-
Notifications
You must be signed in to change notification settings - Fork 9
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
1830: Remove concept of under/overloaded from TemperedWMin
#1860
base: develop
Are you sure you want to change the base?
Conversation
Pipelines resultsPR tests (gcc-6, ubuntu, mpich) Build for 6121bc8
PR tests (clang-3.9, ubuntu, mpich) Build for 6121bc8
PR tests (nvidia cuda 10.1, ubuntu, mpich) Build for 3c8faeb
PR tests (gcc-10, ubuntu, openmpi, no LB) Build for 6121bc8
PR tests (gcc-9, ubuntu, mpich, zoltan) Build for 6121bc8
PR tests (clang-5.0, ubuntu, mpich) Build for 6121bc8
PR tests (nvidia cuda 11.0, ubuntu, mpich) Build for 6121bc8
PR tests (gcc-7, ubuntu, mpich, trace runtime, LB) Build for 6121bc8
PR tests (clang-9, ubuntu, mpich) Build for 6121bc8
PR tests (clang-13, alpine, mpich) Build for 6121bc8
PR tests (intel icpx, ubuntu, mpich) Build for 6121bc8
PR tests (gcc-8, ubuntu, mpich, address sanitizer) Build for 6121bc8
PR tests (clang-12, ubuntu, mpich) Build for 6121bc8
PR tests (clang-11, ubuntu, mpich) Build for 6121bc8
PR tests (clang-13, ubuntu, mpich) Build for 3c8faeb
PR tests (gcc-11, ubuntu, mpich) Build for 6121bc8
PR tests (clang-14, ubuntu, mpich) Build for 6121bc8
PR tests (gcc-12, ubuntu, mpich) Build for cf8804f (2023-02-28 21:41:36 UTC)
PR tests (intel icpc, ubuntu, mpich) Build for 6121bc8
PR tests (clang-10, ubuntu, mpich) Build for 6121bc8
PR tests (gcc-5, ubuntu, mpich) Build for 6121bc8
|
Codecov Report
@@ Coverage Diff @@
## develop #1860 +/- ##
===========================================
- Coverage 84.90% 84.37% -0.54%
===========================================
Files 721 760 +39
Lines 25737 26811 +1074
===========================================
+ Hits 21852 22621 +769
- Misses 3885 4190 +305
|
6a85c17
to
d863bd3
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Within propagateIncomingSync
and propagateIncomingAsync
there are conditionals on isUnderloaded()
that I think should be (effectively) unconditional for TemperedWMin
.
You might consider renaming underloaded_
to potential_recipients_
or something like that.
There are configurations possible for TemperedLB
that are not valid for TemperedWMin
, such as NormByMaxExcludeIneligible
for the CMF. We'll need to validate the options at some point but, for now, know that makeSufficientlyUnderloaded
should never be used with TemperedWMin
.
3c8faeb
to
2ae50dc
Compare
f7b4908
to
f363c13
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Your changes to the algorithms to remove the concept of under/overloaded look good. When doing any testing, be sure to use the default CMF NormByMax
because the others will not be valid for TemperedWMin
.
A good test might be that, with knowledge set to Complete
, informSync
and informAsync
correctly gain knowledge of loads for either all ranks or all underloaded ranks depending if TemperedWMin
or TemperedLB
is in use. I haven't thought through how to set up such a test yet.
f363c13
to
6121bc8
Compare
6121bc8
to
8ca421f
Compare
742d140
to
5f41170
Compare
5f41170
to
7573b94
Compare
89ea20e
to
e2777a3
Compare
e2777a3
to
23608d2
Compare
- extract helper methods and add overloads for TemperedWMin - remove redundant code - remove unused member variable
23608d2
to
cf8804f
Compare
fixes #1830