diff options
| author | Johannes Gäßler <johannesg@5d6.de> | 2024-05-09 14:32:02 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-05-09 14:32:02 +0200 |
| commit | a743d76a01f23038b2c85af1e9048ee836767b44 (patch) | |
| tree | 8182fc85cb9fd055bc9c8268d5d4a05bcf87f57a /tests/test-backend-ops.cpp | |
| parent | f31ec120bc36c6270e4948e6a065a7c4cfa0c404 (diff) | |
CUDA: generalize FP16 fattn vec kernel (#7061)
* CUDA: generalize FP16 fattn vec kernel
* disable unsupported head sizes for AMD in test
* try AMD fix
* fix batch size 2-8
* partially revert changes
Diffstat (limited to 'tests/test-backend-ops.cpp')
| -rw-r--r-- | tests/test-backend-ops.cpp | 4 |
1 files changed, 4 insertions, 0 deletions
diff --git a/tests/test-backend-ops.cpp b/tests/test-backend-ops.cpp index 41718e00..0d66de5d 100644 --- a/tests/test-backend-ops.cpp +++ b/tests/test-backend-ops.cpp @@ -2175,7 +2175,11 @@ static bool test_backend(ggml_backend_t backend, test_mode mode, const char * op test_cases.emplace_back(new test_timestep_embedding()); test_cases.emplace_back(new test_leaky_relu()); +#if defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__) + for (int hs : { 64, 128, }) { // other head sizes not implemented +#else for (int hs : { 64, 80, 128, 256, }) { +#endif // defined(GGML_USE_HIPBLAS) && defined(__HIP_PLATFORM_AMD__) for (int nh : { 32, }) { for (int kv : { 512, 1024, }) { for (int nb : { 1, 2, 4, 8, }) { |