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

update cuda marco define

Browse source
This commit is contained in:
FantasyGmm 2023-12-22 11:24:42 +08:00
parent 3194769235
commit e6d10852d5
2 changed files with 2 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

1
Makefile
View file

@ -91,7 +91,6 @@ gcovr-report: coverage ## Generate gcovr report
mkdir -p gcovr-report
gcovr --root . --html --html-details --output gcovr-report/coverage.html
ifdef RISCV_CROSS_COMPILE
CC := riscv64-unknown-linux-gnu-gcc
CXX := riscv64-unknown-linux-gnu-g++

3
ggml-cuda.cu
View file

@ -84,7 +84,8 @@
#include <cuda_runtime.h>
#include <cublas_v2.h>
#include <cuda_fp16.h>
#if CUDA_VERSION < 1100
// CUDA 10.2 does not have these macro definitions.
#ifndef CUBLAS_TF32_TENSOR_OP_MATH
#define CUBLAS_TF32_TENSOR_OP_MATH CUBLAS_TENSOR_OP_MATH
#define CUBLAS_COMPUTE_16F CUDA_R_16F
#define CUBLAS_COMPUTE_32F CUDA_R_32F