search

rapidsai / rmm

RAPIDS Memory Manager
https://docs.rapids.ai/api/rmm/stable/
Apache License 2.0
492 stars 198 forks source link

[BUG] FailureCallbackResourceAdaptor pytests fail on GPUs with more than 100GiB of device memory #1733

Closed harrism closed 4 days ago

harrism commented 4 days ago

Describe the bug The following two tests expect exceptions when allocating more than 100GB of memory (1e11 bytes).

Steps/Code to reproduce bug Run RMM pytests on a >100GiB GPU, e.g. GH200.

Expected behavior Tests should pass.