[AzureCI] Add all_reduce_bias to rccl-tests CI (#2768)
* [AzureCI] Add all_reduce_bias to rccl-tests CI * Increase rccl-tests timeout to 2 hours --------- Co-authored-by: nileshnegi <Nilesh.Negi@amd.com> Co-authored-by: systems-assistant[bot] <systems-assistant[bot]@users.noreply.github.com> Co-authored-by: Corey Derochie <161367113+corey-derochie-amd@users.noreply.github.com>
Šī revīzija ir iekļauta:
revīziju iesūtīja
GitHub
vecāks
821a6e0700
revīzija
4358cad858
@@ -2,7 +2,7 @@
|
||||
#SBATCH --job-name=rccl-tests
|
||||
#SBATCH --output=%x-%j.out
|
||||
#SBATCH --error=%x-%j.out
|
||||
#SBATCH --time=60
|
||||
#SBATCH --time=120
|
||||
#SBATCH --nodes=1
|
||||
#SBATCH --exclusive
|
||||
#SBATCH --partition=gt
|
||||
@@ -40,7 +40,7 @@ do
|
||||
total=$((n*8))
|
||||
#h_8ppn=`echo ${hosts_8ppn[@]:0:${n}} | tr ' ' ','`
|
||||
|
||||
for coll in all_reduce all_gather reduce_scatter alltoall alltoallv broadcast gather reduce scatter sendrecv
|
||||
for coll in all_reduce all_reduce_bias all_gather reduce_scatter alltoall alltoallv broadcast gather reduce scatter sendrecv
|
||||
do
|
||||
for dtype in float bfloat16 half fp8_e5m2
|
||||
do
|
||||
|
||||
Atsaukties uz šo jaunā problēmā
Block a user