diff --git a/.devops/nix/package.nix b/.devops/nix/package.nix
index cd6146421..b88e6ca80 100644
--- a/.devops/nix/package.nix
+++ b/.devops/nix/package.nix
@@ -126,9 +126,9 @@ effectiveStdenv.mkDerivation (finalAttrs: {
   };
 
   postPatch = ''
-    substituteInPlace ./ggml/src/ggml-metal.m \
+    substituteInPlace ./ggml/src/ggml-metal/ggml-metal.m \
       --replace '[bundle pathForResource:@"ggml-metal" ofType:@"metal"];' "@\"$out/bin/ggml-metal.metal\";"
-    substituteInPlace ./ggml/src/ggml-metal.m \
+    substituteInPlace ./ggml/src/ggml-metal/ggml-metal.m \
       --replace '[bundle pathForResource:@"default" ofType:@"metallib"];' "@\"$out/bin/default.metallib\";"
   '';
 
diff --git a/Makefile b/Makefile
index fc6ef46cd..5ec996c87 100644
--- a/Makefile
+++ b/Makefile
@@ -523,11 +523,11 @@ ifndef GGML_NO_ACCELERATE
 	# Mac OS - include Accelerate framework.
 	# `-framework Accelerate` works both with Apple Silicon and Mac Intel
 	ifeq ($(UNAME_S),Darwin)
-		MK_CPPFLAGS += -DGGML_USE_ACCELERATE -DGGML_USE_BLAS
+		MK_CPPFLAGS += -DGGML_USE_ACCELERATE -DGGML_USE_BLAS -DGGML_BLAS_USE_ACCELERATE
 		MK_CPPFLAGS += -DACCELERATE_NEW_LAPACK
 		MK_CPPFLAGS += -DACCELERATE_LAPACK_ILP64
 		MK_LDFLAGS  += -framework Accelerate
-		OBJ_GGML    += ggml/src/ggml-blas.o
+		OBJ_GGML    += ggml/src/ggml-blas/ggml-blas.o
 	endif
 endif # GGML_NO_ACCELERATE
 
@@ -552,36 +552,36 @@ ifdef GGML_OPENBLAS
 	MK_CPPFLAGS += -DGGML_USE_BLAS $(shell pkg-config --cflags-only-I openblas)
 	MK_CFLAGS   += $(shell pkg-config --cflags-only-other openblas)
 	MK_LDFLAGS  += $(shell pkg-config --libs openblas)
-	OBJ_GGML    += ggml/src/ggml-blas.o
+	OBJ_GGML    += ggml/src/ggml-blas/ggml-blas.o
 endif # GGML_OPENBLAS
 
 ifdef GGML_OPENBLAS64
 	MK_CPPFLAGS += -DGGML_USE_BLAS $(shell pkg-config --cflags-only-I openblas64)
 	MK_CFLAGS   += $(shell pkg-config --cflags-only-other openblas64)
 	MK_LDFLAGS  += $(shell pkg-config --libs openblas64)
-	OBJ_GGML    += ggml/src/ggml-blas.o
+	OBJ_GGML    += src/ggml-blas/ggml-blas.o
 endif # GGML_OPENBLAS64
 
 ifdef GGML_BLIS
 	MK_CPPFLAGS += -DGGML_USE_BLAS -DGGML_BLAS_USE_BLIS -I/usr/local/include/blis -I/usr/include/blis
 	MK_LDFLAGS  += -lblis -L/usr/local/lib
-	OBJ_GGML    += ggml/src/ggml-blas.o
+	OBJ_GGML    += src/ggml-blas/ggml-blas.o
 endif # GGML_BLIS
 
 ifdef GGML_NVPL
 	MK_CPPFLAGS += -DGGML_USE_BLAS -DGGML_BLAS_USE_NVPL -DNVPL_ILP64 -I/usr/local/include/nvpl_blas -I/usr/include/nvpl_blas
 	MK_LDFLAGS  += -L/usr/local/lib -lnvpl_blas_core -lnvpl_blas_ilp64_gomp
-	OBJ_GGML    += ggml/src/ggml-blas.o
+	OBJ_GGML    += src/ggml-blas/ggml-blas.o
 endif # GGML_NVPL
 
 ifndef GGML_NO_LLAMAFILE
 	MK_CPPFLAGS += -DGGML_USE_LLAMAFILE
-	OBJ_GGML    += ggml/src/llamafile/sgemm.o
+	OBJ_GGML    += ggml/src/ggml-cpu/llamafile/sgemm.o
 endif
 
 ifndef GGML_NO_AMX
 	MK_CPPFLAGS += -DGGML_USE_AMX
-	OBJ_GGML    += ggml/src/ggml-amx.o ggml/src/ggml-amx/mmq.o
+	OBJ_GGML    += ggml/src/ggml-amx/ggml-amx.o ggml/src/ggml-amx/mmq.o
 endif
 
 ifdef GGML_RPC
@@ -623,7 +623,7 @@ ifdef GGML_CUDA
 		MK_NVCCFLAGS += -use_fast_math
 	endif # GGML_MUSA
 
-	OBJ_GGML += ggml/src/ggml-cuda.o
+	OBJ_GGML += ggml/src/ggml-cuda/ggml-cuda.o
 	OBJ_GGML += $(patsubst %.cu,%.o,$(wildcard ggml/src/ggml-cuda/*.cu))
 	OBJ_GGML += $(OBJ_CUDA_TMPL)
 
@@ -742,8 +742,8 @@ ggml/src/ggml-cuda/%.o: \
 	ggml/src/ggml-cuda/common.cuh
 	$(NVCC_COMPILE)
 
-ggml/src/ggml-cuda.o: \
-	ggml/src/ggml-cuda.cu \
+ggml/src/ggml-cuda/ggml-cuda.o: \
+	ggml/src/ggml-cuda/ggml-cuda.cu \
 	ggml/include/ggml-cuda.h \
 	ggml/include/ggml.h \
 	ggml/include/ggml-backend.h \
@@ -852,12 +852,12 @@ ifdef GGML_CUDA_NO_PEER_COPY
 	HIPFLAGS += -DGGML_CUDA_NO_PEER_COPY
 endif # GGML_CUDA_NO_PEER_COPY
 
-	OBJ_GGML += ggml/src/ggml-cuda.o
+	OBJ_GGML += ggml/src/ggml-cuda/ggml-cuda.o
 	OBJ_GGML += $(patsubst %.cu,%.o,$(wildcard ggml/src/ggml-cuda/*.cu))
 	OBJ_GGML += $(OBJ_CUDA_TMPL)
 
-ggml/src/ggml-cuda.o: \
-	ggml/src/ggml-cuda.cu \
+ggml/src/ggml-cuda/ggml-cuda.o: \
+	ggml/src/ggml-cuda/ggml-cuda.cu \
 	ggml/include/ggml-cuda.h \
 	ggml/include/ggml.h \
 	ggml/include/ggml-backend.h \
@@ -877,7 +877,7 @@ endif # GGML_HIPBLAS
 ifdef GGML_METAL
 	MK_CPPFLAGS += -DGGML_USE_METAL
 	MK_LDFLAGS  += -framework Foundation -framework Metal -framework MetalKit
-	OBJ_GGML	+= ggml/src/ggml-metal.o
+	OBJ_GGML	+= ggml/src/ggml-metal/ggml-metal.o
 
 ifdef GGML_METAL_USE_BF16
 	MK_CPPFLAGS += -DGGML_METAL_USE_BF16
@@ -892,18 +892,18 @@ endif
 endif # GGML_METAL
 
 ifdef GGML_METAL
-ggml/src/ggml-metal.o: \
-	ggml/src/ggml-metal.m \
+ggml/src/ggml-metal/ggml-metal.o: \
+	ggml/src/ggml-metal/ggml-metal.m \
 	ggml/include/ggml-metal.h \
 	ggml/include/ggml.h
 	$(CC) $(CFLAGS) -c $< -o $@
 
 ifdef GGML_METAL_EMBED_LIBRARY
 ggml/src/ggml-metal-embed.o: \
-	ggml/src/ggml-metal.metal \
+	ggml/src/ggml-metal/ggml-metal.metal \
 	ggml/src/ggml-common.h
 	@echo "Embedding Metal library"
-	@sed -e '/#include "ggml-common.h"/r ggml/src/ggml-common.h' -e '/#include "ggml-common.h"/d' < ggml/src/ggml-metal.metal > ggml/src/ggml-metal-embed.metal
+	@sed -e '/#include "ggml-common.h"/r ggml/src/ggml-common.h' -e '/#include "ggml-common.h"/d' < ggml/src/ggml-metal/ggml-metal.metal > ggml/src/ggml-metal/ggml-metal-embed.metal
 	$(eval TEMP_ASSEMBLY=$(shell mktemp -d))
 	@echo ".section __DATA, __ggml_metallib"            >  $(TEMP_ASSEMBLY)/ggml-metal-embed.s
 	@echo ".globl _ggml_metallib_start"                 >> $(TEMP_ASSEMBLY)/ggml-metal-embed.s
@@ -919,11 +919,16 @@ endif # GGML_METAL
 
 OBJ_GGML += \
 	ggml/src/ggml.o \
-	ggml/src/ggml-cpu.o \
+	ggml/src/ggml-aarch64.o \
 	ggml/src/ggml-alloc.o \
 	ggml/src/ggml-backend.o \
+	ggml/src/ggml-backend-reg.o \
 	ggml/src/ggml-quants.o \
-	ggml/src/ggml-aarch64.o
+	ggml/src/ggml-threading.o \
+	ggml/src/ggml-cpu/ggml-cpu.o \
+	ggml/src/ggml-cpu/ggml-cpu-cpp.o \
+	ggml/src/ggml-cpu/ggml-cpu-aarch64.o \
+	ggml/src/ggml-cpu/ggml-cpu-quants.o
 
 OBJ_LLAMA = \
 	src/llama.o \
@@ -1051,12 +1056,23 @@ ggml/src/ggml.o: \
 	ggml/include/ggml.h
 	$(CC)  $(CFLAGS)   -c $< -o $@
 
-ggml/src/ggml-cpu.o: \
-	ggml/src/ggml-cpu.c \
+ggml/src/ggml-threading.o: \
+	ggml/src/ggml-threading.cpp \
+	ggml/include/ggml.h
+	$(CXX) $(XXCFLAGS)   -c $< -o $@
+
+ggml/src/ggml-cpu/ggml-cpu.o: \
+	ggml/src/ggml-cpu/ggml-cpu.c \
 	ggml/include/ggml.h \
 	ggml/src/ggml-common.h
 	$(CC)  $(CFLAGS)   -c $< -o $@
 
+ggml/src/ggml-cpu/ggml-cpu-cpp.o: \
+	ggml/src/ggml-cpu/ggml-cpu.cpp \
+	ggml/include/ggml.h \
+	ggml/src/ggml-common.h
+	$(CXX) $(CXXFLAGS)   -c $< -o $@
+
 ggml/src/ggml-alloc.o: \
 	ggml/src/ggml-alloc.c \
 	ggml/include/ggml.h \
@@ -1084,22 +1100,22 @@ ggml/src/ggml-aarch64.o: \
 	ggml/src/ggml-common.h
 	$(CC) $(CFLAGS)    -c $< -o $@
 
-ggml/src/ggml-blas.o: \
-	ggml/src/ggml-blas.cpp \
+ggml/src/ggml-blas/ggml-blas.o: \
+	ggml/src/ggml-blas/ggml-blas.cpp \
 	ggml/include/ggml-blas.h
 	$(CXX) $(CXXFLAGS) -c $< -o $@
 
 ifndef GGML_NO_LLAMAFILE
-ggml/src/llamafile/sgemm.o: \
-	ggml/src/llamafile/sgemm.cpp \
-	ggml/src/llamafile/sgemm.h \
+ggml/src/ggml-cpu/llamafile/sgemm.o: \
+	ggml/src/ggml-cpu/llamafile/sgemm.cpp \
+	ggml/src/ggml-cpu/llamafile/sgemm.h \
 	ggml/include/ggml.h
-	$(CXX) $(CXXFLAGS) -c $< -o $@
+	$(CXX) $(CXXFLAGS) -c $< -o $@ -I ggml/src -I ggml/src/ggml-cpu
 endif # GGML_NO_LLAMAFILE
 
 ifndef GGML_NO_AMX
-ggml/src/ggml-amx.o: \
-	ggml/src/ggml-amx.cpp \
+ggml/src/ggml-amx/ggml-amx.o: \
+	ggml/src/ggml-amx/ggml-amx.cpp \
 	ggml/include/ggml-amx.h
 	$(CXX) $(CXXFLAGS) -c $< -o $@
 
@@ -1250,10 +1266,11 @@ clean:
 	rm -rvf ggml/*.a
 	rm -rvf ggml/*.dll
 	rm -rvf ggml/*.so
-	rm -vrf ggml/src/*.o
-	rm -rvf ggml/src/llamafile/*.o
+	rm -rvf ggml/src/*.o
+	rm -rvf ggml/src/ggml-cpu/*.o
+	rm -rvf ggml/src/ggml-cpu/llamafile/*.o
 	rm -rvf common/build-info.cpp
-	rm -vrf ggml/src/ggml-metal-embed.metal
+	rm -vrf ggml/src/ggml-metal/ggml-metal-embed.metal
 	rm -vrf ggml/src/ggml-cuda/*.o
 	rm -vrf ggml/src/ggml-cuda/template-instances/*.o
 	rm -vrf ggml/src/ggml-amx/*.o
diff --git a/Package.swift b/Package.swift
index 0f4f19018..eaa17cdb2 100644
--- a/Package.swift
+++ b/Package.swift
@@ -30,8 +30,8 @@ var cSettings: [CSetting] =  [
 ]
 
 #if canImport(Darwin)
-sources.append("ggml/src/ggml-metal.m")
-resources.append(.process("ggml/src/ggml-metal.metal"))
+sources.append("ggml/src/ggml-metal/ggml-metal.m")
+resources.append(.process("ggml/src/ggml-metal/ggml-metal.metal"))
 linkerSettings.append(.linkedFramework("Accelerate"))
 cSettings.append(
     contentsOf: [
diff --git a/ggml/src/ggml-amx/ggml-amx.cpp b/ggml/src/ggml-amx/ggml-amx.cpp
index 2db0b720a..37da98539 100644
--- a/ggml/src/ggml-amx/ggml-amx.cpp
+++ b/ggml/src/ggml-amx/ggml-amx.cpp
@@ -421,9 +421,18 @@ ggml_backend_reg_t ggml_backend_amx_reg(void) {
 
 #else // if defined(__AMX_INT8__)
 
+ggml_backend_buffer_type_t ggml_backend_amx_buffer_type(void) {
+    return nullptr;
+}
+
+bool ggml_backend_is_amx(ggml_backend_t backend) {
+    GGML_UNUSED(backend);
+    return false;
+}
+
 ggml_backend_t ggml_backend_amx_init(void) {
     fprintf(stderr, "GGML is not compiled with AMX support!\n");
-    return ggml_backend_t{};
+    return nullptr;
 }
 
 void ggml_backend_amx_set_n_threads(ggml_backend_t backend_amx, int n_threads) {
diff --git a/ggml/src/ggml-cpu/ggml-cpu.c b/ggml/src/ggml-cpu/ggml-cpu.c
index bf2bbdef6..4c45146a1 100644
--- a/ggml/src/ggml-cpu/ggml-cpu.c
+++ b/ggml/src/ggml-cpu/ggml-cpu.c
@@ -44,7 +44,7 @@
 #endif
 
 #ifdef GGML_USE_LLAMAFILE
-#include <llamafile/sgemm.h>
+#include "llamafile/sgemm.h"
 #endif
 
 #if defined(_MSC_VER)
diff --git a/ggml/src/ggml-impl.h b/ggml/src/ggml-impl.h
index a040e4b56..aa4d2b85d 100644
--- a/ggml/src/ggml-impl.h
+++ b/ggml/src/ggml-impl.h
@@ -22,6 +22,10 @@
 #include <arm_neon.h>
 #endif
 
+#if defined(__F16C__)
+#include <immintrin.h>
+#endif
+
 #ifdef __cplusplus
 extern "C" {
 #endif
diff --git a/ggml/src/ggml-metal/ggml-metal.metal b/ggml/src/ggml-metal/ggml-metal.metal
index e8b71a9f8..88f06c24e 100644
--- a/ggml/src/ggml-metal/ggml-metal.metal
+++ b/ggml/src/ggml-metal/ggml-metal.metal
@@ -15,8 +15,8 @@ using namespace metal;
 // ref: https://developer.apple.com/metal/Metal-Shading-Language-Specification.pdf
 //
 // cmd:
-//   .../usr/bin/metal -dM -E -c                             ggml/src/ggml-metal.metal
-//   .../usr/bin/metal -dM -E -c -target air64-apple-ios14.0 ggml/src/ggml-metal.metal
+//   .../usr/bin/metal -dM -E -c                             ggml/src/ggml-metal/ggml-metal.metal
+//   .../usr/bin/metal -dM -E -c -target air64-apple-ios14.0 ggml/src/ggml-metal/ggml-metal.metal
 //
 #if __METAL_VERSION__ < 310 && defined(GGML_METAL_USE_BF16)
 #undef GGML_METAL_USE_BF16
diff --git a/ggml/src/ggml.c b/ggml/src/ggml.c
index 11f0803c4..4a97bfc32 100644
--- a/ggml/src/ggml.c
+++ b/ggml/src/ggml.c
@@ -368,7 +368,7 @@ void ggml_fp16_to_fp32_row(const ggml_fp16_t * x, float * y, int64_t n) {
 void ggml_fp32_to_fp16_row(const float * x, ggml_fp16_t * y, int64_t n) {
     int64_t i = 0;
 #if defined(__F16C__)
-    if (ggml_cpu_has_f16c()) {
+    //if (ggml_cpu_has_f16c()) {
         for (; i + 7 < n; i += 8) {
             __m256 x_vec = _mm256_loadu_ps(x + i);
             __m128i y_vec = _mm256_cvtps_ph(x_vec, _MM_FROUND_TO_NEAREST_INT);
@@ -379,7 +379,7 @@ void ggml_fp32_to_fp16_row(const float * x, ggml_fp16_t * y, int64_t n) {
             __m128i y_vec = _mm_cvtps_ph(x_vec, _MM_FROUND_TO_NEAREST_INT);
             _mm_storel_epi64((__m128i *)(y + i), y_vec);
         }
-    }
+    //}
 #endif
     for (; i < n; i++) {
         y[i] = GGML_FP32_TO_FP16(x[i]);
@@ -389,7 +389,7 @@ void ggml_fp32_to_fp16_row(const float * x, ggml_fp16_t * y, int64_t n) {
 void ggml_bf16_to_fp32_row(const ggml_bf16_t * x, float * y, int64_t n) {
     int64_t i = 0;
 #if defined(__AVX512F__)
-    if (ggml_cpu_has_avx512()) {
+    //if (ggml_cpu_has_avx512()) {
         for (; i + 16 <= n; i += 16) {
             _mm512_storeu_ps(y + i,
                             _mm512_castsi512_ps(
@@ -399,10 +399,10 @@ void ggml_bf16_to_fp32_row(const ggml_bf16_t * x, float * y, int64_t n) {
                                             (const __m256i *)(x + i))),
                                     16)));
         }
-    }
+    //}
 #endif
 #if defined(__AVX2__)
-    if (ggml_cpu_has_avx2()) {
+    //if (ggml_cpu_has_avx2()) {
         for (; i + 8 <= n; i += 8) {
             _mm256_storeu_ps(y + i,
                             _mm256_castsi256_ps(
@@ -412,7 +412,7 @@ void ggml_bf16_to_fp32_row(const ggml_bf16_t * x, float * y, int64_t n) {
                                             (const __m128i *)(x + i))),
                                     16)));
         }
-    }
+    //}
 #endif
     for (; i < n; i++) {
         y[i] = GGML_BF16_TO_FP32(x[i]);
diff --git a/pocs/vdot/vdot.cpp b/pocs/vdot/vdot.cpp
index 67c6964b6..2dca62848 100644
--- a/pocs/vdot/vdot.cpp
+++ b/pocs/vdot/vdot.cpp
@@ -237,7 +237,6 @@ int main(int argc, char** argv) {
     int n4 = useQ4_1 ? kVecSize / QK4_1 : kVecSize / QK4_0; n4 = 64*((n4 + 63)/64);
     int n8 = kVecSize / QK8_0; n8 = 64*((n8 + 63)/64);
 
-    const auto * funcs = ggml_get_type_traits(useQ4_1 ? GGML_TYPE_Q4_1 : GGML_TYPE_Q4_0);
     const auto * funcs_cpu = ggml_get_type_traits_cpu(useQ4_1 ? GGML_TYPE_Q4_1 : GGML_TYPE_Q4_0);
 
     std::vector<block_q4_0> q40;