[PR #554] Add architecture during device linking #634

New issue

Open

opened 2026-10-27 21:04:43 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-27 21:04:43 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/deepseek-ai/DeepEP/pull/554
Author: @rzambre
Created: 1/3/2026
Status: 🔄 Open

Base: main ← Head: rzambre/fix-dlinking

📝 Commits (1)

190d7e6 Add architecture during device linking

📊 Changes

1 file changed (+3 additions, -1 deletions)

View changed files

📝 setup.py (+3 -1)

📄 Description

Without this fix, I see the following warnings when compiling DeepEP for a specific architecture, which leads to runtime errors with test_low_latency.py

nvlink warning : SM Arch ('sm_75') not found in 'DeepEP/build/temp.linux-x86_64-cpython-312/csrc/kernels/internode.o'
nvlink warning : SM Arch ('sm_75') not found in 'DeepEP/build/temp.linux-x86_64-cpython-312/csrc/kernels/internode_ll.o'
nvlink warning : SM Arch ('sm_75') not found in 'DeepEP/build/temp.linux-x86_64-cpython-312/csrc/kernels/intranode.o'
nvlink warning : SM Arch ('sm_75') not found in 'DeepEP/build/temp.linux-x86_64-cpython-312/csrc/kernels/layout.o'
nvlink warning : SM Arch ('sm_75') not found in 'DeepEP/build/temp.linux-x86_64-cpython-312/csrc/kernels/runtime.o'

Before this PR:

> cuobjdump -lelf DeepEP/build/lib.linux-x86_64-cpython-312/deep_ep_cpp.cpython-312-x86_64-linux-gnu.so 2>&1 | grep "sm_"
ELF file    1: deep_ep_cpp.cpython-312-x86_64-linux-gnu.1.sm_75.cubin

After this PR:

> cuobjdump -lelf DeepEP/build/lib.linux-x86_64-cpython-312/deep_ep_cpp.cpython-312-x86_64-linux-gnu.so 2>&1 | grep "sm_"
ELF file    1: deep_ep_cpp.cpython-312-x86_64-linux-gnu.1.sm_103.cubin

_{🔄 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/DeepEP/pull/554 **Author:** [@rzambre](https://github.com/rzambre) **Created:** 1/3/2026 **Status:** 🔄 Open **Base:** `main` ← **Head:** `rzambre/fix-dlinking` --- ### 📝 Commits (1) - [`190d7e6`](https://github.com/deepseek-ai/DeepEP/commit/190d7e63e86c94c8ecc4fe8131ba0a11fa2d9f1c) Add architecture during device linking ### 📊 Changes **1 file changed** (+3 additions, -1 deletions) <details> <summary>View changed files</summary> 📝 `setup.py` (+3 -1) </details> ### 📄 Description Without this fix, I see the following warnings when compiling DeepEP for a specific architecture, which leads to runtime errors with test_low_latency.py ``` nvlink warning : SM Arch ('sm_75') not found in 'DeepEP/build/temp.linux-x86_64-cpython-312/csrc/kernels/internode.o' nvlink warning : SM Arch ('sm_75') not found in 'DeepEP/build/temp.linux-x86_64-cpython-312/csrc/kernels/internode_ll.o' nvlink warning : SM Arch ('sm_75') not found in 'DeepEP/build/temp.linux-x86_64-cpython-312/csrc/kernels/intranode.o' nvlink warning : SM Arch ('sm_75') not found in 'DeepEP/build/temp.linux-x86_64-cpython-312/csrc/kernels/layout.o' nvlink warning : SM Arch ('sm_75') not found in 'DeepEP/build/temp.linux-x86_64-cpython-312/csrc/kernels/runtime.o' ``` Before this PR: ``` > cuobjdump -lelf DeepEP/build/lib.linux-x86_64-cpython-312/deep_ep_cpp.cpython-312-x86_64-linux-gnu.so 2>&1 | grep "sm_" ELF file 1: deep_ep_cpp.cpython-312-x86_64-linux-gnu.1.sm_75.cubin ``` After this PR: ``` > cuobjdump -lelf DeepEP/build/lib.linux-x86_64-cpython-312/deep_ep_cpp.cpython-312-x86_64-linux-gnu.so 2>&1 | grep "sm_" ELF file 1: deep_ep_cpp.cpython-312-x86_64-linux-gnu.1.sm_103.cubin ``` --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi added the

pull-request

label

2026-10-27 21:04:43 +00:00

bsakaguchi referenced this issue

2026-10-27 21:05:09 +00:00

[PR #634] [MERGED] Hybrid ep nixl restripe #675