diff --git a/base-foc/mk/spec-foc_arndale.mk b/base-foc/mk/spec-foc_arndale.mk
index 2425d44a85..6a76d78de7 100644
--- a/base-foc/mk/spec-foc_arndale.mk
+++ b/base-foc/mk/spec-foc_arndale.mk
@@ -1,4 +1,5 @@
 SPECS += foc_arm platform_arndale uboot
 
+include $(call select_from_repositories,mk/spec-fpu_vfpv3.mk)
 include $(call select_from_repositories,mk/spec-platform_arndale.mk)
 include $(call select_from_repositories,mk/spec-foc_arm.mk)
diff --git a/base/include/arm/vfp/cpu/string.h b/base/include/arm/vfp/cpu/string.h
new file mode 100644
index 0000000000..c4a1b1bdc3
--- /dev/null
+++ b/base/include/arm/vfp/cpu/string.h
@@ -0,0 +1,69 @@
+/*
+ * \brief  ARM-specific memcpy using VFP
+ * \author Sebastian Sumpf
+ * \date   2013-06-19
+ *
+ * Should work for VFPv2, VFPv3, and Advanced SIMD.
+ */
+
+/*
+ * Copyright (C) 2012-2013 Genode Labs GmbH
+ *
+ * This file is part of the Genode OS framework, which is distributed
+ * under the terms of the GNU General Public License version 2.
+ */
+
+#ifndef _INCLUDE__ARM__VFP__STRING_H_
+#define _INCLUDE__ARM__VFP__STRING_H_
+
+namespace Genode
+{
+	/**
+	 * Copy memory block
+	 *
+	 * \param dst   destination memory block
+	 * \param src   source memory block
+	 * \param size  number of bytes to copy
+	 *
+	 * \return      Number of bytes not copied
+	 */
+	inline size_t memcpy_cpu(void *dst, const void *src, size_t size)
+	{
+		unsigned char *d = (unsigned char *)dst, *s = (unsigned char *)src;
+		/* check 4 byte; alignment */
+		size_t d_align = (size_t)d & 0x3;
+		size_t s_align = (size_t)s & 0x3;
+
+		/* only same alignments work for the following loops */
+		if (d_align != s_align)
+			return size;
+
+		/* copy to 4 byte alignment */
+		for (; (size > 0) && (s_align > 0) && (s_align < 4);
+		     s_align++, *d++ = *s++, size--);
+
+		/* copy 64 byte chunks using FPU */
+		for (; size >= 64; size -= 64)
+			asm volatile ("pld [%0, #0xc0]  \n\t"
+			              "vldm %0!,{d0-d7} \n\t"
+			              "vstm %1!,{d0-d7} \n\t"
+			              : "+r"(s), "+r" (d)
+                    :: "d0", "d1", "d2", "d3", "d4", "d5", "d6", "d7");
+
+		/* copy left over 32 byte chunk */
+		for (; size >= 32; size -= 32)
+			asm volatile ("ldmia %0!, {r3 - r10} \n\t"
+			              "stmia %1!, {r3 - r10} \n\t"
+			              : "+r" (s), "+r" (d)
+			              :: "r3","r4","r5","r6","r7","r8","r9","r10");
+
+		for(; size >= 4; size -= 4)
+			asm volatile ("ldr r3, [%0], #4 \n\t"
+			              "str r3, [%1], #4 \n\t"
+			              : "+r" (s), "+r" (d)
+			              :: "r3");
+		return size;
+	}
+}
+
+#endif /* _INCLUDE__ARM__VFP__STRING_H_ */
diff --git a/base/mk/spec-fpu_vfpv3.mk b/base/mk/spec-fpu_vfpv3.mk
new file mode 100644
index 0000000000..a4b7ae4a6c
--- /dev/null
+++ b/base/mk/spec-fpu_vfpv3.mk
@@ -0,0 +1,15 @@
+#
+# \brief  Enable VFPV3-FPU on ARM
+# \author Sebastian Sumpf
+# \date   2013-06-16
+#
+
+#
+# Enable floating point support in compiler
+#
+CC_MARCH    += -mfpu=vfpv3 -mfloat-abi=softfp
+
+#
+# Include floating-point unit code
+#
+REP_INC_DIR += include/arm/vfp