Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

TOOLS/PERF: rocm perf modules not available #4433

Closed
paklui opened this issue Nov 12, 2019 · 2 comments
Closed

TOOLS/PERF: rocm perf modules not available #4433

paklui opened this issue Nov 12, 2019 · 2 comments

Comments

@paklui
Copy link
Contributor

paklui commented Nov 12, 2019

The ucx_perftest for rocm cannot run because the perf modules for rocm are not available.

@paklui
Copy link
Contributor Author

paklui commented Nov 12, 2019

The proposed changes in PR #4434 should provide the ucx_perftest for rocm and rocm-managed.

paklui@ixt-hq-50:~/ucx$ 
        HIP_VISIBLE_DEVICES=0 /home/paklui/ucx/install/bin/ucx_perftest -m rocm -t tag_bw &
        HIP_VISIBLE_DEVICES=1 /home/paklui/ucx/install/bin/ucx_perftest -m rocm -t tag_bw -s 4194304 localhost 2>&1 | tee -a $LOG
[1573535643.539054] [ixt-hq-50:18615:0]       perftest.c:1394 UCX  WARN  CPU affinity is not set (bound to 56 cpus). Performance may be impacted.
Waiting for connection...
+------------------------------------------------------------------------------------------+
| API:          protocol layer                                                             |
| Test:         tag match bandwidth                                                        |
| Data layout:  (automatic)                                                                |
| Message size: 4194304                                                                    |
+------------------------------------------------------------------------------------------+
[1573535643.539053] [ixt-hq-50:18616:0]       perftest.c:1394 UCX  WARN  CPU affinity is not set (bound to 56 cpus). Performance may be impacted.
+--------------+-----------------------------+---------------------+-----------------------+
|              |       latency (usec)        |   bandwidth (MB/s)  |  message rate (msg/s) |
+--------------+---------+---------+---------+----------+----------+-----------+-----------+
| # iterations | typical | average | overall |  average |  overall |   average |   overall |
+--------------+---------+---------+---------+----------+----------+-----------+-----------+
          6828   146.402   146.472   146.472   27309.02   27309.02        6827        6827
         13655   146.386   146.478   146.475   27307.80   27308.41        6827        6827
...
        996892   146.375   146.460   146.464   27311.26   27310.42        6828        6828
       1000000   146.369   146.512   146.464   27301.54   27310.39        6825        6828
paklui@ixt-hq-50:~/ucx$

@paklui
Copy link
Contributor Author

paklui commented Dec 21, 2019

It is fixed in #4587

@paklui paklui closed this as completed Dec 21, 2019
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant