diff options
| author | Peter Sugihara <peter@campsh.com> | 2023-11-03 12:18:18 -0700 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2023-11-03 21:18:18 +0200 |
| commit | d9b33fe95bd257b36c84ee5769cc048230067d6f (patch) | |
| tree | eb85223d65bda18d6f5e68c5f57d1a2df5ec01a1 | |
| parent | 5ba37461711095c0284233dbd14f0d9010cdbf56 (diff) | |
metal : round up to 16 to fix MTLDebugComputeCommandEncoder assertion (#3938)
| -rw-r--r-- | ggml-metal.m | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/ggml-metal.m b/ggml-metal.m index acdb8384..78ae4485 100644 --- a/ggml-metal.m +++ b/ggml-metal.m @@ -1017,7 +1017,7 @@ void ggml_metal_graph_compute( [encoder setBytes:&ne00 length:sizeof(ne00) atIndex:2]; [encoder setBytes:&ne01 length:sizeof(ne01) atIndex:3]; [encoder setBytes:&ne02 length:sizeof(ne02) atIndex:4]; - [encoder setThreadgroupMemoryLength:nth/32*sizeof(float) atIndex:0]; + [encoder setThreadgroupMemoryLength:MAX(16, nth/32*sizeof(float)) atIndex:0]; [encoder dispatchThreadgroups:MTLSizeMake(ne01*ne02*ne03, 1, 1) threadsPerThreadgroup:MTLSizeMake(nth, 1, 1)]; } break; @@ -1348,7 +1348,7 @@ void ggml_metal_graph_compute( [encoder setBytes:&ne00 length:sizeof( int64_t) atIndex:2]; [encoder setBytes:&nb01 length:sizeof(uint64_t) atIndex:3]; [encoder setBytes:&eps length:sizeof( float) atIndex:4]; - [encoder setThreadgroupMemoryLength:nth*sizeof(float) atIndex:0]; + [encoder setThreadgroupMemoryLength:MAX(16, nth*sizeof(float)) atIndex:0]; const int64_t nrows = ggml_nrows(src0); |