vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

sycl : Fixes to broken builds and test-backend-ops (#10257)

Browse source 
* Fixes broken build for the SYCL CUDA backend caused by non-explicit gemm call in outprod (merged in with RWKV6 in
Optimize RWKV6 Operator Naming and Implement Multi-core CPU/ SYCL Acceleration #10133)

* Marks permuted MUL_MAT as unsupported to be able to run test-backend-ops

* Fixes asserts in norm to fix debug builds.
This commit is contained in:
Alberto Cabrera Pérez 2024-11-13 09:40:57 +00:00 • committed by GitHub
parent 80dd7ff22f
commit 2e82ffa4af
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 9 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

4
ggml/src/ggml-sycl.cpp
View file

@ -4350,6 +4350,10 @@ static bool ggml_backend_sycl_device_supports_op(ggml_backend_dev_t dev, const g
if (op->op == GGML_OP_MUL_MAT) {
a = op->src[0];
b = op->src[1];
if (ggml_is_permuted(a) || ggml_is_permuted(b)) {
// TODO: fix like https://github.com/ggerganov/llama.cpp/pull/10021
return false;
}
} else {
a = op->src[2];
b = op->src[1];