ClementLinCF cab25f919e Optimize NCHANNELS and MSCCL config for gfx942 80CUs (#1195) ...

* Optimize NCHANNELS and MSCCL config for gfx942 80CUs

Set appropriately for different NCCL_MIN_NCHANNELS and MSCCL config,
potentially improving communication perf on the MI300x 80CUs

* Delete tools/msccl-algorithms/allreduce_1step_mccl_8_2_16777216_LL.xml

* Change the factor of gfx94 and update msccl config

2024-06-01 07:07:46 -07:00

..

allgather-8n-0-8kb.xml

msccl algorithms tuning for allgather on MI300 (#1110)

2024-03-14 12:18:26 -04:00

allgather-8n-1mb-40mb.xml

msccl algorithms tuning for allgather on MI300 (#1110)

2024-03-14 12:18:26 -04:00

allgather-8n-8kb-128kb.xml

msccl algorithms tuning for allgather on MI300 (#1110)

2024-03-14 12:18:26 -04:00

allgather-8n-128kb-1mb.xml

msccl algorithms tuning for allgather on MI300 (#1110)

2024-03-14 12:18:26 -04:00

allgather-allpairs-16n-16tb.xml

Introduce allgather MSCCL XML specification for MI250X up to 320KB. (#930)

2023-10-24 18:35:55 -05:00

allreduce-1step-4n-ll-1pass.xml

Optimize NCHANNELS and MSCCL config for gfx942 80CUs (#1195)

2024-06-01 07:07:46 -07:00

allreduce-allpairs-8n-ll-1pass-op.xml

MSCCL: add support for out-of-place all reduce (#1156)

2024-04-28 19:49:09 -07:00

allreduce-allpairs-8n-ll-1pass.xml

msccl algorithms tuning for allreduce on MI300 (#1088)

2024-02-21 11:31:56 -05:00

allreduce-allpairs-8n-ll-32tb-op.xml

MSCCL: add support for out-of-place all reduce (#1156)

2024-04-28 19:49:09 -07:00

allreduce-allpairs-8n-ll-32tb.xml

msccl algorithms tuning for allreduce on MI300 (#1088)

2024-02-21 11:31:56 -05:00

allreduce-allpairs-8n-ll-64tb-op.xml

MSCCL: add support for out-of-place all reduce (#1156)

2024-04-28 19:49:09 -07:00

allreduce-allpairs-8n-ll-64tb.xml

msccl algorithms tuning for allreduce on MI300 (#1088)

2024-02-21 11:31:56 -05:00

allreduce-allpairs-8n-simple_2.xml

msccl algorithms tuning for allreduce on MI300 (#1088)

2024-02-21 11:31:56 -05:00

allreduce-allpairs-8n-simple-op.xml

MSCCL: add support for out-of-place all reduce (#1156)

2024-04-28 19:49:09 -07:00

allreduce-allpairs-8n-simple.xml

msccl algorithms tuning for allreduce on MI300 (#1088)

2024-02-21 11:31:56 -05:00

allreduce-allpairs-16n-16tb.xml

Introduce 1-shot allreduce for MI250X Hayabusa. (#929)

2023-10-24 16:31:18 -05:00

allreduce-allpairs-16n-32tb.xml

MSCCL: Improve executor and integrate scheduler (#694)

2023-03-14 14:34:25 -07:00

alltoall-8n-0-9kb.xml

msccl algorithms tuning for alltoall on MI300 (#1120)

2024-03-21 20:35:29 -04:00

alltoall-8n-7mb-43mb.xml

msccl algorithms tuning for alltoall on MI300 (#1120)

2024-03-21 20:35:29 -04:00

alltoall-8n-9kb-190kb.xml

msccl algorithms tuning for alltoall on MI300 (#1120)

2024-03-21 20:35:29 -04:00

alltoall-8n-190kb-512kb.xml

msccl algorithms tuning for alltoall on MI300 (#1120)

2024-03-21 20:35:29 -04:00

alltoall-8n-512kb-7mb.xml

msccl algorithms tuning for alltoall on MI300 (#1120)

2024-03-21 20:35:29 -04:00