Closed Sudha247 closed 1 year ago
Okay, good news is the CI is happy - I've marked it as ready for review. Question (to @punchagan or @shakthimaan): does the parallel runs include graph500 benchmarks? I saw some failures on it in my local setup with this PR.
Could you share the failures you see?
From the latest run in Navajo, we run graph500_gen
benchmarks
The log says
$ (cd _build/5.2.0+trunk_1/benchmarks/graph500seq && ./gen.exe -scale 21 -edgefactor 16 edges.data)
> Generating edge list...
[got signal KILL]
$ (cd _build/5.2.0+trunk_1/benchmarks/graph500par && ./gen.exe -scale 21 -edgefactor 16 -ndomains 16 edges.data)
> Generating edge list...
> Generated. Time: 173.464429 s.
Maybe the benchmark's just too big for my machine. It might help to do a run on navajo or another larger machine, but otherwise I don't have any other changes to this PR.
It might help to do a run on navajo or another larger machine
The changes look good to me. I'll do a test run on navajo
and merge the PR. Thanks for the PR, @Sudha247!
Okay, good news is the CI is happy - I've marked it as ready for review. Question (to @punchagan or @shakthimaan): does the parallel runs include graph500 benchmarks? I saw some failures on it in my local setup with this PR.