Lumpiasty/llama.cpp
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

ggml: support concat for scalar types at cuda backend (#24011)

Browse Source 
* cuda: support concat for scalar types

* Update concat.cu

* fix metal ci issue
This commit is contained in:
ZihaoMu
2026-06-12 09:32:44 +03:00
committed by GitHub
parent 099ea76fb4
commit 85f99dca8b
4 changed files with 106 additions and 62 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-cuda/ggml-cuda.cu
+9 -1
View File
@@ -5345,7 +5345,15 @@ static bool ggml_backend_cuda_device_supports_op(ggml_backend_dev_t dev, const g
case GGML_OP_CONCAT:
{
ggml_type src0_type = op->src[0]->type;
return src0_type != GGML_TYPE_I32 && src0_type != GGML_TYPE_I16;
ggml_type src1_type = op->src[1]->type;
return src0_type == src1_type &&
src0_type == op->type &&
!ggml_is_quantized(src0_type) &&
ggml_blck_size(src0_type) == 1 &&
(ggml_type_size(src0_type) == 1 ||
ggml_type_size(src0_type) == 2 ||
ggml_type_size(src0_type) == 4 ||
ggml_type_size(src0_type) == 8);
} break;
case GGML_OP_CONV_TRANSPOSE_1D:
{