mirror of
https://github.com/ikawrakow/ik_llama.cpp.git
synced 2026-03-03 10:30:27 +00:00
13 lines
422 B
Markdown
13 lines
422 B
Markdown
### 🔀 [#343](https://github.com/ikawrakow/ik_llama.cpp/pull/343) - cuda: use switch in constexpr funcs
|
|
|
|
| **Author** | `ikawrakow` |
|
|
| :--- | :--- |
|
|
| **State** | ❌ **Closed** |
|
|
| **Created** | 2025-04-24 |
|
|
| **Updated** | 2025-04-24 |
|
|
|
|
---
|
|
|
|
#### Description
|
|
|
|
Based on [PR 13095](https://github.com/ggml-org/llama.cpp/pull/13095) in mainline. Did not measure, but had the impression that CUDA compile time is reduced. |