tree a3824e2e7d606abacc1ed279751a8bdf4b4dc1eb
parent 4bdbe4ad9dd25b432d405d8a6ac6b7410da9b7a2
author Artem Belevich <tra@google.com> 1596222068 -0700
committer Artem Belevich <tra@google.com> 1596222235 -0700

[test-suite, CUDA] Run test kernel with just one thread.

For some reason this test is flaky on sm_60+, with the flaky failures irrelevant
to what we're testing here. Reducing grid size should reduce the failure rate.