Skip to content

CUDA: enable Gemma FA for HIP/Pascal #9581

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Sep 22, 2024
Merged

CUDA: enable Gemma FA for HIP/Pascal #9581

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 8 additions & 8 deletions ggml/src/ggml-cuda.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2968,19 +2968,19 @@ GGML_CALL static bool ggml_backend_cuda_supports_op(ggml_backend_t backend, cons
case GGML_OP_TIMESTEP_EMBEDDING:
case GGML_OP_LEAKY_RELU:
return true;
case GGML_OP_FLASH_ATTN_EXT:
#if defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)
return (op->src[0]->ne[0] == 64 && op->src[1]->type == GGML_TYPE_F16) || op->src[0]->ne[0] == 128;
#else
case GGML_OP_FLASH_ATTN_EXT: {
if (op->src[0]->ne[0] == 64 && op->src[1]->type == GGML_TYPE_F16) {
return true;
}
if (op->src[0]->ne[0] == 128) {
return true;
}
if (op->src[0]->ne[0] == 64 && op->src[1]->type == GGML_TYPE_F16) {
if (op->src[0]->ne[0] == 256 && op->src[1]->type == GGML_TYPE_F16 && op->src[2]->type == GGML_TYPE_F16) {
return true;
}
return ggml_cuda_info().devices[cuda_ctx->device].cc >= CC_VOLTA &&
op->src[1]->type == GGML_TYPE_F16 && op->src[2]->type == GGML_TYPE_F16;
#endif // defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__)
const int cc = ggml_cuda_info().devices[cuda_ctx->device].cc;
return cc >= CC_VOLTA && cc < CC_OFFSET_AMD && op->src[1]->type == GGML_TYPE_F16 && op->src[2]->type == GGML_TYPE_F16;
}
case GGML_OP_CROSS_ENTROPY_LOSS:
case GGML_OP_CROSS_ENTROPY_LOSS_BACK:
case GGML_OP_OPT_STEP_ADAMW:
Expand Down
2 changes: 1 addition & 1 deletion ggml/src/ggml-cuda/fattn.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -314,7 +314,7 @@ void ggml_cuda_flash_attn_ext(ggml_backend_cuda_context & ctx, ggml_tensor * dst
}

if (!fast_fp16_available(cc)) {
if (Q->ne[1] <= 8) {
if (Q->ne[1] <= 8 || Q->ne[0] == 256) {
ggml_cuda_flash_attn_ext_vec_f32(ctx, dst);
} else {
ggml_cuda_flash_attn_ext_tile_f32(ctx, dst);
Expand Down
2 changes: 1 addition & 1 deletion tests/test-backend-ops.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3564,7 +3564,7 @@ static bool test_backend(ggml_backend_t backend, test_mode mode, const char * op
if (hs != 128 && logit_softcap != 0.0f) continue;
for (int nh : { 32, }) {
for (int kv : { 512, 1024, }) {
for (int nb : { 1, 2, 4, 8, }) {
for (int nb : { 1, 3, 32, 35, }) {
for (ggml_type type_KV : {GGML_TYPE_F16, GGML_TYPE_Q8_0, GGML_TYPE_Q4_0}) {
test_cases.emplace_back(new test_flash_attn_ext(hs, nh, kv, nb, mask, max_bias, logit_softcap, type_KV));
}
Expand Down
Loading