Skip to content

Introduction of CUDA Programmatic Dependent Launch to Llama.cpp#15480

Closed
agray3 wants to merge 1 commit into
ggml-org:masterfrom
agray3:ag_cuda_programmatic_dependent_launch
Closed

Introduction of CUDA Programmatic Dependent Launch to Llama.cpp#15480
agray3 wants to merge 1 commit into
ggml-org:masterfrom
agray3:ag_cuda_programmatic_dependent_launch

Conversation

@agray3

@agray3 agray3 commented Aug 21, 2025

Copy link
Copy Markdown
Contributor

See #15479

Make sure to read the contributing guidelines before submitting a PR

Comment thread ggml/src/ggml-cuda/acc.cu
static __global__ void acc_f32(const float * x, const float * y, float * dst, const int64_t ne,
const int64_t ne10, const int64_t ne11, const int64_t ne12, const int64_t ne13,
const int64_t s11, const int64_t s12, const int64_t s13, const int64_t offset) {
#if !defined(GGML_USE_HIP) && __CUDA_ARCH__ >= GGML_CUDA_CC_HOPPER

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It might be better to define a dedicated macro and use it wherever needed. For example:

#if !defined(GGML_USE_HIP) && __CUDA_ARCH__ >= GGML_CUDA_CC_HOPPER
#define XXX_AVAILABLE
#endif // !defined(GGML_USE_HIP) && __CUDA_ARCH__ >= GGML_CUDA_CC_HOPPER

@agray3

agray3 commented Sep 26, 2025

Copy link
Copy Markdown
Contributor Author

Closing as per comments on #15479

@agray3 agray3 closed this Sep 26, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants