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

llama : fix buffer checks for mamba and rwk (#10111)

Browse source 
* llama : fix buffer checks for mamba and rwk

* llama : fix missing worst case flag during reserve

* cuda : fix supports_op for norm

* disable sched SET_CAUSE
This commit is contained in:
Diego Devesa 2024-10-31 22:54:23 +01:00 • committed by GitHub
parent ab3d71f97f
commit c02e5ab2a6
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
4 changed files with 35 additions and 12 deletions
Download patch file Download diff file Expand all files Collapse all files

6
ggml/src/ggml-cuda.cu
View file

@ -3107,18 +3107,20 @@ static bool ggml_backend_cuda_device_supports_op(ggml_backend_dev_t dev, const g
}
return false;
} break;
case GGML_OP_NORM:
case GGML_OP_RMS_NORM:
return ggml_is_contiguous(op->src[0]) && op->ne[0] % WARP_SIZE == 0;
break;
case GGML_OP_NONE:
case GGML_OP_RESHAPE:
case GGML_OP_VIEW:
case GGML_OP_PERMUTE:
case GGML_OP_TRANSPOSE:
case GGML_OP_NORM:
case GGML_OP_ADD:
case GGML_OP_ADD1:
case GGML_OP_SUB:
case GGML_OP_MUL:
case GGML_OP_DIV:
case GGML_OP_RMS_NORM:
case GGML_OP_SCALE:
case GGML_OP_SQR:
case GGML_OP_SQRT: