ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-02-24 07:04:11 +00:00
Code Issues Packages Projects Releases Wiki Activity

Try this

Browse Source
This commit is contained in:
Iwan Kawrakow
2025-10-21 19:37:16 +03:00
parent d5cfbca86b
commit c291fc056c
1 changed files with 1 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
ggml/src/ggml-cuda.cu
View File

@@ -3143,7 +3143,7 @@ static bool ggml_cuda_compute_forward(ggml_backend_cuda_context & ctx, struct gg
ggml_cuda_op_fused_add_add_rms_norm(ctx, dst, cgraph->nodes[i+1], cgraph->nodes[i+2]);
i += 2;
}
else if (i + 1 < cgraph->n_nodes &&
else if (false && i + 1 < cgraph->n_nodes &&
cgraph->nodes[i+1]->op == GGML_OP_FUSED_RMS_NORM &&
ggml_is_contiguous(dst->src[0]) &&
ggml_is_contiguous(dst->src[1]) &&
@@ -3154,7 +3154,6 @@ static bool ggml_cuda_compute_forward(ggml_backend_cuda_context & ctx, struct gg
} else {
ggml_cuda_op_add(ctx, dst);
}
//ggml_cuda_op_add(ctx, dst);
break;
case GGML_OP_ADD_ID:
ggml_cuda_op_add_id(ctx, dst);