QMCPACK / QMCPACK/qmcpack

deterministic-unit_test_afqmc_numerics broken in CUDA offload build

Open
#3,114 0 comments 0 reactions 1 assignee Claimed by @mmorale3 View on GitHub
bug
Dominant language
C++
Stars
403
Forks
154
Avg merge
1d 12h
Merged PRs (30d)
82

Description

**Describe the bug**

e.g. https://cdash.qmcpack.org/CDash/testDetails.php?test=12931331&build=202774

Looks to be a simple issue around memory:

```
Rank = 0 Free Memory = 22150 MB
Running in node with 1 GPUs.
CUDA compute capability: 6.1
Device Name: Quadro P1000
Warning CUDA major compute capability < 6.0
Error from calling cudaFree: an illegal memory access was encountered
Error from calling cudaFree: an illegal memory access was encountered
cusolver_gqr_strided_status
...
[0;37m/scratch/QMCPACK_CI_BUILDS_DO_NOT_REMOVE/qmcpack-develop/src/AFQMC/Numerics/tests/test_dense_numerics.cpp:754: [0m [1;31mFAILED:
[0m [0;36m {Unknown expression after the reported line}
[0mdue to unexpected exception with message:
Error code returned by cusolver.
...
... (more)
```

@ye-luo , @mmorale3 ?

**To Reproduce**

Nightly test configuration as per top URL. Looks to be offload CUDA builds only.

**Expected behavior**

Test passes

**System:**

bora.

**Additional context**

Deterministic tests and unit tests should always pass.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.