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

llama : fix platforms without mmap (#4578)

Browse source 
* llama : fix platforms without mmap

* win32 : limit prefetch size to the file size

* fix win32 error clobber, unnecessary std::string in std::runtime_error
This commit is contained in:
slaren 2023-12-22 12:12:53 +01:00 • committed by GitHub
parent 48b24b170e
commit 48b7ff193e
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
3 changed files with 24 additions and 21 deletions
Download patch file Download diff file Expand all files Collapse all files

3
ggml-cuda.cu
View file

@ -7702,7 +7702,8 @@ inline void ggml_cuda_op_scale(
GGML_ASSERT(src0->type == GGML_TYPE_F32);
GGML_ASSERT( dst->type == GGML_TYPE_F32);
const float scale = ((float *) dst->op_params)[0];
float scale;
memcpy(&scale, dst->op_params, sizeof(float));
scale_f32_cuda(src0_dd, dst_dd, scale, ggml_nelements(src0), main_stream);
CUDA_CHECK(cudaGetLastError());