Hi Howard,

                 What was the latency with the 4.3.x branch?

 

Rajeev

 

 

From: Howard Pritchard via discuss <discuss@mpich.org>
Reply-To: "discuss@mpich.org" <discuss@mpich.org>
Date: Wednesday, July 30, 2025 at 1:43 PM
To: "Zhou, Hui" <zhouh@anl.gov>
Cc: Howard Pritchard <hppritcha@gmail.com>, "discuss@mpich.org" <discuss@mpich.org>
Subject: Re: [mpich-discuss] MPICH 5.0.1 performance on HPE SS11 plus more - a slurm problem

 

Hi Hui That didn’t help. I am not surprised though as our cluster is an NVIDIA free zone. What did help is to switch to the mpich 4. 3. x branch and latency results are nominal and the slurm problem went away too. So we will stick with that branch. 

ZjQcmQRYFpfptBannerStart

This Message Is From an External Sender This message came from outside your organization.

ZjQcmQRYFpfptBannerEnd

Hi Hui

 

That didn’t help.  I am not surprised though as our cluster is an NVIDIA free zone.  What did help is to switch to the mpich 4.3.x branch and latency results are nominal and the slurm problem went away too.  So we will stick with that branch.

 

Howard

 

On Mon, Jul 28, 2025 at 4:15 PM Zhou, Hui <zhouh@anl.gov> wrote:

Hi Howard,

 

 I wonder whether it is due to the overhead of querying pointer attributes. Could you try disable GPU support with `MPIR_CVAR_ENABLE_GPU=0` and see if the latency improves?

 

Hui

---

From: Howard Pritchard via discuss <discuss@mpich.org>
Sent: Monday, July 28, 2025 9:41 AM
To: discuss@mpich.org <discuss@mpich.org>
Cc: Howard Pritchard <hppritcha@gmail.com>
Subject: [mpich-discuss] MPICH 5.0.1 performance on HPE SS11 plus more - a slurm problem

 

Hi Folks, We are seeing a strange performance issue on our HPE SS11 system when testing osu_latency inter-node with MPICH. First the info: system using libfabric 1. 22. 0 slurm - 24. 11. 5 Here's my mpichversion output: MPICH Version:       5. 0. 0a1

ZjQcmQRYFpfptBannerStart

This Message Is From an External Sender

This message came from outside your organization.

 

ZjQcmQRYFpfptBannerEnd

Hi Folks,

 

We are seeing a strange performance issue on our HPE SS11 system when testing osu_latency inter-node with MPICH.

 

First the info:

system using libfabric 1.22.0

slurm - 24.11.5

 

Here's my mpichversion output:

 

MPICH Version:      5.0.0a1

MPICH Release date: unreleased development copy

MPICH ABI:          0:0:0

MPICH Device:       ch4:ofi

MPICH configure:    --prefix=/XXXX/mpich_again/install --enable-g=no --enable-error-checking=no --with-device=ch4:ofi --enable-threads=multiple --with-ch4-shmmods=posix,xpmem --enable-thread-cs=per-vci --with-libfabric=/opt/cray/libfabric/1.22.0 --with-xpmem=/opt/cray/xpmem/default --with-pmix=/opt/pmix/gcc4x/5.0.8 --enable-fast=O3

MPICH CC:           gcc     -O3

MPICH CXX:          g++   -O3

MPICH F77:          gfortran   -O3

MPICH FC:           gfortran   -O3

MPICH features:     threadcomm



And here's the OSU latency results:

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_belong_chk: nid001439 [1]: pmixp_coll.c:280: No process controlled by this slurmstepd is involved in this collective.

slurmstepd: error:  mpi/pmix_v4: _process_server_request: nid001439 [1]: pmixp_server.c:923: Unable to pmixp_state_coll_get()

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_check: nid001438 [0]: pmixp_coll_ring.c:614: 0x15005c005dc0: unexpected contrib from nid001439:1, expected is 0

slurmstepd: error:  mpi/pmix_v4: _process_server_request: nid001438 [0]: pmixp_server.c:937: 0x15005c005dc0: unexpected contrib from nid001439:1, coll->seq=0, seq=0

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_reset_if_to: nid001438 [0]: pmixp_coll_ring.c:738: 0x1500580532f0: collective timeout seq=0

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_log: nid001438 [0]: pmixp_coll.c:286: Dumping collective state

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001438 [0]: pmixp_coll_ring.c:756: 0x1500580532f0: COLL_FENCE_RING state seq=0

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001438 [0]: pmixp_coll_ring.c:758: my peerid: 0:nid001438

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001438 [0]: pmixp_coll_ring.c:765: neighbor id: next 1:nid001439, prev 1:nid001439

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001438 [0]: pmixp_coll_ring.c:775: Context ptr=0x150058053368, #0, in-use=0

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001438 [0]: pmixp_coll_ring.c:775: Context ptr=0x1500580533a0, #1, in-use=0

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001438 [0]: pmixp_coll_ring.c:775: Context ptr=0x1500580533d8, #2, in-use=1

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001438 [0]: pmixp_coll_ring.c:786:  seq=0 contribs: loc=1/prev=0/fwd=1

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001438 [0]: pmixp_coll_ring.c:788:  neighbor contribs [2]:

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001438 [0]: pmixp_coll_ring.c:821:  done contrib: -

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001438 [0]: pmixp_coll_ring.c:823:  wait contrib: nid001439

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001438 [0]: pmixp_coll_ring.c:825:  status=PMIXP_COLL_RING_PROGRESS

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001438 [0]: pmixp_coll_ring.c:829:  buf (offset/size): 36/16384

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_reset_if_to: nid001439 [1]: pmixp_coll_ring.c:738: 0x151d0c053400: collective timeout seq=0

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_log: nid001439 [1]: pmixp_coll.c:286: Dumping collective state

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001439 [1]: pmixp_coll_ring.c:756: 0x151d0c053400: COLL_FENCE_RING state seq=0

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001439 [1]: pmixp_coll_ring.c:758: my peerid: 1:nid001439

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001439 [1]: pmixp_coll_ring.c:765: neighbor id: next 0:nid001438, prev 0:nid001438

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001439 [1]: pmixp_coll_ring.c:775: Context ptr=0x151d0c053478, #0, in-use=0

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001439 [1]: pmixp_coll_ring.c:775: Context ptr=0x151d0c0534b0, #1, in-use=0

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001439 [1]: pmixp_coll_ring.c:775: Context ptr=0x151d0c0534e8, #2, in-use=1

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001439 [1]: pmixp_coll_ring.c:786:  seq=0 contribs: loc=1/prev=0/fwd=1

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001439 [1]: pmixp_coll_ring.c:788:  neighbor contribs [2]:

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001439 [1]: pmixp_coll_ring.c:821:  done contrib: -

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001439 [1]: pmixp_coll_ring.c:823:  wait contrib: nid001438

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001439 [1]: pmixp_coll_ring.c:825:  status=PMIXP_COLL_RING_PROGRESS

slurmstepd: error:  mpi/pmix_v4: pmixp_coll_ring_log: nid001439 [1]: pmixp_coll_ring.c:829:  buf (offset/size): 36/16384

# OSU MPI Latency Test v5.8

# Size          Latency (us)

0                       1.66

1                       9.29

2                       9.57

4                       9.69

8                       9.76

16                      9.77

32                      9.76

64                      9.77

128                    10.32

256                     7.54

512                     7.45

1024                    7.38

2048                    7.37

4096                    7.45

8192                    9.21

16384                   9.70

32768                  10.63

65536                  13.15

131072                 16.96

262144                 23.84

524288                 36.16

1048576                60.36

2097152               108.43

4194304               228.31

Note the slurm behavior is - I launch the job.  Go get coffee, do some duo-lingo, read some emails, then after about 10 minutes the osu latency runs.

I did not get the slurm problems using an older mpich 4.3.1 but did get the same performance issue.  9 usecs doesn't seem right for an 8-byte pingpong over libfabric S11.  I was expecting more like 1.6 or so.

I am confident the slurm issue is unrelated to the latency issue.

Thanks for any suggestions on how to address either issue however.