[PR #151] [MERGED] Make various updates and fixes #266

New issue

Closed

opened 2026-10-26 21:15:31 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-26 21:15:31 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/deepseek-ai/DeepGEMM/pull/151
Author: @RayWang96
Created: 8/3/2025
Status: ✅ Merged
Merged: 8/3/2025
Merged by: @RayWang96

Base: main ← Head: update_jit

📝 Commits (1)

d9c363f Make various updates and fixes:

📊 Changes

36 files changed (+592 additions, -362 deletions)

View changed files

📝 CMakeLists.txt (+2 -2)
📝 README.md (+4 -3)
➕ build.sh (+12 -0)
📝 csrc/jit/cache.hpp (+1 -1)
📝 csrc/jit/compiler.hpp (+128 -21)
📝 csrc/jit/device_runtime.hpp (+12 -2)
➕ csrc/jit/handle.hpp (+135 -0)
📝 csrc/jit/kernel_runtime.hpp (+20 -42)
📝 csrc/jit_kernels/heuristics/common.hpp (+11 -5)
📝 csrc/jit_kernels/heuristics/sm100.hpp (+7 -2)
📝 csrc/jit_kernels/impls/sm100_fp8_gemm_1d1d.hpp (+8 -16)
📝 csrc/jit_kernels/impls/sm100_fp8_gemm_1d2d.hpp (+5 -11)
📝 csrc/jit_kernels/impls/sm90_fp8_gemm_1d2d.hpp (+4 -11)
📝 csrc/jit_kernels/impls/smxx_layout.hpp (+6 -51)
📝 csrc/python_api.cpp (+27 -17)
📝 csrc/utils/exception.hpp (+10 -0)
➕ csrc/utils/lazy_init.hpp (+27 -0)
📝 csrc/utils/system.hpp (+2 -2)
📝 deep_gemm/__init__.py (+7 -2)
➕ deep_gemm/include/deep_gemm/common/cute_tie.cuh (+48 -0)

...and 16 more files

📄 Description

Add support for legacy CUDA versions; now compatible with CUDA 12.3 and newer
Add support for NVRTC compilation
Other fixes and code refactoring

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/deepseek-ai/DeepGEMM/pull/151 **Author:** [@RayWang96](https://github.com/RayWang96) **Created:** 8/3/2025 **Status:** ✅ Merged **Merged:** 8/3/2025 **Merged by:** [@RayWang96](https://github.com/RayWang96) **Base:** `main` ← **Head:** `update_jit` --- ### 📝 Commits (1) - [`d9c363f`](https://github.com/deepseek-ai/DeepGEMM/commit/d9c363f86f20b38dd852191d6680b10059fb61d5) Make various updates and fixes: ### 📊 Changes **36 files changed** (+592 additions, -362 deletions) <details> <summary>View changed files</summary> 📝 `CMakeLists.txt` (+2 -2) 📝 `README.md` (+4 -3) ➕ `build.sh` (+12 -0) 📝 `csrc/jit/cache.hpp` (+1 -1) 📝 `csrc/jit/compiler.hpp` (+128 -21) 📝 `csrc/jit/device_runtime.hpp` (+12 -2) ➕ `csrc/jit/handle.hpp` (+135 -0) 📝 `csrc/jit/kernel_runtime.hpp` (+20 -42) 📝 `csrc/jit_kernels/heuristics/common.hpp` (+11 -5) 📝 `csrc/jit_kernels/heuristics/sm100.hpp` (+7 -2) 📝 `csrc/jit_kernels/impls/sm100_fp8_gemm_1d1d.hpp` (+8 -16) 📝 `csrc/jit_kernels/impls/sm100_fp8_gemm_1d2d.hpp` (+5 -11) 📝 `csrc/jit_kernels/impls/sm90_fp8_gemm_1d2d.hpp` (+4 -11) 📝 `csrc/jit_kernels/impls/smxx_layout.hpp` (+6 -51) 📝 `csrc/python_api.cpp` (+27 -17) 📝 `csrc/utils/exception.hpp` (+10 -0) ➕ `csrc/utils/lazy_init.hpp` (+27 -0) 📝 `csrc/utils/system.hpp` (+2 -2) 📝 `deep_gemm/__init__.py` (+7 -2) ➕ `deep_gemm/include/deep_gemm/common/cute_tie.cuh` (+48 -0) _...and 16 more files_ </details> ### 📄 Description - Add support for legacy CUDA versions; now compatible with CUDA 12.3 and newer - Add support for NVRTC compilation - Other fixes and code refactoring --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-26 21:15:31 +00:00