thek0tyara/llama-cpp-turboquant
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 4

llama: use host memory if device reports 0 memory (#18587)

Browse source
This commit is contained in:
Aaron Teo 2026-01-09 05:34:56 +08:00 • committed by GitHub
parent 480160d472
commit 046d5fd44e
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
4 changed files with 28 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml/src/ggml-backend-impl.h
View file

@ -144,7 +144,7 @@ extern "C" {
// device description: short informative description of the device, could be the model name
const char * (*get_description)(ggml_backend_dev_t dev);
// device memory in bytes
// device memory in bytes: 0 bytes to indicate no memory to report
void (*get_memory)(ggml_backend_dev_t dev, size_t * free, size_t * total);
// device type

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

@ -4287,8 +4287,8 @@ static const char * ggml_backend_opencl_device_get_description(ggml_backend_dev_
}
static void ggml_backend_opencl_device_get_memory(ggml_backend_dev_t dev, size_t * free, size_t * total) {
*free = 1;
*total = 1;
*free = 0;
*total = 0;
GGML_UNUSED(dev);
}