deps(llama.cpp): bump to latest, update build variables (#2669)

* arrow_up: Update ggerganov/llama.cpp

Signed-off-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Signed-off-by: Ettore Di Giacinto <[email protected]>

* deps(llama.cpp): update build variables to follow upstream

Update build recipes with ggerganov/llama.cpp#8006

Signed-off-by: Ettore Di Giacinto <[email protected]>

* Disable shared libs by default in llama.cpp

Signed-off-by: Ettore Di Giacinto <[email protected]>

* Disable shared libs in llama.cpp Makefile

Signed-off-by: Ettore Di Giacinto <[email protected]>

* Disable metal embedding for now, until it is tested

Signed-off-by: Ettore Di Giacinto <[email protected]>

* fix(mac): explicitly enable metal

Signed-off-by: Ettore Di Giacinto <[email protected]>

* debug

Signed-off-by: Ettore Di Giacinto <[email protected]>

* fix typo

Signed-off-by: Ettore Di Giacinto <[email protected]>

---------

Signed-off-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Signed-off-by: Ettore Di Giacinto <[email protected]>
Co-authored-by: mudler <[email protected]>

.github/workflows/test.yml

@@ -220,7 +220,7 @@ jobs:
           export CPLUS_INCLUDE_PATH=/usr/local/include
           # Used to run the newer GNUMake version from brew that supports --output-sync
           export PATH="/opt/homebrew/opt/make/libexec/gnubin:$PATH"
-          BUILD_TYPE="GITHUB_CI_HAS_BROKEN_METAL" CMAKE_ARGS="-DLLAMA_F16C=OFF -DLLAMA_AVX512=OFF -DLLAMA_AVX2=OFF -DLLAMA_FMA=OFF" make --jobs 4 --output-sync=target test
+          BUILD_TYPE="GITHUB_CI_HAS_BROKEN_METAL" CMAKE_ARGS="-DGGML_F16C=OFF -DGGML_AVX512=OFF -DGGML_AVX2=OFF -DGGML_FMA=OFF" make --jobs 4 --output-sync=target test
       - name: Setup tmate session if tests fail
         if: ${{ failure() }}
         uses: mxschmitt/[email protected]

Makefile

@@ -5,7 +5,7 @@ BINARY_NAME=local-ai
 
 # llama.cpp versions
 GOLLAMA_STABLE_VERSION?=2b57a8ae43e4699d3dc5d1496a1ccd42922993be
-CPPLLAMA_VERSION?=e112b610a1a75cb7fa8351e1a933e2e7a755a5ce
+CPPLLAMA_VERSION?=ae5d0f4b899ff2842bfca561370c945ad8d4368b
 
 # gpt4all version
 GPT4ALL_REPO?=https://github.com/nomic-ai/gpt4all
@@ -80,8 +80,8 @@ ifeq ($(OS),Darwin)
 		BUILD_TYPE=metal
 	# disable metal if on Darwin and any other value is explicitly passed.
 	else ifneq ($(BUILD_TYPE),metal)
-		CMAKE_ARGS+=-DLLAMA_METAL=OFF
-		export LLAMA_NO_ACCELERATE=1
+		CMAKE_ARGS+=-DGGML_METAL=OFF
+		export GGML_NO_ACCELERATE=1
 	endif
 
 	ifeq ($(BUILD_TYPE),metal)
@@ -98,13 +98,13 @@ endif
 
 ifeq ($(BUILD_TYPE),cublas)
 	CGO_LDFLAGS+=-lcublas -lcudart -L$(CUDA_LIBPATH)
-	export LLAMA_CUBLAS=1
+	export GGML_CUDA=1
 	export WHISPER_CUDA=1
 	CGO_LDFLAGS_WHISPER+=-L$(CUDA_LIBPATH)/stubs/ -lcuda -lcufft
 endif
 
 ifeq ($(BUILD_TYPE),vulkan)
-	CMAKE_ARGS+=-DLLAMA_VULKAN=1
+	CMAKE_ARGS+=-DGGML_VULKAN=1
 endif
 
 ifeq ($(BUILD_TYPE),hipblas)
@@ -118,13 +118,13 @@ ifeq ($(BUILD_TYPE),hipblas)
 	export WHISPER_HIPBLAS=1
 	GPU_TARGETS ?= gfx900,gfx906,gfx908,gfx940,gfx941,gfx942,gfx90a,gfx1030,gfx1031,gfx1100,gfx1101
 	AMDGPU_TARGETS ?= "$(GPU_TARGETS)"
-	CMAKE_ARGS+=-DLLAMA_HIPBLAS=ON -DAMDGPU_TARGETS="$(AMDGPU_TARGETS)" -DGPU_TARGETS="$(GPU_TARGETS)"
+	CMAKE_ARGS+=-DGGML_HIPBLAS=ON -DAMDGPU_TARGETS="$(AMDGPU_TARGETS)" -DGPU_TARGETS="$(GPU_TARGETS)"
 	CGO_LDFLAGS += -O3 --rtlib=compiler-rt -unwindlib=libgcc -lhipblas -lrocblas --hip-link -L${ROCM_HOME}/lib/llvm/lib
 endif
 
 ifeq ($(BUILD_TYPE),metal)
 	CGO_LDFLAGS+=-framework Foundation -framework Metal -framework MetalKit -framework MetalPerformanceShaders
-	export LLAMA_METAL=1
+	export GGML_METAL=1
 	export WHISPER_METAL=1
 endif
 
@@ -354,7 +354,7 @@ else
 endif
 
 dist-cross-linux-arm64: 
-	CMAKE_ARGS="$(CMAKE_ARGS) -DLLAMA_NATIVE=off" GRPC_BACKENDS="backend-assets/grpc/llama-cpp-fallback backend-assets/grpc/llama-cpp-grpc backend-assets/util/llama-cpp-rpc-server" \
+	CMAKE_ARGS="$(CMAKE_ARGS) -DGGML_NATIVE=off" GRPC_BACKENDS="backend-assets/grpc/llama-cpp-fallback backend-assets/grpc/llama-cpp-grpc backend-assets/util/llama-cpp-rpc-server" \
 	STATIC=true $(MAKE) build
 	mkdir -p release
 # if BUILD_ID is empty, then we don't append it to the binary name
@@ -711,21 +711,21 @@ backend-assets/grpc/llama-cpp-avx2: backend-assets/grpc
 	cp -rf backend/cpp/llama backend/cpp/llama-avx2
 	$(MAKE) -C backend/cpp/llama-avx2 purge
 	$(info ${GREEN}I llama-cpp build info:avx2${RESET})
-	CMAKE_ARGS="$(CMAKE_ARGS) -DLLAMA_AVX=on -DLLAMA_AVX2=on -DLLAMA_AVX512=off -DLLAMA_FMA=on -DLLAMA_F16C=on" $(MAKE) VARIANT="llama-avx2" build-llama-cpp-grpc-server
+	CMAKE_ARGS="$(CMAKE_ARGS) -DGGML_AVX=on -DGGML_AVX2=on -DGGML_AVX512=off -DGGML_FMA=on -DGGML_F16C=on" $(MAKE) VARIANT="llama-avx2" build-llama-cpp-grpc-server
 	cp -rfv backend/cpp/llama-avx2/grpc-server backend-assets/grpc/llama-cpp-avx2
 
 backend-assets/grpc/llama-cpp-avx: backend-assets/grpc
 	cp -rf backend/cpp/llama backend/cpp/llama-avx
 	$(MAKE) -C backend/cpp/llama-avx purge
 	$(info ${GREEN}I llama-cpp build info:avx${RESET})
-	CMAKE_ARGS="$(CMAKE_ARGS) -DLLAMA_AVX=on -DLLAMA_AVX2=off -DLLAMA_AVX512=off -DLLAMA_FMA=off -DLLAMA_F16C=off" $(MAKE) VARIANT="llama-avx" build-llama-cpp-grpc-server
+	CMAKE_ARGS="$(CMAKE_ARGS) -DGGML_AVX=on -DGGML_AVX2=off -DGGML_AVX512=off -DGGML_FMA=off -DGGML_F16C=off" $(MAKE) VARIANT="llama-avx" build-llama-cpp-grpc-server
 	cp -rfv backend/cpp/llama-avx/grpc-server backend-assets/grpc/llama-cpp-avx
 
 backend-assets/grpc/llama-cpp-fallback: backend-assets/grpc
 	cp -rf backend/cpp/llama backend/cpp/llama-fallback
 	$(MAKE) -C backend/cpp/llama-fallback purge
 	$(info ${GREEN}I llama-cpp build info:fallback${RESET})
-	CMAKE_ARGS="$(CMAKE_ARGS) -DLLAMA_AVX=off -DLLAMA_AVX2=off -DLLAMA_AVX512=off -DLLAMA_FMA=off -DLLAMA_F16C=off" $(MAKE) VARIANT="llama-fallback" build-llama-cpp-grpc-server
+	CMAKE_ARGS="$(CMAKE_ARGS) -DGGML_AVX=off -DGGML_AVX2=off -DGGML_AVX512=off -DGGML_FMA=off -DGGML_F16C=off" $(MAKE) VARIANT="llama-fallback" build-llama-cpp-grpc-server
 	cp -rfv backend/cpp/llama-fallback/grpc-server backend-assets/grpc/llama-cpp-fallback
 # TODO: every binary should have its own folder instead, so can have different metal implementations
 ifeq ($(BUILD_TYPE),metal)
@@ -736,7 +736,7 @@ backend-assets/grpc/llama-cpp-cuda: backend-assets/grpc
 	cp -rf backend/cpp/llama backend/cpp/llama-cuda
 	$(MAKE) -C backend/cpp/llama-cuda purge
 	$(info ${GREEN}I llama-cpp build info:cuda${RESET})
-	CMAKE_ARGS="$(CMAKE_ARGS) -DLLAMA_AVX=on -DLLAMA_AVX2=off -DLLAMA_AVX512=off -DLLAMA_FMA=off -DLLAMA_F16C=off -DLLAMA_CUDA=ON" $(MAKE) VARIANT="llama-cuda" build-llama-cpp-grpc-server
+	CMAKE_ARGS="$(CMAKE_ARGS) -DGGML_AVX=on -DGGML_AVX2=off -DGGML_AVX512=off -DGGML_FMA=off -DGGML_F16C=off -DGGML_CUDA=ON" $(MAKE) VARIANT="llama-cuda" build-llama-cpp-grpc-server
 	cp -rfv backend/cpp/llama-cuda/grpc-server backend-assets/grpc/llama-cpp-cuda
 
 backend-assets/grpc/llama-cpp-hipblas: backend-assets/grpc
@@ -764,7 +764,7 @@ backend-assets/grpc/llama-cpp-grpc: backend-assets/grpc
 	cp -rf backend/cpp/llama backend/cpp/llama-grpc
 	$(MAKE) -C backend/cpp/llama-grpc purge
 	$(info ${GREEN}I llama-cpp build info:grpc${RESET})
-	CMAKE_ARGS="$(CMAKE_ARGS) -DLLAMA_RPC=ON -DLLAMA_AVX=off -DLLAMA_AVX2=off -DLLAMA_AVX512=off -DLLAMA_FMA=off -DLLAMA_F16C=off" TARGET="--target grpc-server --target rpc-server" $(MAKE) VARIANT="llama-grpc" build-llama-cpp-grpc-server
+	CMAKE_ARGS="$(CMAKE_ARGS) -DGGML_RPC=ON -DGGML_AVX=off -DGGML_AVX2=off -DGGML_AVX512=off -DGGML_FMA=off -DGGML_F16C=off" TARGET="--target grpc-server --target rpc-server" $(MAKE) VARIANT="llama-grpc" build-llama-cpp-grpc-server
 	cp -rfv backend/cpp/llama-grpc/grpc-server backend-assets/grpc/llama-cpp-grpc
 
 backend-assets/util/llama-cpp-rpc-server: backend-assets/grpc/llama-cpp-grpc

backend/cpp/llama/Makefile

@@ -6,35 +6,42 @@ BUILD_TYPE?=
 ONEAPI_VARS?=/opt/intel/oneapi/setvars.sh
 TARGET?=--target grpc-server
 
-# If build type is cublas, then we set -DLLAMA_CUBLAS=ON to CMAKE_ARGS automatically
+# Disable Shared libs as we are linking on static gRPC and we can't mix shared and static
+CMAKE_ARGS+=-DBUILD_SHARED_LIBS=OFF
+
+# If build type is cublas, then we set -DGGML_CUDA=ON to CMAKE_ARGS automatically
 ifeq ($(BUILD_TYPE),cublas)
-	CMAKE_ARGS+=-DLLAMA_CUBLAS=ON
-# If build type is openblas then we set -DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS
+	CMAKE_ARGS+=-DGGML_CUDA=ON
+# If build type is openblas then we set -DGGML_BLAS=ON -DGGML_BLAS_VENDOR=OpenBLAS
 # to CMAKE_ARGS automatically
 else ifeq ($(BUILD_TYPE),openblas)
-	CMAKE_ARGS+=-DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS
-# If build type is clblas (openCL) we set -DLLAMA_CLBLAST=ON -DCLBlast_DIR=/some/path
+	CMAKE_ARGS+=-DGGML_BLAS=ON -DGGML_BLAS_VENDOR=OpenBLAS
+# If build type is clblas (openCL) we set -DGGML_CLBLAST=ON -DCLBlast_DIR=/some/path
 else ifeq ($(BUILD_TYPE),clblas)
-	CMAKE_ARGS+=-DLLAMA_CLBLAST=ON -DCLBlast_DIR=/some/path
+	CMAKE_ARGS+=-DGGML_CLBLAST=ON -DCLBlast_DIR=/some/path
 # If it's hipblas we do have also to set CC=/opt/rocm/llvm/bin/clang CXX=/opt/rocm/llvm/bin/clang++ 
 else ifeq ($(BUILD_TYPE),hipblas)
-	CMAKE_ARGS+=-DLLAMA_HIPBLAS=ON
-# If it's OSX, DO NOT embed the metal library - -DLLAMA_METAL_EMBED_LIBRARY=ON requires further investigation
+	CMAKE_ARGS+=-DGGML_HIPBLAS=ON
+# If it's OSX, DO NOT embed the metal library - -DGGML_METAL_EMBED_LIBRARY=ON requires further investigation
 # But if it's OSX without metal, disable it here
 else ifeq ($(OS),Darwin)
 	ifneq ($(BUILD_TYPE),metal)
-		CMAKE_ARGS+=-DLLAMA_METAL=OFF
+		CMAKE_ARGS+=-DGGML_METAL=OFF
 	else
+		CMAKE_ARGS+=-DGGML_METAL=ON
+# Until this is tested properly, we disable embedded metal file
+# as we already embed it as part of the LocalAI assets
+		CMAKE_ARGS+=-DGGML_METAL_EMBED_LIBRARY=OFF
 		TARGET+=--target ggml-metal
 	endif
 endif
 
 ifeq ($(BUILD_TYPE),sycl_f16)
-	CMAKE_ARGS+=-DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA_SYCL_F16=ON
+	CMAKE_ARGS+=-DGGML_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DGGML_SYCL_F16=ON
 endif
 
 ifeq ($(BUILD_TYPE),sycl_f32)
-	CMAKE_ARGS+=-DLLAMA_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx
+	CMAKE_ARGS+=-DGGML_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx
 endif
 
 llama.cpp:

docs/content/docs/advanced/fine-tuning.md

@@ -118,7 +118,7 @@ And we convert it to the gguf format that LocalAI can consume:
 
 # Convert to gguf
 git clone https://github.com/ggerganov/llama.cpp.git
-pushd llama.cpp && make LLAMA_CUBLAS=1 && popd
+pushd llama.cpp && make GGML_CUDA=1 && popd
 
 # We need to convert the pytorch model into ggml for quantization
 # It crates 'ggml-model-f16.bin' in the 'merged' directory.

docs/content/docs/faq.md

@@ -55,4 +55,4 @@ This typically happens when your prompt exceeds the context size. Try to reduce
 
 ### I'm getting a 'SIGILL' error, what's wrong?
 
-Your CPU probably does not have support for certain instructions that are compiled by default in the pre-built binaries. If you are running in a container, try setting `REBUILD=true` and disable the CPU instructions that are not compatible with your CPU. For instance: `CMAKE_ARGS="-DLLAMA_F16C=OFF -DLLAMA_AVX512=OFF -DLLAMA_AVX2=OFF -DLLAMA_FMA=OFF" make build`
+Your CPU probably does not have support for certain instructions that are compiled by default in the pre-built binaries. If you are running in a container, try setting `REBUILD=true` and disable the CPU instructions that are not compatible with your CPU. For instance: `CMAKE_ARGS="-DGGML_F16C=OFF -DGGML_AVX512=OFF -DGGML_AVX2=OFF -DGGML_FMA=OFF" make build`

docs/content/docs/getting-started/build.md

@@ -101,14 +101,14 @@ Here is the list of the variables available that can be used to customize the bu
 LocalAI uses different backends based on ggml and llama.cpp to run models. If your CPU doesn't support common instruction sets, you can disable them during build:
 
 ```
-CMAKE_ARGS="-DLLAMA_F16C=OFF -DLLAMA_AVX512=OFF -DLLAMA_AVX2=OFF -DLLAMA_AVX=OFF -DLLAMA_FMA=OFF" make build
+CMAKE_ARGS="-DGGML_F16C=OFF -DGGML_AVX512=OFF -DGGML_AVX2=OFF -DGGML_AVX=OFF -DGGML_FMA=OFF" make build
 ```
 
 To have effect on the container image, you need to set `REBUILD=true`:
 
 ```
 docker run  quay.io/go-skynet/localai
-docker run --rm -ti -p 8080:8080 -e DEBUG=true -e MODELS_PATH=/models -e THREADS=1 -e REBUILD=true -e CMAKE_ARGS="-DLLAMA_F16C=OFF -DLLAMA_AVX512=OFF -DLLAMA_AVX2=OFF -DLLAMA_AVX=OFF -DLLAMA_FMA=OFF" -v $PWD/models:/models quay.io/go-skynet/local-ai:latest
+docker run --rm -ti -p 8080:8080 -e DEBUG=true -e MODELS_PATH=/models -e THREADS=1 -e REBUILD=true -e CMAKE_ARGS="-DGGML_F16C=OFF -DGGML_AVX512=OFF -DGGML_AVX2=OFF -DGGML_AVX=OFF -DGGML_FMA=OFF" -v $PWD/models:/models quay.io/go-skynet/local-ai:latest
 ```
 
 {{% /alert %}}

entrypoint.sh

@@ -22,7 +22,7 @@ else
 	echo "@@@@@"
 	echo "If you are experiencing issues with the pre-compiled builds, try setting REBUILD=true"
 	echo "If you are still experiencing issues with the build, try setting CMAKE_ARGS and disable the instructions set as needed:"
-	echo 'CMAKE_ARGS="-DLLAMA_F16C=OFF -DLLAMA_AVX512=OFF -DLLAMA_AVX2=OFF -DLLAMA_FMA=OFF"'
+	echo 'CMAKE_ARGS="-DGGML_F16C=OFF -DGGML_AVX512=OFF -DGGML_AVX2=OFF -DGGML_FMA=OFF"'
 	echo "see the documentation at: https://localai.io/basics/build/index.html"
 	echo "Note: See also https://github.com/go-skynet/LocalAI/issues/288"
 	echo "@@@@@"

examples/e2e-fine-tuning/README.md

@@ -65,7 +65,7 @@ And we convert it to the gguf format that LocalAI can consume:
 
 # Convert to gguf
 git clone https://github.com/ggerganov/llama.cpp.git
-pushd llama.cpp && make LLAMA_CUBLAS=1 && popd
+pushd llama.cpp && make GGML_CUDA=1 && popd
 
 # We need to convert the pytorch model into ggml for quantization
 # It crates 'ggml-model-f16.bin' in the 'merged' directory.

examples/e2e-fine-tuning/notebook.ipynb

@@ -1600,7 +1600,7 @@
       "source": [
         "\n",
         "!git clone https://github.com/ggerganov/llama.cpp.git\n",
-        "!cd llama.cpp && make LLAMA_CUBLAS=1\n",
+        "!cd llama.cpp && make GGML_CUDA=1\n",
         "\n"
       ]
     },