS CPU support via Intel MKL
Encoder processing can be accelerated on the CPU via the BLAS compatible interface of Intel's Math Kernel Library.
First, make sure you have installed Intel's MKL runtime and development packages: https://www.intel.com/content/www/us/en/developer/tools/oneapi/onemkl-download.html
Now build `whisper.cpp` with Intel MKL BLAS support:
```
source /opt/intel/oneapi/setvars.sh
mkdir build
cd build
cmake -DWHISPER_MKL=ON ..
WHISPER_MKL=1 make -j
```
## Ascend NPU support
Ascend NPU provides inference acceleration via [`CANN`](https://www.hiascend.com/en/software/cann) and AI cores.
First, check if your Ascend NPU device is supported:
**Verified devices**
| Ascend NPU | Status |
|:-----------------------------:|:-------:|
| Atlas 300T A2 | Support |
Then, make sure you have installed [`CANN toolkit`](https://www.hiascend.com/en/software/cann/community) . The lasted version of CANN is recommanded.
- If you have trouble with Ascend NPU device, please create a issue with **[CANN]** prefix/tag.
- If you run successfully with your Ascend NPU device, please help update the table `Verified devices`.
## Docker
### Prerequisites
* Docker must be installed and running on your system.
* Create a folder to store big models & intermediate files (ex. /whisper/models)
- Docker must be installed and running on your system.
- Create a folder to store big models & intermediate files (ex. /whisper/models)
### Images
We have two Docker images available for this project:
1. `ghcr.io/ggerganov/whisper.cpp:main`: This image includes the main executable file as well as `curl` and `ffmpeg`. (platforms: `linux/amd64`, `linux/arm64`)
Use the [extra/bench-wts.sh](https://github.com/ggerganov/whisper.cpp/blob/master/extra/bench-wts.sh) script to generate a video in the following format:
Use the [scripts/bench-wts.sh](https://github.com/ggerganov/whisper.cpp/blob/master/scripts/bench-wts.sh) script to generate a video in the following format:
```java
./extra/bench-wts.sh samples/jfk.wav
```bash
./scripts/bench-wts.sh samples/jfk.wav
ffplay ./samples/jfk.wav.all.mp4
```
@ -739,20 +803,19 @@ took to execute it. The results are summarized in the following Github issue:
| [whisper.android](examples/whisper.android) | | Android mobile application using whisper.cpp |
| [whisper.nvim](examples/whisper.nvim) | | Speech-to-text plugin for Neovim |
| [generate-karaoke.sh](examples/generate-karaoke.sh) | | Helper script to easily [generate a karaoke video](https://youtu.be/uj7hVta4blM) of raw audio capture |
| [whisper.android](examples/whisper.android) | | Android mobile application using whisper.cpp |
| [whisper.nvim](examples/whisper.nvim) | | Speech-to-text plugin for Neovim |
| [generate-karaoke.sh](examples/generate-karaoke.sh) | | Helper script to easily [generate a karaoke video](https://youtu.be/uj7hVta4blM) of raw audio capture |
SYCL is a higher-level programming model to improve programming productivity on various hardware accelerators<72>such as CPUs, GPUs, and FPGAs. It is a single-source embedded domain-specific language based on pure C++17.
oneAPI is a specification that is open and standards-based, supporting multiple architecture types including but not limited to GPU, CPU, and FPGA. The spec has both direct programming and API-based programming paradigms.
Intel uses the SYCL as direct programming language to support CPU, GPUs and FPGAs.
To avoid re-inventing the wheel, this code refers other code paths in llama.cpp (like OpenBLAS, cuBLAS, CLBlast). We use a open-source tool [SYCLomatic](https://github.com/oneapi-src/SYCLomatic) (Commercial release [Intel<EFBFBD> DPC++ Compatibility Tool](https://www.intel.com/content/www/us/en/developer/tools/oneapi/dpc-compatibility-tool.html)) migrate to SYCL.
The whisper.cpp for SYCL is used to support Intel GPUs.
For Intel CPU, recommend to use whisper.cpp for X86 (Intel MKL build).
## OS
|OS|Status|Verified|
|-|-|-|
|Linux|Support|Ubuntu 22.04|
|Windows|Ongoing| |
## Intel GPU
|Intel GPU| Status | Verified Model|
|-|-|-|
|Intel Data Center Max Series| Support| Max 1550|
|Intel Data Center Flex Series| Support| Flex 170|
a. Please follow the procedure in [Get the Intel<65> oneAPI Base Toolkit ](https://www.intel.com/content/www/us/en/developer/tools/oneapi/base-toolkit.html).
Recommend to install to default folder: **/opt/intel/oneapi**.
Following guide use the default folder as example. If you use other folder, please modify the following guide info with your folder.
b. Check
```
source /opt/intel/oneapi/setvars.sh
sycl-ls
```
There should be one or more level-zero devices. Like **[ext_oneapi_level_zero:gpu:0]**.
fprintf(stderr,"ggml_backend_tensor_copy: neither cpy_tensor_from nor cpy_tensor_to are implemented for backends %s and %s, falling back to get/set\n",ggml_backend_name(src->buffer->backend),ggml_backend_name(dst->buffer->backend));
// pass 4: split graph, find tensors that need to be copied
// TODO:
// - when switching from a less preferred backend to a more preferred backend, check if it is possible to move the switch to an earlier point for the same cost
# As of cmake 3.27, there is no official cmake support for FindFFmpeg.
# Consequnelty we added a FindFFmpeg.cmake script the cmake subfolder:
# whisper.cpp does not need the full ffmpeg libs, just AVFORMAT AVCODEC AVUTIL SWRESAMPLE
# libswresample performs highly optimized audio resampling, rematrixing and sample format conversion operations
# libavcodec provides a generic encoding/decoding framework and contains multiple decoders and encoders for audio, video and subtitle streams, and several bitstream filters.
# libavformat provides a generic framework for multiplexing and demultiplexing (muxing and demuxing) audio, video and subtitle streams.
Processes an audio file using a specified model and returns the processed string.
:param wav_file: Path to the WAV file
:param model_name: Name of the model to use
:return: Processed string output from the audio processing
:raises: Exception if an error occurs during processing
"""
model=f"./models/ggml-{model_name}.bin"
# Check if the file exists
ifnotos.path.exists(model):
raiseFileNotFoundError(f"Model file not found: {model}\n\nDownload a model with this command:\n\n> bash ./models/download-ggml-model.sh {model_name}\n\n")
ifnotos.path.exists(wav_file):
raiseFileNotFoundError(f"WAV file not found: {wav_file}")
--convert, [false ] Convert audio to WAV, requires ffmpeg on the server
```
> [!WARNING]
> [!WARNING]
> **Do not run the server example with administrative privileges and ensure it's operated in a sandbox environment, especially since it involves risky operations like accepting user file uploads and using ffmpeg for format conversions. Always validate and sanitize inputs to guard against potential security threats.**
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
