ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

CUDA: rename macros to avoid conflicts with WinAPI (#10736)

Browse source 
* Renames NVIDIA GPU-architecture flags to avoid name clashes with WinAPI. (e.g. CC_PASCAL, GPU architecture or WinAPI pascal compiler flag?)

* Reverts erroneous rename in SYCL-code.

* Renames GGML_CUDA_MIN_CC_DP4A to GGML_CUDA_CC_DP4A.

* Renames the rest of the compute capability macros for consistency.
This commit is contained in:
Andreas Kieslinger 2024-12-10 18:23:24 +01:00 • committed by GitHub
parent a86ad841f1
commit 750cb3e246
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
10 changed files with 69 additions and 71 deletions
Download patch file Download diff file Expand all files Collapse all files

10
ggml/src/ggml-cuda/mmq.cu
View file

@ -27,7 +27,7 @@ void ggml_cuda_op_mul_mat_q(
// The stream-k decomposition is only faster for recent NVIDIA GPUs.
// Also its fixup needs to allocate a temporary buffer in the memory pool.
// There are multiple parallel CUDA streams for src1_ncols != ne11 which would introduce a race condition for this buffer.
const bool use_stream_k = compute_capability >= CC_VOLTA && compute_capability < CC_OFFSET_AMD && src1_ncols == ne11;
const bool use_stream_k = compute_capability >= GGML_CUDA_CC_VOLTA && compute_capability < GGML_CUDA_CC_OFFSET_AMD && src1_ncols == ne11;
const mmq_args args = {src0_dd_i, src1_ddq_i, dst_dd_i, ne00, row_diff, stride00, src1_padded_row_size, src1_ncols, ne11, nrows_dst, use_stream_k};
switch (src0->type) {
@ -136,7 +136,7 @@ bool ggml_cuda_should_use_mmq(enum ggml_type type, int cc, int64_t ne11) {
return true;
}
if (cc < MIN_CC_DP4A) {
if (cc < GGML_CUDA_CC_DP4A) {
return false;
}
@ -144,9 +144,9 @@ bool ggml_cuda_should_use_mmq(enum ggml_type type, int cc, int64_t ne11) {
return true;
#endif //GGML_CUDA_FORCE_MMQ
if (cc < CC_OFFSET_AMD) {
return cc < CC_VOLTA || ne11 < MMQ_DP4A_MAX_BATCH_SIZE;
if (cc < GGML_CUDA_CC_OFFSET_AMD) {
return cc < GGML_CUDA_CC_VOLTA || ne11 < MMQ_DP4A_MAX_BATCH_SIZE;
}
return (cc < CC_RDNA3 && cc != CC_CDNA && cc != CC_VEGA20) || ne11 < MMQ_DP4A_MAX_BATCH_SIZE;
return (cc < GGML_CUDA_CC_RDNA3 && cc != GGML_CUDA_CC_CDNA && cc != GGML_CUDA_CC_VEGA20) || ne11 < MMQ_DP4A_MAX_BATCH_SIZE;
}