[PR #246] [CLOSED] docs: add DG_MINIMIZE_NUM_SMS environment variable to README #310

Closed
opened 2026-10-26 21:16:07 +00:00 by bsakaguchi · 0 comments
Owner

📋 Pull Request Information

Original PR: https://github.com/deepseek-ai/DeepGEMM/pull/246
Author: @yurekami
Created: 12/25/2025
Status: ❌ Closed

Base: main ← Head: docs/add-minimize-sms-envvar


📝 Commits (1)

  • 4899190 docs: add DG_MINIMIZE_NUM_SMS environment variable to README

📊 Changes

1 file changed (+1 additions, -0 deletions)

View changed files

📝 README.md (+1 -0)

📄 Description

Summary

Add documentation for the new DG_MINIMIZE_NUM_SMS environment variable to the README.

This environment variable was added in PR #244 (addressing issue #239) to control whether the kernel should minimize the number of SMs used.

Changes

Added to the "Heuristic selection" section:

  • DG_MINIMIZE_NUM_SMS: 0 or 1, minimize the number of SMs used for better L2 cache usage and reduced GPU frequency drops; may cause additional kernel compilations on first run, 1 by default

Usage

# Disable min SMS optimization for faster startup (TTFT)
export DG_MINIMIZE_NUM_SMS=0

# Enable min SMS optimization (default behavior)
export DG_MINIMIZE_NUM_SMS=1

🤖 Generated with Claude Code


🔄 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/246 **Author:** [@yurekami](https://github.com/yurekami) **Created:** 12/25/2025 **Status:** ❌ Closed **Base:** `main` ← **Head:** `docs/add-minimize-sms-envvar` --- ### 📝 Commits (1) - [`4899190`](https://github.com/deepseek-ai/DeepGEMM/commit/48991906c240cbc3b214beb429e97aa5b80f7299) docs: add DG_MINIMIZE_NUM_SMS environment variable to README ### 📊 Changes **1 file changed** (+1 additions, -0 deletions) <details> <summary>View changed files</summary> 📝 `README.md` (+1 -0) </details> ### 📄 Description ## Summary Add documentation for the new `DG_MINIMIZE_NUM_SMS` environment variable to the README. This environment variable was added in PR #244 (addressing issue #239) to control whether the kernel should minimize the number of SMs used. ## Changes Added to the "Heuristic selection" section: - `DG_MINIMIZE_NUM_SMS`: `0` or `1`, minimize the number of SMs used for better L2 cache usage and reduced GPU frequency drops; may cause additional kernel compilations on first run, `1` by default ## Usage ```bash # Disable min SMS optimization for faster startup (TTFT) export DG_MINIMIZE_NUM_SMS=0 # Enable min SMS optimization (default behavior) export DG_MINIMIZE_NUM_SMS=1 ``` 🤖 Generated with [Claude Code](https://claude.com/claude-code) --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
bsakaguchi 2026-10-26 21:16:07 +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#310
No description provided.