NVIDIA / NVIDIA/MagnumIO

running cufile_sample_034 with difference parameters fail

Open
#18 2 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
C++
Stars
122
Forks
27
PR merge metrics
No merged PRs in 30d

Description

The test case with default parameters runs fine.

After changing MAX_BATCH_SIZE from 4 to 256, test report failure. (MAX_BATCH_SIZE 133 is the golden value to pass, 134 will fail)
cuFileReadAsync return value is still CU_FILE_SUCCESS while when sync stream and check for result in below code:
Image
read_bytes_done return -5 and test exit.

I have already put all big arrays in the test from stack to heap to avoid stack overflow.

Also, with 128 BATCH size, if i change MAX_BUF_SIZE from (1024 * 1024UL) to (64 * 1024 * 1024UL), test will also fail.
io error issued size:67108864 read:67104768 write:67108864
4096 bytes are missing

GPU is Nvidia A2
OS is debian 9
Env: NVIDIA-SMI 535.216.01 Driver Version: 535.216.01 CUDA Version: 12.2
FS: EXT-4

Contributor guide

Open the contributing guide

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 with the cufile_sample_034 test and reproduce the failure using the reported MAX_BATCH_SIZE and MAX_BUF_SIZE values on the stated NVIDIA A2, Debian 9, CUDA 12.2, and ext4 environment. Compare the asynchronous return status with read_bytes_done and verify that larger batches and buffers complete without missing bytes.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
testing-qa
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.