EngineX-Ascend/enginex-ascend-910-llama.cpp
10
0
Fork 0
Code Issues Pull Requests Actions 4 Projects Releases Wiki Activity

vulkan: handle large sizes for get_rows (#15686)

Browse Source
This commit is contained in:
Jeff Bolz
2025-08-31 03:13:27 -05:00
committed by GitHub
parent c37052ab4d
commit bbbf5ecccb
3 changed files with 46 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
ggml/src/ggml-vulkan/ggml-vulkan.cpp
View File

@@ -7849,6 +7849,8 @@ static void ggml_vk_op_f32(ggml_backend_vk_context * ctx, vk_context& subctx, co
break;
case GGML_OP_GET_ROWS:
elements = { (uint32_t)ne00, (uint32_t)ne10, (uint32_t)(ne11 * ne12) };
elements[1] = std::min(elements[1], ctx->device->properties.limits.maxComputeWorkGroupCount[1]);
elements[2] = std::min(elements[2], ctx->device->properties.limits.maxComputeWorkGroupCount[2]);
break;
case GGML_OP_ARGSORT:
elements = { (uint32_t)ne00, (uint32_t)ggml_nrows(src0), 1 };