mirror of
https://github.com/LostRuins/koboldcpp.git
synced 2026-09-03 03:21:24 +02:00
Gavin Zhao
82ca83db3c
Supercedes #4024 and #4813. CMake's native HIP support has become the recommended way to add HIP code into a project (see [here](https://rocm.docs.amd.com/en/docs-6.0.0/conceptual/cmake-packages.html#using-hip-in-cmake)). This PR makes the following changes: 1. The environment variable `HIPCXX` or CMake option `CMAKE_HIP_COMPILER` should be used to specify the HIP compiler. Notably this shouldn't be `hipcc`, but ROCm's clang, which usually resides in `$ROCM_PATH/llvm/bin/clang`. Previously this was control by `CMAKE_C_COMPILER` and `CMAKE_CXX_COMPILER`. Note that since native CMake HIP support is not yet available on Windows, on Windows we fall back to the old behavior. 2. CMake option `CMAKE_HIP_ARCHITECTURES` is used to control the GPU architectures to build for. Previously this was controled by `GPU_TARGETS`. 3. Updated the Nix recipe to account for these new changes. 4. The GPU targets to build against in the Nix recipe is now consistent with the supported GPU targets in nixpkgs. 5. Added CI checks for HIP on both Linux and Windows. On Linux, we test both the new and old behavior. The most important part about this PR is the separation of the HIP compiler and the C/C++ compiler. This allows users to choose a different C/C++ compiler if desired, compared to the current situation where when building for ROCm support, everything must be compiled with ROCm's clang. ~~Makefile is unchanged. Please let me know if we want to be consistent on variables' naming because Makefile still uses `GPU_TARGETS` to control architectures to build for, but I feel like setting `CMAKE_HIP_ARCHITECTURES` is a bit awkward when you're calling `make`.~~ Makefile used `GPU_TARGETS` but the README says to use `AMDGPU_TARGETS`. For consistency with CMake, all usage of `GPU_TARGETS` in Makefile has been updated to `AMDGPU_TARGETS`. Thanks to the suggestion of @jin-eld, to maintain backwards compatibility (and not break too many downstream users' builds), if `CMAKE_CXX_COMPILER` ends with `hipcc`, then we still compile using the original behavior and emit a warning that recommends switching to the new HIP support. Similarly, if `AMDGPU_TARGETS` is set but `CMAKE_HIP_ARCHITECTURES` is not, then we forward `AMDGPU_TARGETS` to `CMAKE_HIP_ARCHITECTURES` to ease the transition to the new HIP support. Signed-off-by: Gavin Zhao <git@gzgz.dev>
646 lines
27 KiB
Makefile
646 lines
27 KiB
Makefile
default: koboldcpp_default koboldcpp_failsafe koboldcpp_openblas koboldcpp_noavx2 koboldcpp_clblast koboldcpp_clblast_noavx2 koboldcpp_cublas koboldcpp_hipblas koboldcpp_vulkan koboldcpp_vulkan_noavx2
|
|
tools: quantize_gpt2 quantize_gptj quantize_gguf quantize_neox quantize_mpt quantize_clip sdmain gguf-split
|
|
dev: koboldcpp_openblas
|
|
dev2: koboldcpp_clblast
|
|
|
|
|
|
ifndef UNAME_S
|
|
UNAME_S := $(shell uname -s)
|
|
endif
|
|
|
|
ifndef UNAME_P
|
|
UNAME_P := $(shell uname -p)
|
|
endif
|
|
|
|
ifndef UNAME_M
|
|
UNAME_M := $(shell uname -m)
|
|
endif
|
|
|
|
ifneq ($(shell grep -e "Arch Linux" -e "ID_LIKE=arch" /etc/os-release 2>/dev/null),)
|
|
ARCH_ADD = -lcblas
|
|
endif
|
|
|
|
|
|
# Mac OS + Arm can report x86_64
|
|
# ref: https://github.com/ggerganov/whisper.cpp/issues/66#issuecomment-1282546789
|
|
ifeq ($(UNAME_S),Darwin)
|
|
ifneq ($(UNAME_P),arm)
|
|
SYSCTL_M := $(shell sysctl -n hw.optional.arm64 2>/dev/null)
|
|
ifeq ($(SYSCTL_M),1)
|
|
# UNAME_P := arm
|
|
# UNAME_M := arm64
|
|
warn := $(warning Your arch is announced as x86_64, but it seems to actually be ARM64. Not fixing that can lead to bad performance. For more info see: https://github.com/ggerganov/whisper.cpp/issues/66\#issuecomment-1282546789)
|
|
endif
|
|
endif
|
|
endif
|
|
|
|
#
|
|
# Compile flags
|
|
#
|
|
|
|
# keep standard at C11 and C++11
|
|
CFLAGS = -I. -I./include -I./include/CL -I./otherarch -I./otherarch/tools -I./otherarch/sdcpp -I./otherarch/sdcpp/thirdparty -I./include/vulkan -O3 -DNDEBUG -std=c11 -fPIC -DLOG_DISABLE_LOGS -D_GNU_SOURCE -DGGML_USE_LLAMAFILE
|
|
CXXFLAGS = -I. -I./common -I./include -I./include/CL -I./otherarch -I./otherarch/tools -I./otherarch/sdcpp -I./otherarch/sdcpp/thirdparty -I./include/vulkan -O3 -DNDEBUG -std=c++11 -fPIC -DLOG_DISABLE_LOGS -D_GNU_SOURCE -DGGML_USE_LLAMAFILE
|
|
LDFLAGS =
|
|
FASTCFLAGS = $(subst -O3,-Ofast,$(CFLAGS))
|
|
FASTCXXFLAGS = $(subst -O3,-Ofast,$(CXXFLAGS))
|
|
|
|
# these are used on windows, to build some libraries with extra old device compatibility
|
|
SIMPLECFLAGS =
|
|
FULLCFLAGS =
|
|
NONECFLAGS =
|
|
|
|
OPENBLAS_FLAGS = -DGGML_USE_OPENBLAS -I/usr/local/include/openblas
|
|
CLBLAST_FLAGS = -DGGML_USE_CLBLAST
|
|
FAILSAFE_FLAGS = -DUSE_FAILSAFE
|
|
VULKAN_FLAGS = -DGGML_USE_VULKAN
|
|
ifdef LLAMA_CUBLAS
|
|
CUBLAS_FLAGS = -DGGML_USE_CUDA -DSD_USE_CUBLAS
|
|
else
|
|
CUBLAS_FLAGS =
|
|
endif
|
|
CUBLASLD_FLAGS =
|
|
CUBLAS_OBJS =
|
|
|
|
OBJS_FULL += ggml-alloc.o ggml-quants.o unicode.o unicode-data.o sgemm.o common.o grammar-parser.o
|
|
OBJS_SIMPLE += ggml-alloc.o ggml-quants_noavx2.o unicode.o unicode-data.o sgemm_noavx2.o common.o grammar-parser.o
|
|
OBJS_FAILSAFE += ggml-alloc.o ggml-quants_failsafe.o unicode.o unicode-data.o sgemm_failsafe.o common.o grammar-parser.o
|
|
|
|
#lets try enabling everything
|
|
CFLAGS += -pthread -s -Wno-deprecated -Wno-deprecated-declarations
|
|
CXXFLAGS += -pthread -s -Wno-multichar -Wno-write-strings -Wno-deprecated -Wno-deprecated-declarations
|
|
|
|
# OS specific
|
|
# TODO: support Windows
|
|
ifeq ($(UNAME_S),Linux)
|
|
CFLAGS += -pthread
|
|
CXXFLAGS += -pthread
|
|
endif
|
|
|
|
ifeq ($(UNAME_S),Darwin)
|
|
CFLAGS += -pthread
|
|
CXXFLAGS += -pthread
|
|
CLANG_VER = $(shell clang -v 2>&1 | head -n 1 | awk 'BEGIN {FS="[. ]"};{print $$1 $$2 $$4}')
|
|
ifeq ($(CLANG_VER),Appleclang15)
|
|
LDFLAGS += -ld_classic
|
|
endif
|
|
endif
|
|
ifeq ($(UNAME_S),FreeBSD)
|
|
CFLAGS += -pthread
|
|
CXXFLAGS += -pthread
|
|
endif
|
|
ifeq ($(UNAME_S),NetBSD)
|
|
CFLAGS += -pthread
|
|
CXXFLAGS += -pthread
|
|
endif
|
|
ifeq ($(UNAME_S),OpenBSD)
|
|
CFLAGS += -pthread
|
|
CXXFLAGS += -pthread
|
|
endif
|
|
ifeq ($(UNAME_S),Haiku)
|
|
CFLAGS += -pthread
|
|
CXXFLAGS += -pthread
|
|
endif
|
|
|
|
ifdef LLAMA_GPROF
|
|
CFLAGS += -pg
|
|
CXXFLAGS += -pg
|
|
endif
|
|
ifdef LLAMA_PERF
|
|
CFLAGS += -DGGML_PERF
|
|
CXXFLAGS += -DGGML_PERF
|
|
endif
|
|
|
|
# Architecture specific
|
|
# TODO: probably these flags need to be tweaked on some architectures
|
|
# feel free to update the Makefile for your architecture and send a pull request or issue
|
|
ifeq ($(UNAME_M),$(filter $(UNAME_M),x86_64 i686))
|
|
# Use all CPU extensions that are available:
|
|
# old library NEEDS mf16c to work. so we must build with it. new one doesnt
|
|
ifeq ($(OS),Windows_NT)
|
|
CFLAGS +=
|
|
NONECFLAGS +=
|
|
SIMPLECFLAGS += -mavx -msse3
|
|
FULLCFLAGS += -mavx2 -msse3 -mfma -mf16c -mavx
|
|
else
|
|
# if not on windows, they are clearly building it themselves, so lets just use whatever is supported
|
|
ifdef LLAMA_PORTABLE
|
|
CFLAGS +=
|
|
NONECFLAGS +=
|
|
SIMPLECFLAGS += -mavx -msse3
|
|
FULLCFLAGS += -mavx2 -msse3 -mfma -mf16c -mavx
|
|
else
|
|
CFLAGS += -march=native -mtune=native
|
|
endif
|
|
endif
|
|
endif
|
|
|
|
ifndef LLAMA_NO_ACCELERATE
|
|
# Mac M1 - include Accelerate framework.
|
|
# `-framework Accelerate` works on Mac Intel as well, with negliable performance boost (as of the predict time).
|
|
ifeq ($(UNAME_S),Darwin)
|
|
CFLAGS += -DGGML_USE_ACCELERATE
|
|
LDFLAGS += -framework Accelerate
|
|
endif
|
|
endif
|
|
|
|
# it is recommended to use the CMAKE file to build for cublas if you can - will likely work better
|
|
ifdef LLAMA_CUBLAS
|
|
CUBLAS_FLAGS = -DGGML_USE_CUDA -DSD_USE_CUBLAS -I/usr/local/cuda/include -I/opt/cuda/include -I$(CUDA_PATH)/targets/x86_64-linux/include
|
|
CUBLASLD_FLAGS = -lcuda -lcublas -lcudart -lcublasLt -lpthread -ldl -lrt -L/usr/local/cuda/lib64 -L/opt/cuda/lib64 -L$(CUDA_PATH)/targets/x86_64-linux/lib -L/usr/local/cuda/targets/aarch64-linux/lib -L/usr/local/cuda/targets/sbsa-linux/lib -L/usr/lib/wsl/lib
|
|
CUBLAS_OBJS = ggml-cuda.o ggml_v3-cuda.o ggml_v2-cuda.o ggml_v2-cuda-legacy.o
|
|
CUBLAS_OBJS += $(patsubst %.cu,%.o,$(wildcard ggml-cuda/*.cu))
|
|
NVCC = nvcc
|
|
NVCCFLAGS = --forward-unknown-to-host-compiler -use_fast_math
|
|
|
|
ifdef LLAMA_ADD_CONDA_PATHS
|
|
CUBLASLD_FLAGS += -Lconda/envs/linux/lib -Lconda/envs/linux/lib/stubs
|
|
endif
|
|
|
|
ifdef CUDA_DOCKER_ARCH
|
|
NVCCFLAGS += -Wno-deprecated-gpu-targets -arch=$(CUDA_DOCKER_ARCH)
|
|
else
|
|
ifdef LLAMA_PORTABLE
|
|
ifdef LLAMA_COLAB #colab does not need all targets, all-major doesnt work correctly with pascal
|
|
NVCCFLAGS += -Wno-deprecated-gpu-targets -arch=all-major
|
|
else
|
|
NVCCFLAGS += -Wno-deprecated-gpu-targets -arch=all
|
|
endif #LLAMA_COLAB
|
|
else
|
|
NVCCFLAGS += -arch=native
|
|
endif #LLAMA_PORTABLE
|
|
endif # CUDA_DOCKER_ARCH
|
|
|
|
ifdef LLAMA_CUDA_FORCE_DMMV
|
|
NVCCFLAGS += -DGGML_CUDA_FORCE_DMMV
|
|
endif # LLAMA_CUDA_FORCE_DMMV
|
|
ifdef LLAMA_CUDA_DMMV_X
|
|
NVCCFLAGS += -DGGML_CUDA_DMMV_X=$(LLAMA_CUDA_DMMV_X)
|
|
else
|
|
NVCCFLAGS += -DGGML_CUDA_DMMV_X=32
|
|
endif # LLAMA_CUDA_DMMV_X
|
|
ifdef LLAMA_CUDA_MMV_Y
|
|
NVCCFLAGS += -DGGML_CUDA_MMV_Y=$(LLAMA_CUDA_MMV_Y)
|
|
else ifdef LLAMA_CUDA_DMMV_Y
|
|
NVCCFLAGS += -DGGML_CUDA_MMV_Y=$(LLAMA_CUDA_DMMV_Y) # for backwards compatibility
|
|
else
|
|
NVCCFLAGS += -DGGML_CUDA_MMV_Y=1
|
|
endif # LLAMA_CUDA_MMV_Y
|
|
ifdef LLAMA_CUDA_F16
|
|
NVCCFLAGS += -DGGML_CUDA_F16
|
|
endif # LLAMA_CUDA_F16
|
|
ifdef LLAMA_CUDA_DMMV_F16
|
|
NVCCFLAGS += -DGGML_CUDA_F16
|
|
endif # LLAMA_CUDA_DMMV_F16
|
|
ifdef LLAMA_CUDA_KQUANTS_ITER
|
|
NVCCFLAGS += -DK_QUANTS_PER_ITERATION=$(LLAMA_CUDA_KQUANTS_ITER)
|
|
else
|
|
NVCCFLAGS += -DK_QUANTS_PER_ITERATION=2
|
|
endif
|
|
ifdef LLAMA_CUDA_MMQ_Y
|
|
NVCCFLAGS += -DGGML_CUDA_MMQ_Y=$(LLAMA_CUDA_MMQ_Y)
|
|
else
|
|
NVCCFLAGS += -DGGML_CUDA_MMQ_Y=64
|
|
endif
|
|
ifdef LLAMA_CUDA_CCBIN
|
|
NVCCFLAGS += -ccbin $(LLAMA_CUDA_CCBIN)
|
|
endif
|
|
|
|
ggml-cuda/%.o: ggml-cuda/%.cu ggml-cuda/%.cuh ggml.h ggml-common.h ggml-cuda/common.cuh
|
|
$(NVCC) $(NVCCFLAGS) $(subst -Ofast,-O3,$(CXXFLAGS)) $(CUBLAS_FLAGS) $(CUBLAS_CXXFLAGS) -Wno-pedantic -c $< -o $@
|
|
ggml-cuda.o: ggml-cuda.cu ggml-cuda.h ggml.h ggml-backend.h ggml-backend-impl.h ggml-common.h $(wildcard ggml-cuda/*.cuh)
|
|
$(NVCC) $(NVCCFLAGS) $(subst -Ofast,-O3,$(CXXFLAGS)) $(CUBLAS_FLAGS) $(CUBLAS_CXXFLAGS) -Wno-pedantic -c $< -o $@
|
|
ggml_v2-cuda.o: otherarch/ggml_v2-cuda.cu otherarch/ggml_v2-cuda.h
|
|
$(NVCC) $(NVCCFLAGS) $(subst -Ofast,-O3,$(CXXFLAGS)) $(CUBLAS_FLAGS) $(CUBLAS_CXXFLAGS) -Wno-pedantic -c $< -o $@
|
|
ggml_v2-cuda-legacy.o: otherarch/ggml_v2-cuda-legacy.cu otherarch/ggml_v2-cuda-legacy.h
|
|
$(NVCC) $(NVCCFLAGS) $(subst -Ofast,-O3,$(CXXFLAGS)) $(CUBLAS_FLAGS) $(CUBLAS_CXXFLAGS) -Wno-pedantic -c $< -o $@
|
|
ggml_v3-cuda.o: otherarch/ggml_v3-cuda.cu otherarch/ggml_v3-cuda.h
|
|
$(NVCC) $(NVCCFLAGS) $(subst -Ofast,-O3,$(CXXFLAGS)) $(CUBLAS_FLAGS) $(CUBLAS_CXXFLAGS) -Wno-pedantic -c $< -o $@
|
|
endif # LLAMA_CUBLAS
|
|
|
|
ifdef LLAMA_HIPBLAS
|
|
ifeq ($(wildcard /opt/rocm),)
|
|
ROCM_PATH ?= /usr
|
|
GPU_TARGETS ?= $(shell $(shell which amdgpu-arch))
|
|
HCC := $(ROCM_PATH)/bin/hipcc
|
|
HCXX := $(ROCM_PATH)/bin/hipcc
|
|
else
|
|
ROCM_PATH ?= /opt/rocm
|
|
GPU_TARGETS ?= gfx803 gfx900 gfx906 gfx908 gfx90a gfx1030 gfx1100 $(shell $(ROCM_PATH)/llvm/bin/amdgpu-arch)
|
|
HCC := $(ROCM_PATH)/llvm/bin/clang
|
|
HCXX := $(ROCM_PATH)/llvm/bin/clang++
|
|
endif
|
|
LLAMA_CUDA_DMMV_X ?= 32
|
|
LLAMA_CUDA_MMV_Y ?= 1
|
|
LLAMA_CUDA_KQUANTS_ITER ?= 2
|
|
HIPFLAGS += -DGGML_USE_HIPBLAS -DGGML_USE_CUDA -DSD_USE_CUBLAS $(shell $(ROCM_PATH)/bin/hipconfig -C)
|
|
HIPLDFLAGS += -L$(ROCM_PATH)/lib -Wl,-rpath=$(ROCM_PATH)/lib -lhipblas -lamdhip64 -lrocblas
|
|
HIP_OBJS += ggml-cuda.o ggml_v3-cuda.o ggml_v2-cuda.o ggml_v2-cuda-legacy.o
|
|
HIP_OBJS += $(patsubst %.cu,%.o,$(wildcard ggml-cuda/*.cu))
|
|
|
|
HIPFLAGS2 += $(addprefix --offload-arch=,$(GPU_TARGETS))
|
|
HIPFLAGS2 += -DGGML_CUDA_DMMV_X=$(LLAMA_CUDA_DMMV_X)
|
|
HIPFLAGS2 += -DGGML_CUDA_MMV_Y=$(LLAMA_CUDA_MMV_Y)
|
|
HIPFLAGS2 += -DK_QUANTS_PER_ITERATION=$(LLAMA_CUDA_KQUANTS_ITER)
|
|
|
|
ggml-cuda/%.o: ggml-cuda/%.cu ggml-cuda/%.cuh ggml.h ggml-common.h ggml-cuda/common.cuh
|
|
$(HCXX) $(CXXFLAGS) $(HIPFLAGS) $(HIPFLAGS2) -x hip -c -o $@ $<
|
|
ggml-cuda.o: ggml-cuda.cu ggml-cuda.h ggml.h ggml-backend.h ggml-backend-impl.h ggml-common.h $(wildcard ggml-cuda/*.cuh)
|
|
$(HCXX) $(CXXFLAGS) $(HIPFLAGS) $(HIPFLAGS2) -x hip -c -o $@ $<
|
|
ggml_v2-cuda.o: otherarch/ggml_v2-cuda.cu otherarch/ggml_v2-cuda.h
|
|
$(HCXX) $(CXXFLAGS) $(HIPFLAGS) $(HIPFLAGS2) -x hip -c -o $@ $<
|
|
ggml_v2-cuda-legacy.o: otherarch/ggml_v2-cuda-legacy.cu otherarch/ggml_v2-cuda-legacy.h
|
|
$(HCXX) $(CXXFLAGS) $(HIPFLAGS) $(HIPFLAGS2) -x hip -c -o $@ $<
|
|
ggml_v3-cuda.o: otherarch/ggml_v3-cuda.cu otherarch/ggml_v3-cuda.h
|
|
$(HCXX) $(CXXFLAGS) $(HIPFLAGS) $(HIPFLAGS2) -x hip -c -o $@ $<
|
|
endif # LLAMA_HIPBLAS
|
|
|
|
|
|
ifdef LLAMA_METAL
|
|
CFLAGS += -DGGML_USE_METAL -DGGML_METAL_NDEBUG -DSD_USE_METAL
|
|
CXXFLAGS += -DGGML_USE_METAL -DSD_USE_METAL
|
|
LDFLAGS += -framework Foundation -framework Metal -framework MetalKit -framework MetalPerformanceShaders
|
|
OBJS += ggml-metal.o
|
|
|
|
ggml-metal.o: ggml-metal.m ggml-metal.h
|
|
@echo "== Preparing merged Metal file =="
|
|
@sed -e '/#include "ggml-common.h"/r ggml-common.h' -e '/#include "ggml-common.h"/d' < ggml-metal.metal > ggml-metal-merged.metal
|
|
$(CC) $(CFLAGS) -c $< -o $@
|
|
endif # LLAMA_METAL
|
|
|
|
ifneq ($(filter aarch64%,$(UNAME_M)),)
|
|
# Apple M1, M2, etc.
|
|
# Raspberry Pi 3, 4, Zero 2 (64-bit)
|
|
CFLAGS +=
|
|
CXXFLAGS +=
|
|
endif
|
|
ifneq ($(filter armv6%,$(UNAME_M)),)
|
|
# Raspberry Pi 1, Zero
|
|
CFLAGS += -mfpu=neon-fp-armv8 -mfp16-format=ieee -mno-unaligned-access
|
|
CXXFLAGS += -mfpu=neon-fp-armv8 -mfp16-format=ieee -mno-unaligned-access
|
|
endif
|
|
ifneq ($(filter armv7%,$(UNAME_M)),)
|
|
# Raspberry Pi 2
|
|
CFLAGS += -mfpu=neon-fp-armv8 -mfp16-format=ieee -mno-unaligned-access -funsafe-math-optimizations
|
|
CXXFLAGS += -mfpu=neon-fp-armv8 -mfp16-format=ieee -mno-unaligned-access -funsafe-math-optimizations
|
|
endif
|
|
ifneq ($(filter armv8%,$(UNAME_M)),)
|
|
# Raspberry Pi 3, 4, Zero 2 (32-bit)
|
|
CFLAGS += -mfp16-format=ieee -mno-unaligned-access
|
|
CXXFLAGS += -mfp16-format=ieee -mno-unaligned-access
|
|
endif
|
|
ifneq ($(filter ppc64%,$(UNAME_M)),)
|
|
POWER9_M := $(shell grep "POWER9" /proc/cpuinfo)
|
|
ifneq (,$(findstring POWER9,$(POWER9_M)))
|
|
CFLAGS += -mcpu=power9
|
|
CXXFLAGS += -mcpu=power9
|
|
endif
|
|
endif
|
|
|
|
|
|
DEFAULT_BUILD =
|
|
FAILSAFE_BUILD =
|
|
OPENBLAS_BUILD =
|
|
NOAVX2_BUILD =
|
|
CLBLAST_BUILD =
|
|
CUBLAS_BUILD =
|
|
HIPBLAS_BUILD =
|
|
VULKAN_BUILD =
|
|
|
|
ifeq ($(OS),Windows_NT)
|
|
DEFAULT_BUILD = $(CXX) $(CXXFLAGS) $^ -shared -o $@.dll $(LDFLAGS)
|
|
FAILSAFE_BUILD = $(CXX) $(CXXFLAGS) $^ -shared -o $@.dll $(LDFLAGS)
|
|
OPENBLAS_BUILD = $(CXX) $(CXXFLAGS) $^ lib/libopenblas.lib -shared -o $@.dll $(LDFLAGS)
|
|
NOAVX2_BUILD = $(CXX) $(CXXFLAGS) $^ -shared -o $@.dll $(LDFLAGS)
|
|
CLBLAST_BUILD = $(CXX) $(CXXFLAGS) $^ lib/OpenCL.lib lib/clblast.lib -shared -o $@.dll $(LDFLAGS)
|
|
VULKAN_BUILD = $(CXX) $(CXXFLAGS) $^ lib/vulkan-1.lib -shared -o $@.dll $(LDFLAGS)
|
|
|
|
ifdef LLAMA_CUBLAS
|
|
CUBLAS_BUILD = $(CXX) $(CXXFLAGS) $(CUBLAS_FLAGS) $^ -shared -o $@.dll $(CUBLASLD_FLAGS) $(LDFLAGS)
|
|
endif
|
|
ifdef LLAMA_HIPBLAS
|
|
HIPBLAS_BUILD = $(HCXX) $(CXXFLAGS) $(HIPFLAGS) $^ -shared -o $@.dll $(HIPLDFLAGS) $(LDFLAGS)
|
|
endif
|
|
else
|
|
DEFAULT_BUILD = $(CXX) $(CXXFLAGS) $^ -shared -o $@.so $(LDFLAGS)
|
|
ifdef LLAMA_PORTABLE
|
|
FAILSAFE_BUILD = $(CXX) $(CXXFLAGS) $^ -shared -o $@.so $(LDFLAGS)
|
|
NOAVX2_BUILD = $(CXX) $(CXXFLAGS) $^ -shared -o $@.so $(LDFLAGS)
|
|
endif
|
|
|
|
ifdef LLAMA_OPENBLAS
|
|
OPENBLAS_BUILD = $(CXX) $(CXXFLAGS) $^ $(ARCH_ADD) -lopenblas -shared -o $@.so $(LDFLAGS)
|
|
endif
|
|
ifdef LLAMA_CLBLAST
|
|
ifeq ($(UNAME_S),Darwin)
|
|
CLBLAST_BUILD = $(CXX) $(CXXFLAGS) $^ -lclblast -framework OpenCL $(ARCH_ADD) -lopenblas -shared -o $@.so $(LDFLAGS)
|
|
else
|
|
CLBLAST_BUILD = $(CXX) $(CXXFLAGS) $^ -lclblast -lOpenCL $(ARCH_ADD) -lopenblas -shared -o $@.so $(LDFLAGS)
|
|
endif
|
|
endif
|
|
ifdef LLAMA_CUBLAS
|
|
CUBLAS_BUILD = $(CXX) $(CXXFLAGS) $(CUBLAS_FLAGS) $^ -shared -o $@.so $(CUBLASLD_FLAGS) $(LDFLAGS)
|
|
endif
|
|
ifdef LLAMA_HIPBLAS
|
|
HIPBLAS_BUILD = $(HCXX) $(CXXFLAGS) $(HIPFLAGS) $^ -shared -o $@.so $(HIPLDFLAGS) $(LDFLAGS)
|
|
endif
|
|
ifdef LLAMA_VULKAN
|
|
VULKAN_BUILD = $(CXX) $(CXXFLAGS) $^ -lvulkan -shared -o $@.so $(LDFLAGS)
|
|
endif
|
|
|
|
ifndef LLAMA_OPENBLAS
|
|
ifndef LLAMA_CLBLAST
|
|
ifndef LLAMA_CUBLAS
|
|
ifndef LLAMA_HIPBLAS
|
|
ifndef LLAMA_VULKAN
|
|
OPENBLAS_BUILD = @echo 'Your OS $(OS) does not appear to be Windows. For faster speeds, install and link a BLAS library. Set LLAMA_OPENBLAS=1 to compile with OpenBLAS support or LLAMA_CLBLAST=1 to compile with ClBlast support. This is just a reminder, not an error.'
|
|
endif
|
|
endif
|
|
endif
|
|
endif
|
|
endif
|
|
endif
|
|
|
|
CCV := $(shell $(CC) --version | head -n 1)
|
|
CXXV := $(shell $(CXX) --version | head -n 1)
|
|
|
|
#
|
|
# Print build information
|
|
#
|
|
|
|
$(info I llama.cpp build info: )
|
|
$(info I UNAME_S: $(UNAME_S))
|
|
$(info I UNAME_P: $(UNAME_P))
|
|
$(info I UNAME_M: $(UNAME_M))
|
|
$(info I CFLAGS: $(CFLAGS))
|
|
$(info I CXXFLAGS: $(CXXFLAGS))
|
|
$(info I LDFLAGS: $(LDFLAGS))
|
|
$(info I CC: $(CCV))
|
|
$(info I CXX: $(CXXV))
|
|
$(info )
|
|
|
|
#
|
|
# Build library
|
|
#
|
|
|
|
ggml.o: ggml.c ggml.h ggml-cuda.h ggml-common.h
|
|
$(CC) $(FASTCFLAGS) $(FULLCFLAGS) -c $< -o $@
|
|
ggml_v4_openblas.o: ggml.c ggml.h ggml-cuda.h ggml-common.h
|
|
$(CC) $(FASTCFLAGS) $(FULLCFLAGS) $(OPENBLAS_FLAGS) -c $< -o $@
|
|
ggml_v4_failsafe.o: ggml.c ggml.h ggml-cuda.h ggml-common.h
|
|
$(CC) $(FASTCFLAGS) $(NONECFLAGS) -c $< -o $@
|
|
ggml_v4_noavx2.o: ggml.c ggml.h ggml-cuda.h ggml-common.h
|
|
$(CC) $(FASTCFLAGS) $(SIMPLECFLAGS) -c $< -o $@
|
|
ggml_v4_clblast.o: ggml.c ggml.h ggml-cuda.h ggml-common.h
|
|
$(CC) $(FASTCFLAGS) $(FULLCFLAGS) $(CLBLAST_FLAGS) -c $< -o $@
|
|
ggml_v4_cublas.o: ggml.c ggml.h ggml-cuda.h ggml-common.h
|
|
$(CC) $(FASTCFLAGS) $(FULLCFLAGS) $(CUBLAS_FLAGS) $(HIPFLAGS) -c $< -o $@
|
|
ggml_v4_clblast_noavx2.o: ggml.c ggml.h ggml-cuda.h ggml-common.h
|
|
$(CC) $(FASTCFLAGS) $(SIMPLECFLAGS) $(CLBLAST_FLAGS) -c $< -o $@
|
|
ggml_v4_vulkan.o: ggml.c ggml.h ggml-cuda.h ggml-common.h
|
|
$(CC) $(FASTCFLAGS) $(FULLCFLAGS) $(VULKAN_FLAGS) -c $< -o $@
|
|
ggml_v4_vulkan_noavx2.o: ggml.c ggml.h ggml-cuda.h ggml-common.h
|
|
$(CC) $(FASTCFLAGS) $(SIMPLECFLAGS) $(VULKAN_FLAGS) -c $< -o $@
|
|
|
|
#quants
|
|
ggml-quants.o: ggml-quants.c ggml.h ggml-quants.h ggml-cuda.h ggml-common.h
|
|
$(CC) $(CFLAGS) $(FULLCFLAGS) -c $< -o $@
|
|
ggml-quants_noavx2.o: ggml-quants.c ggml.h ggml-quants.h ggml-cuda.h ggml-common.h
|
|
$(CC) $(CFLAGS) $(SIMPLECFLAGS) -c $< -o $@
|
|
ggml-quants_failsafe.o: ggml-quants.c ggml.h ggml-quants.h ggml-cuda.h ggml-common.h
|
|
$(CC) $(CFLAGS) $(NONECFLAGS) -c $< -o $@
|
|
|
|
#sgemm
|
|
sgemm.o: sgemm.cpp sgemm.h ggml.h
|
|
$(CXX) $(CXXFLAGS) $(FULLCFLAGS) -c $< -o $@
|
|
sgemm_noavx2.o: sgemm.cpp sgemm.h ggml.h
|
|
$(CXX) $(CXXFLAGS) $(SIMPLECFLAGS) -c $< -o $@
|
|
sgemm_failsafe.o: sgemm.cpp sgemm.h ggml.h
|
|
$(CXX) $(CXXFLAGS) $(NONECFLAGS) -c $< -o $@
|
|
|
|
#there's no intrinsics or special gpu ops used here, so we can have a universal object
|
|
ggml-alloc.o: ggml-alloc.c ggml.h ggml-alloc.h
|
|
$(CC) $(CFLAGS) -c $< -o $@
|
|
llava.o: examples/llava/llava.cpp examples/llava/llava.h
|
|
$(CXX) $(CXXFLAGS) -c $< -o $@
|
|
unicode.o: unicode.cpp unicode.h
|
|
$(CXX) $(CXXFLAGS) -c $< -o $@
|
|
unicode-data.o: unicode-data.cpp unicode-data.h
|
|
$(CXX) $(CXXFLAGS) -c $< -o $@
|
|
|
|
#these have special gpu defines
|
|
ggml-backend_default.o: ggml-backend.c ggml.h ggml-backend.h
|
|
$(CC) $(CFLAGS) -c $< -o $@
|
|
ggml-backend_vulkan.o: ggml-backend.c ggml.h ggml-backend.h
|
|
$(CC) $(CFLAGS) $(VULKAN_FLAGS) -c $< -o $@
|
|
ggml-backend_cublas.o: ggml-backend.c ggml.h ggml-backend.h
|
|
$(CC) $(CFLAGS) $(CUBLAS_FLAGS) -c $< -o $@
|
|
llavaclip_default.o: examples/llava/clip.cpp examples/llava/clip.h
|
|
$(CXX) $(CXXFLAGS) -c $< -o $@
|
|
llavaclip_cublas.o: examples/llava/clip.cpp examples/llava/clip.h
|
|
$(CXX) $(CXXFLAGS) $(CUBLAS_FLAGS) -c $< -o $@
|
|
|
|
#version 3 libs
|
|
ggml_v3.o: otherarch/ggml_v3.c otherarch/ggml_v3.h
|
|
$(CC) $(FASTCFLAGS) $(FULLCFLAGS) -c $< -o $@
|
|
ggml_v3_openblas.o: otherarch/ggml_v3.c otherarch/ggml_v3.h
|
|
$(CC) $(FASTCFLAGS) $(FULLCFLAGS) $(OPENBLAS_FLAGS) -c $< -o $@
|
|
ggml_v3_failsafe.o: otherarch/ggml_v3.c otherarch/ggml_v3.h
|
|
$(CC) $(FASTCFLAGS) $(NONECFLAGS) -c $< -o $@
|
|
ggml_v3_noavx2.o: otherarch/ggml_v3.c otherarch/ggml_v3.h
|
|
$(CC) $(FASTCFLAGS) $(SIMPLECFLAGS) -c $< -o $@
|
|
ggml_v3_clblast.o: otherarch/ggml_v3.c otherarch/ggml_v3.h
|
|
$(CC) $(FASTCFLAGS) $(FULLCFLAGS) $(CLBLAST_FLAGS) -c $< -o $@
|
|
ggml_v3_cublas.o: otherarch/ggml_v3.c otherarch/ggml_v3.h
|
|
$(CC) $(FASTCFLAGS) $(FULLCFLAGS) $(CUBLAS_FLAGS) $(HIPFLAGS) -c $< -o $@
|
|
ggml_v3_clblast_noavx2.o: otherarch/ggml_v3.c otherarch/ggml_v3.h
|
|
$(CC) $(FASTCFLAGS) $(SIMPLECFLAGS) $(CLBLAST_FLAGS) -c $< -o $@
|
|
|
|
#version 2 libs
|
|
ggml_v2.o: otherarch/ggml_v2.c otherarch/ggml_v2.h
|
|
$(CC) $(FASTCFLAGS) $(FULLCFLAGS) -c $< -o $@
|
|
ggml_v2_openblas.o: otherarch/ggml_v2.c otherarch/ggml_v2.h
|
|
$(CC) $(FASTCFLAGS) $(FULLCFLAGS) $(OPENBLAS_FLAGS) -c $< -o $@
|
|
ggml_v2_failsafe.o: otherarch/ggml_v2.c otherarch/ggml_v2.h
|
|
$(CC) $(FASTCFLAGS) $(NONECFLAGS) -c $< -o $@
|
|
ggml_v2_noavx2.o: otherarch/ggml_v2.c otherarch/ggml_v2.h
|
|
$(CC) $(FASTCFLAGS) $(SIMPLECFLAGS) -c $< -o $@
|
|
ggml_v2_clblast.o: otherarch/ggml_v2.c otherarch/ggml_v2.h
|
|
$(CC) $(FASTCFLAGS) $(FULLCFLAGS) $(CLBLAST_FLAGS) -c $< -o $@
|
|
ggml_v2_cublas.o: otherarch/ggml_v2.c otherarch/ggml_v2.h
|
|
$(CC) $(FASTCFLAGS) $(FULLCFLAGS) $(CUBLAS_FLAGS) $(HIPFLAGS) -c $< -o $@
|
|
ggml_v2_clblast_noavx2.o: otherarch/ggml_v2.c otherarch/ggml_v2.h
|
|
$(CC) $(FASTCFLAGS) $(SIMPLECFLAGS) $(CLBLAST_FLAGS) -c $< -o $@
|
|
|
|
#extreme old version compat
|
|
ggml_v1.o: otherarch/ggml_v1.c otherarch/ggml_v1.h
|
|
$(CC) $(FASTCFLAGS) $(FULLCFLAGS) -c $< -o $@
|
|
ggml_v1_failsafe.o: otherarch/ggml_v1.c otherarch/ggml_v1.h
|
|
$(CC) $(FASTCFLAGS) $(NONECFLAGS) -c $< -o $@
|
|
|
|
#opencl
|
|
ggml-opencl.o: ggml-opencl.cpp ggml-opencl.h
|
|
$(CXX) $(CXXFLAGS) $(CLBLAST_FLAGS) -c $< -o $@
|
|
ggml_v2-opencl.o: otherarch/ggml_v2-opencl.cpp otherarch/ggml_v2-opencl.h
|
|
$(CXX) $(CXXFLAGS) $(CLBLAST_FLAGS) -c $< -o $@
|
|
ggml_v2-opencl-legacy.o: otherarch/ggml_v2-opencl-legacy.c otherarch/ggml_v2-opencl-legacy.h
|
|
$(CC) $(CFLAGS) -c $< -o $@
|
|
ggml_v3-opencl.o: otherarch/ggml_v3-opencl.cpp otherarch/ggml_v3-opencl.h
|
|
$(CXX) $(CXXFLAGS) $(CLBLAST_FLAGS) -c $< -o $@
|
|
|
|
#vulkan
|
|
ggml-vulkan.o: ggml-vulkan.cpp ggml-vulkan.h
|
|
$(CXX) $(CXXFLAGS) $(VULKAN_FLAGS) -c $< -o $@
|
|
|
|
# intermediate objects
|
|
llama.o: llama.cpp ggml.h ggml-alloc.h ggml-backend.h ggml-cuda.h ggml-metal.h llama.h otherarch/llama-util.h
|
|
$(CXX) $(CXXFLAGS) -c $< -o $@
|
|
common.o: common/common.cpp common/common.h common/log.h
|
|
$(CXX) $(CXXFLAGS) -c $< -o $@
|
|
console.o: common/console.cpp common/console.h
|
|
$(CXX) $(CXXFLAGS) -c $< -o $@
|
|
grammar-parser.o: common/grammar-parser.cpp common/grammar-parser.h
|
|
$(CXX) $(CXXFLAGS) -c $< -o $@
|
|
expose.o: expose.cpp expose.h
|
|
$(CXX) $(CXXFLAGS) -c $< -o $@
|
|
|
|
# sd.cpp objects
|
|
sdcpp_default.o: otherarch/sdcpp/sdtype_adapter.cpp otherarch/sdcpp/stable-diffusion.h otherarch/sdcpp/stable-diffusion.cpp otherarch/sdcpp/util.cpp otherarch/sdcpp/upscaler.cpp otherarch/sdcpp/model.cpp otherarch/sdcpp/thirdparty/zip.c
|
|
$(CXX) $(CXXFLAGS) -c $< -o $@
|
|
sdcpp_cublas.o: otherarch/sdcpp/sdtype_adapter.cpp otherarch/sdcpp/stable-diffusion.h otherarch/sdcpp/stable-diffusion.cpp otherarch/sdcpp/util.cpp otherarch/sdcpp/upscaler.cpp otherarch/sdcpp/model.cpp otherarch/sdcpp/thirdparty/zip.c
|
|
$(CXX) $(CXXFLAGS) $(CUBLAS_FLAGS) $(HIPFLAGS) -c $< -o $@
|
|
|
|
# idiotic "for easier compilation"
|
|
GPTTYPE_ADAPTER = gpttype_adapter.cpp otherarch/llama_v2.cpp otherarch/llama_v3.cpp llama.cpp otherarch/utils.cpp otherarch/gptj_v1.cpp otherarch/gptj_v2.cpp otherarch/gptj_v3.cpp otherarch/gpt2_v1.cpp otherarch/gpt2_v2.cpp otherarch/gpt2_v3.cpp otherarch/rwkv_v2.cpp otherarch/rwkv_v3.cpp otherarch/neox_v2.cpp otherarch/neox_v3.cpp otherarch/mpt_v3.cpp ggml.h ggml-cuda.h llama.h otherarch/llama-util.h
|
|
gpttype_adapter_failsafe.o: $(GPTTYPE_ADAPTER)
|
|
$(CXX) $(CXXFLAGS) $(FAILSAFE_FLAGS) -c $< -o $@
|
|
gpttype_adapter.o: $(GPTTYPE_ADAPTER)
|
|
$(CXX) $(CXXFLAGS) -c $< -o $@
|
|
gpttype_adapter_clblast.o: $(GPTTYPE_ADAPTER)
|
|
$(CXX) $(CXXFLAGS) $(CLBLAST_FLAGS) -c $< -o $@
|
|
gpttype_adapter_cublas.o: $(GPTTYPE_ADAPTER)
|
|
$(CXX) $(CXXFLAGS) $(CUBLAS_FLAGS) $(HIPFLAGS) -c $< -o $@
|
|
gpttype_adapter_clblast_noavx2.o: $(GPTTYPE_ADAPTER)
|
|
$(CXX) $(CXXFLAGS) $(FAILSAFE_FLAGS) $(CLBLAST_FLAGS) -c $< -o $@
|
|
gpttype_adapter_vulkan.o: $(GPTTYPE_ADAPTER)
|
|
$(CXX) $(CXXFLAGS) $(VULKAN_FLAGS) -c $< -o $@
|
|
gpttype_adapter_vulkan_noavx2.o: $(GPTTYPE_ADAPTER)
|
|
$(CXX) $(CXXFLAGS) $(FAILSAFE_FLAGS) $(VULKAN_FLAGS) -c $< -o $@
|
|
|
|
clean:
|
|
rm -vf *.o main sdmain quantize_gguf quantize_clip quantize_gpt2 quantize_gptj quantize_neox quantize_mpt quantize-stats perplexity embedding benchmark-matmult save-load-state gguf imatrix imatrix.exe gguf.exe main.exe quantize_clip.exe quantize_gguf.exe quantize_gptj.exe quantize_gpt2.exe quantize_neox.exe quantize_mpt.exe koboldcpp_default.dll koboldcpp_openblas.dll koboldcpp_failsafe.dll koboldcpp_noavx2.dll koboldcpp_clblast.dll koboldcpp_clblast_noavx2.dll koboldcpp_cublas.dll koboldcpp_hipblas.dll koboldcpp_vulkan.dll koboldcpp_vulkan_noavx2.dll koboldcpp_default.so koboldcpp_openblas.so koboldcpp_failsafe.so koboldcpp_noavx2.so koboldcpp_clblast.so koboldcpp_clblast_noavx2.so koboldcpp_cublas.so koboldcpp_hipblas.so koboldcpp_vulkan.so koboldcpp_vulkan_noavx2.so
|
|
rm -vrf ggml-cuda/*.o
|
|
|
|
# useful tools
|
|
main: examples/main/main.cpp common/sampling.cpp build-info.h ggml.o llama.o console.o llavaclip_default.o llava.o ggml-backend_default.o $(OBJS_FULL) $(OBJS)
|
|
$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
|
|
@echo '==== Run ./main -h for help. ===='
|
|
sdmain: otherarch/sdcpp/util.cpp otherarch/sdcpp/main.cpp otherarch/sdcpp/stable-diffusion.cpp otherarch/sdcpp/upscaler.cpp otherarch/sdcpp/model.cpp otherarch/sdcpp/thirdparty/zip.c build-info.h ggml.o llama.o console.o ggml-backend_default.o $(OBJS_FULL) $(OBJS)
|
|
$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
|
|
imatrix: examples/imatrix/imatrix.cpp common/sampling.cpp build-info.h ggml.o llama.o console.o llavaclip_default.o llava.o ggml-backend_default.o $(OBJS_FULL) $(OBJS)
|
|
$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
|
|
gguf: examples/gguf/gguf.cpp build-info.h ggml.o llama.o llavaclip_default.o llava.o ggml-backend_default.o $(OBJS_FULL) $(OBJS)
|
|
$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
|
|
gguf-split: examples/gguf-split/gguf-split.cpp ggml.o llama.o common/sampling.cpp build-info.h llavaclip_default.o llava.o ggml-backend_default.o $(OBJS_FULL) $(OBJS)
|
|
$(CXX) $(CXXFLAGS) $(filter-out %.h,$^) -o $@ $(LDFLAGS)
|
|
|
|
|
|
#generated libraries
|
|
koboldcpp_default: ggml.o ggml_v3.o ggml_v2.o ggml_v1.o expose.o gpttype_adapter.o sdcpp_default.o llavaclip_default.o llava.o ggml-backend_default.o $(OBJS_FULL) $(OBJS)
|
|
$(DEFAULT_BUILD)
|
|
|
|
ifdef OPENBLAS_BUILD
|
|
koboldcpp_openblas: ggml_v4_openblas.o ggml_v3_openblas.o ggml_v2_openblas.o ggml_v1.o expose.o gpttype_adapter.o sdcpp_default.o llavaclip_default.o llava.o ggml-backend_default.o $(OBJS_FULL) $(OBJS)
|
|
$(OPENBLAS_BUILD)
|
|
else
|
|
koboldcpp_openblas:
|
|
$(DONOTHING)
|
|
endif
|
|
|
|
ifdef FAILSAFE_BUILD
|
|
koboldcpp_failsafe: ggml_v4_failsafe.o ggml_v3_failsafe.o ggml_v2_failsafe.o ggml_v1_failsafe.o expose.o gpttype_adapter_failsafe.o sdcpp_default.o llavaclip_default.o llava.o ggml-backend_default.o $(OBJS_FAILSAFE) $(OBJS)
|
|
$(FAILSAFE_BUILD)
|
|
else
|
|
koboldcpp_failsafe:
|
|
$(DONOTHING)
|
|
endif
|
|
|
|
ifdef NOAVX2_BUILD
|
|
koboldcpp_noavx2: ggml_v4_noavx2.o ggml_v3_noavx2.o ggml_v2_noavx2.o ggml_v1_failsafe.o expose.o gpttype_adapter_failsafe.o sdcpp_default.o llavaclip_default.o llava.o ggml-backend_default.o $(OBJS_SIMPLE) $(OBJS)
|
|
$(NOAVX2_BUILD)
|
|
else
|
|
koboldcpp_noavx2:
|
|
$(DONOTHING)
|
|
endif
|
|
|
|
ifdef CLBLAST_BUILD
|
|
koboldcpp_clblast: ggml_v4_clblast.o ggml_v3_clblast.o ggml_v2_clblast.o ggml_v1.o expose.o gpttype_adapter_clblast.o ggml-opencl.o ggml_v3-opencl.o ggml_v2-opencl.o ggml_v2-opencl-legacy.o sdcpp_default.o llavaclip_default.o llava.o ggml-backend_default.o $(OBJS_FULL) $(OBJS)
|
|
$(CLBLAST_BUILD)
|
|
ifdef NOAVX2_BUILD
|
|
koboldcpp_clblast_noavx2: ggml_v4_clblast_noavx2.o ggml_v3_clblast_noavx2.o ggml_v2_clblast_noavx2.o ggml_v1_failsafe.o expose.o gpttype_adapter_clblast_noavx2.o ggml-opencl.o ggml_v3-opencl.o ggml_v2-opencl.o ggml_v2-opencl-legacy.o sdcpp_default.o llavaclip_default.o llava.o ggml-backend_default.o $(OBJS_SIMPLE) $(OBJS)
|
|
$(CLBLAST_BUILD)
|
|
else
|
|
koboldcpp_clblast_noavx2:
|
|
$(DONOTHING)
|
|
endif
|
|
else
|
|
koboldcpp_clblast:
|
|
$(DONOTHING)
|
|
koboldcpp_clblast_noavx2:
|
|
$(DONOTHING)
|
|
endif
|
|
|
|
ifdef CUBLAS_BUILD
|
|
koboldcpp_cublas: ggml_v4_cublas.o ggml_v3_cublas.o ggml_v2_cublas.o ggml_v1.o expose.o gpttype_adapter_cublas.o sdcpp_cublas.o llavaclip_cublas.o llava.o ggml-backend_cublas.o $(CUBLAS_OBJS) $(OBJS_FULL) $(OBJS)
|
|
$(CUBLAS_BUILD)
|
|
else
|
|
koboldcpp_cublas:
|
|
$(DONOTHING)
|
|
endif
|
|
|
|
ifdef HIPBLAS_BUILD
|
|
koboldcpp_hipblas: ggml_v4_cublas.o ggml_v3_cublas.o ggml_v2_cublas.o ggml_v1.o expose.o gpttype_adapter_cublas.o sdcpp_cublas.o llavaclip_cublas.o llava.o ggml-backend_cublas.o $(HIP_OBJS) $(OBJS_FULL) $(OBJS)
|
|
$(HIPBLAS_BUILD)
|
|
else
|
|
koboldcpp_hipblas:
|
|
$(DONOTHING)
|
|
endif
|
|
|
|
ifdef VULKAN_BUILD
|
|
koboldcpp_vulkan: ggml_v4_vulkan.o ggml_v3.o ggml_v2.o ggml_v1.o expose.o gpttype_adapter_vulkan.o ggml-vulkan.o sdcpp_default.o llavaclip_default.o llava.o ggml-backend_vulkan.o $(OBJS_FULL) $(OBJS)
|
|
$(VULKAN_BUILD)
|
|
ifdef NOAVX2_BUILD
|
|
koboldcpp_vulkan_noavx2: ggml_v4_vulkan_noavx2.o ggml_v3_noavx2.o ggml_v2_noavx2.o ggml_v1_failsafe.o expose.o gpttype_adapter_vulkan_noavx2.o ggml-vulkan.o sdcpp_default.o llavaclip_default.o llava.o ggml-backend_vulkan.o $(OBJS_SIMPLE) $(OBJS)
|
|
$(VULKAN_BUILD)
|
|
else
|
|
koboldcpp_vulkan_noavx2:
|
|
$(DONOTHING)
|
|
endif
|
|
else
|
|
koboldcpp_vulkan:
|
|
$(DONOTHING)
|
|
koboldcpp_vulkan_noavx2:
|
|
$(DONOTHING)
|
|
endif
|
|
|
|
# tools
|
|
quantize_gguf: examples/quantize/quantize.cpp ggml.o llama.o llavaclip_default.o llava.o ggml-backend_default.o $(OBJS_FULL)
|
|
$(CXX) $(CXXFLAGS) $^ -o $@ $(LDFLAGS)
|
|
quantize_gptj: otherarch/tools/gptj_quantize.cpp otherarch/tools/common-ggml.cpp ggml_v3.o ggml.o llama.o llavaclip_default.o llava.o ggml-backend_default.o $(OBJS_FULL)
|
|
$(CXX) $(CXXFLAGS) $^ -o $@ $(LDFLAGS)
|
|
quantize_gpt2: otherarch/tools/gpt2_quantize.cpp otherarch/tools/common-ggml.cpp ggml_v3.o ggml.o llama.o llavaclip_default.o llava.o ggml-backend_default.o $(OBJS_FULL)
|
|
$(CXX) $(CXXFLAGS) $^ -o $@ $(LDFLAGS)
|
|
quantize_neox: otherarch/tools/neox_quantize.cpp otherarch/tools/common-ggml.cpp ggml_v3.o ggml.o llama.o llavaclip_default.o llava.o ggml-backend_default.o $(OBJS_FULL)
|
|
$(CXX) $(CXXFLAGS) $^ -o $@ $(LDFLAGS)
|
|
quantize_mpt: otherarch/tools/mpt_quantize.cpp otherarch/tools/common-ggml.cpp ggml_v3.o ggml.o llama.o llavaclip_default.o llava.o ggml-backend_default.o $(OBJS_FULL)
|
|
$(CXX) $(CXXFLAGS) $^ -o $@ $(LDFLAGS)
|
|
quantize_clip: examples/llava/clip.cpp examples/llava/clip.h examples/llava/quantclip.cpp ggml_v3.o ggml.o llama.o ggml-backend_default.o $(OBJS_FULL)
|
|
$(CXX) $(CXXFLAGS) $^ -o $@ $(LDFLAGS)
|
|
|
|
#window simple clinfo
|
|
simpleclinfo: simpleclinfo.cpp
|
|
$(CXX) $(CXXFLAGS) $^ lib/OpenCL.lib lib/clblast.lib -o $@ $(LDFLAGS)
|
|
|
|
build-info.h:
|
|
$(DONOTHING)
|