NVIDIA / NVIDIA/cutlass

[BUG] Potential overflow issue in packed_stride.hpp

Open
#3,269 4 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

inactive-30d
Dominant language
C++
Stars
10.5k
Forks
2.1k
Avg merge
3d 11h
Merged PRs (30d)
7

Description

Hello,

I am no CUDA expert but stumble on a potential bug in my code and it seems there might be an issue here:
https://github.com/NVIDIA/cutlass/blob/546c3efa899ed1793d178a26fe764d63f93b49ea/tools/util/include/cutlass/util/packed_stride.hpp#L85

There is an upcast to IntT but I encoutered a situation where the two values are both int32 but their multiplication is overflowing, in my case:
49 152 × 65 536, leading to negative value even after upcast.
This issue might be present in the next block as well.

Kind regards

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by inspecting tools/util/include/cutlass/util/packed_stride.hpp at line 85 and the following block. Reproduce the reported case with 49,152 × 65,536 using int32 values, then verify that the packed-stride calculation does not overflow and that both mentioned blocks behave consistently.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
hpc
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
55/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.