nccl-tests/verifiable
John Bachan 51af5572bf Resync with NCCL 2.13
* Added "verifiable", a suite of kernels for generating and verifying reduction
  input and output arrays in a bit-precise way.
* Data corruption errors now reported in number of wrong elements instead of max
  deviation.
* Use ncclGetLastError.
* Don't run hypercube on non-powers of 2 ranks.
* Fix to hypercube data verification.
* Use "thread local" as the defaut CUDA capture mode.
* Replaced pthread_yield -> sched_yield()
* Bugfix to the cpu-side barrier/allreduce implementations.
2022-08-22 17:51:06 -07:00
..
inexact_regress.cu Resync with NCCL 2.13 2022-08-22 17:51:06 -07:00
Makefile Resync with NCCL 2.13 2022-08-22 17:51:06 -07:00
verifiable.cu Resync with NCCL 2.13 2022-08-22 17:51:06 -07:00
verifiable.h Resync with NCCL 2.13 2022-08-22 17:51:06 -07:00
verifiable.mk Resync with NCCL 2.13 2022-08-22 17:51:06 -07:00