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-metal/ggml-metal-device.m
+10 -1
View File
@@ -1120,8 +1120,17 @@ bool ggml_metal_device_supports_op(ggml_metal_device_t dev, const struct ggml_te
case GGML_OP_VIEW:
case GGML_OP_TRANSPOSE:
case GGML_OP_PERMUTE:
case GGML_OP_CONCAT:
return true;
case GGML_OP_CONCAT:
{
// kernel_concat copies one float-sized value per element.
// Other scalar types need a type-generic copy kernel first.
const enum ggml_type src0_type = op->src[0]->type;
const enum ggml_type src1_type = op->src[1]->type;
return src0_type == src1_type &&
src0_type == op->type &&
(src0_type == GGML_TYPE_F32 || src0_type == GGML_TYPE_I32);
}
case GGML_OP_ADD:
case GGML_OP_SUB:
case GGML_OP_MUL: