vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

CUDA: Quantized matrix matrix multiplication (#2160)

Browse source 
* mmq implementation for non k-quants

* q6_K

* q2_K

* q3_k

* q4_K

* vdr

* q5_K

* faster q8_1 loading

* loop unrolling

* add __restrict__

* q2_K sc_high

* GGML_CUDA_MMQ_Y

* Updated Makefile

* Update Makefile

* DMMV_F16 -> F16

* Updated README, CMakeLists

* Fix CMakeLists.txt

* Fix CMakeLists.txt

* Fix multi GPU out-of-bounds
This commit is contained in:
Johannes Gäßler 2023-07-29 23:04:44 +02:00 • committed by GitHub
parent 9baf9ef304
commit 11f3ca06b8
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
4 changed files with 1295 additions and 324 deletions
Download patch file Download diff file Expand all files Collapse all files

15
Makefile
View file

@ -194,7 +194,7 @@ ifdef LLAMA_CUBLAS
CXXFLAGS += -DGGML_USE_CUBLAS -I/usr/local/cuda/include -I/opt/cuda/include -I$(CUDA_PATH)/targets/x86_64-linux/include
LDFLAGS += -lcublas -lculibos -lcudart -lcublasLt -lpthread -ldl -lrt -L/usr/local/cuda/lib64 -L/opt/cuda/lib64 -L$(CUDA_PATH)/targets/x86_64-linux/lib
OBJS += ggml-cuda.o
NVCCFLAGS = --forward-unknown-to-host-compiler
NVCCFLAGS = --forward-unknown-to-host-compiler -use_fast_math
ifdef LLAMA_CUDA_NVCC
NVCC = $(LLAMA_CUDA_NVCC)
else
@ -220,14 +220,25 @@ else ifdef LLAMA_CUDA_DMMV_Y
else
NVCCFLAGS += -DGGML_CUDA_MMV_Y=1
endif # LLAMA_CUDA_MMV_Y
ifdef LLAMA_CUDA_F16
NVCCFLAGS += -DGGML_CUDA_F16
endif # LLAMA_CUDA_F16
ifdef LLAMA_CUDA_DMMV_F16
NVCCFLAGS += -DGGML_CUDA_DMMV_F16
NVCCFLAGS += -DGGML_CUDA_F16
endif # LLAMA_CUDA_DMMV_F16
ifdef LLAMA_CUDA_KQUANTS_ITER
NVCCFLAGS += -DK_QUANTS_PER_ITERATION=$(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 # LLAMA_CUDA_MMQ_Y
ifdef LLAMA_CUDA_CUBLAS
NVCCFLAGS += -DGGML_CUDA_CUBLAS
endif # LLAMA_CUDA_CUBLAS
ifdef LLAMA_CUDA_CCBIN
NVCCFLAGS += -ccbin $(LLAMA_CUDA_CCBIN)
endif