[PR #61] [CLOSED] Add kernel for fp8 gemm with 1x128 LHS scaling and 1x128 RHS scaling. #224

Closed
opened 2026-10-26 21:15:03 +00:00 by bsakaguchi · 0 comments
Owner

📋 Pull Request Information

Original PR: https://github.com/deepseek-ai/DeepGEMM/pull/61
Author: @yukavio
Created: 3/13/2025
Status: ❌ Closed

Base: main ← Head: main


📝 Commits (4)

📊 Changes

6 files changed (+673 additions, -4 deletions)

View changed files

📝 deep_gemm/__init__.py (+1 -0)
➕ deep_gemm/include/deep_gemm/fp8_gemm_backward_w.cuh (+440 -0)
📝 deep_gemm/include/deep_gemm/scheduler.cuh (+1 -0)
📝 deep_gemm/jit_kernels/__init__.py (+1 -0)
➕ deep_gemm/jit_kernels/gemm_bw.py (+186 -0)
📝 tests/test_core.py (+44 -4)

📄 Description

Add kernel for fp8 gemm with 1x128 LHS scaling and 1x128 RHS scaling for backward of weight in training.


🔄 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/61 **Author:** [@yukavio](https://github.com/yukavio) **Created:** 3/13/2025 **Status:** ❌ Closed **Base:** `main` ← **Head:** `main` --- ### 📝 Commits (4) - [`6e53c66`](https://github.com/deepseek-ai/DeepGEMM/commit/6e53c6613d35cc770bb1d04ddebe3b17d2698be2) tested - [`e22d67d`](https://github.com/deepseek-ai/DeepGEMM/commit/e22d67d4517e27efe3c0502a8f515aa8ec892f4e) Merge pull request #1 from yukavio/add_backward_w - [`094d042`](https://github.com/deepseek-ai/DeepGEMM/commit/094d0421eca0fb7378029f395e7535fb13802d36) refine - [`bcd5fe8`](https://github.com/deepseek-ai/DeepGEMM/commit/bcd5fe8271e1ada20083ef9bee4a9caffb4e79b1) Merge pull request #2 from yukavio/add_backward_w ### 📊 Changes **6 files changed** (+673 additions, -4 deletions) <details> <summary>View changed files</summary> 📝 `deep_gemm/__init__.py` (+1 -0) ➕ `deep_gemm/include/deep_gemm/fp8_gemm_backward_w.cuh` (+440 -0) 📝 `deep_gemm/include/deep_gemm/scheduler.cuh` (+1 -0) 📝 `deep_gemm/jit_kernels/__init__.py` (+1 -0) ➕ `deep_gemm/jit_kernels/gemm_bw.py` (+186 -0) 📝 `tests/test_core.py` (+44 -4) </details> ### 📄 Description Add kernel for fp8 gemm with 1x128 LHS scaling and 1x128 RHS scaling for backward of weight in training. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
bsakaguchi 2026-10-26 21:15:03 +00:00
Sign in to join this conversation.
No labels
pull-request
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
deepseek-ai/DeepGEMM#224
No description provided.