Env conf debug #1534
Env conf debug #1534
Conversation
Nikhil-Nunna
requested review from
wenkaidu,
gilbertlee-amd,
akolliasAMD,
PedramAlizadeh,
nusislam,
nileshnegi,
KawtharShafie,
AtlantaPepsi,
mberenjk,
corey-derochie-amd,
mustafabar,
thananon,
JhaShweta1,
BertanDogancay,
rahulvaidya20,
jee7s,
isaki001,
PJAvinash,
AbandiGa and
haripriya-amd
as code owners
|
|
||
| local output_file="rccl_tests_output.txt" | ||
|
|
||
| $1/build/all_reduce_perf -b 8 -e 16M -f 8 -g 2 > "${output_file}" |
There was a problem hiding this comment.
If we are only running to get the version info, you can just run from -b 8 -e 8 -g 1, no need to do any kind of sweep.
|
|
||
| ##################################################### README ##################################################### | ||
|
|
||
| # This script only requires 1 input arguement, it is the path to the rccl-tests repo. EX: /path/to/rccl-tests/ |
There was a problem hiding this comment.
| # This script only requires 1 input arguement, it is the path to the rccl-tests repo. EX: /path/to/rccl-tests/ | |
| # This script only requires 1 input argument, it is the path to the rccl-tests repo. EX: /path/to/rccl-tests/ |
| # 10. ibv_devinfo is on path | ||
| # 11. ibstat device GUIDs | ||
|
|
||
| # All output will be in a folder called conf-script-output that will be created in the same directory as the script |
There was a problem hiding this comment.
I'd suggest making this path user configurable, including allowing it to output to stdout. Perhaps default to cwd.
| echo "An error occurred during ${func_call}" | ||
| echo "${error_message}" | ||
| echo "in step ${func_name}" |
There was a problem hiding this comment.
Only successful output is captured to log files, which may not be desirable. These should probably tee to the conf-script-output location, or to stderr if no output location provided.
|
|
||
| ##################################################### setup output folder ##################################################### | ||
|
|
||
| mkdir conf-script-output |
There was a problem hiding this comment.
| mkdir conf-script-output | |
| mkdir -p conf-script-output |
|
|
||
| ##################################################### end define necessary functions ##################################################### | ||
|
|
||
|
|
There was a problem hiding this comment.
Since this script is meant to be used by customers, and has a required argument, it should probably display help text. Probably also recognize --help.
|
Should there be README or similar documentation to steer users toward this, or should we rely on support to instruct them? |
|
Should we call this script in our CI to capture our test configurations for debugging? |
Details
Do not mention proprietary info or link to internal work items in this PR.
Work item: Internal
What were the changes?
Developed Customer Support script that gathers customer system configuration.
Why were the changes made?
Script was made per request of the RCCL team to aid RCCL setup and debug.
How was the outcome achieved?
Technical details behind the work. Explain any publicly-available hardware peculiarities.
Additional Documentation:
What else should the reviewer know?
Approval Checklist
Do not approve until these items are satisfied.