mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2025-04-16 03:26:08 +00:00
Code Issues Packages Projects Releases Wiki Activity

cuda/vulkan: specify fp32-only support for some operations in supports_op (ggml/1129)

Browse Source 
ggml-ci
This commit is contained in:
cmdr2 2025-02-28 12:36:46 +02:00 committed by Georgi Gerganov
parent 8371d44595
commit 0cbee131ad
3 changed files with 8 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
ggml/src/ggml-cuda/ggml-cuda.cu
View File

@ -3155,7 +3155,7 @@ static bool ggml_backend_cuda_device_supports_op(ggml_backend_dev_t dev, const g
return false;
} break;
case GGML_OP_SILU_BACK:
return ggml_is_contiguous(op->src[0]);
return ggml_is_contiguous(op->src[0]) && op->src[0]->type == GGML_TYPE_F32;
break;
case GGML_OP_NORM:
case GGML_OP_RMS_NORM:

11
ggml/src/ggml-vulkan/ggml-vulkan.cpp
View File

@ -8452,7 +8452,7 @@ static bool ggml_backend_vk_device_supports_op(ggml_backend_dev_t dev, const ggm
case GGML_UNARY_OP_RELU:
case GGML_UNARY_OP_TANH:
case GGML_UNARY_OP_SIGMOID:
return ggml_is_contiguous(op->src[0]);
return ggml_is_contiguous(op->src[0]) && op->src[0]->type == GGML_TYPE_F32;
default:
return false;
}
@ -8653,19 +8653,20 @@ static bool ggml_backend_vk_device_supports_op(ggml_backend_dev_t dev, const ggm
case GGML_OP_RMS_NORM:
return ggml_is_contiguous(op->src[0]);
case GGML_OP_ADD:
case GGML_OP_ACC:
case GGML_OP_SUB:
case GGML_OP_MUL:
case GGML_OP_DIV:
case GGML_OP_CONCAT:
case GGML_OP_SILU_BACK:
case GGML_OP_RMS_NORM_BACK:
case GGML_OP_UPSCALE:
case GGML_OP_SCALE:
case GGML_OP_SQR:
case GGML_OP_SIN:
case GGML_OP_COS:
case GGML_OP_CLAMP:
return op->src[0]->type == GGML_TYPE_F32;
case GGML_OP_ACC:
case GGML_OP_CONCAT:
case GGML_OP_UPSCALE:
case GGML_OP_SCALE:
case GGML_OP_PAD:
case GGML_OP_DIAG_MASK_INF:
case GGML_OP_SOFT_MAX:

5
tests/test-backend-ops.cpp
View File

@ -3998,10 +3998,7 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
test_cases.emplace_back(new test_add1());
test_cases.emplace_back(new test_scale());
for (ggml_type type : {GGML_TYPE_F16, GGML_TYPE_F32}) {
test_cases.emplace_back(new test_silu_back());
}
test_cases.emplace_back(new test_silu_back());
for (float eps : {0.0f, 1e-6f, 1e-4f, 1e-1f}) {
for (bool v : {false, true}) {