Skip to content

Fix scheduler bug where it rounds to 0 divisions for a chunk. - #6760

Merged
sklam merged 4 commits into
numba:masterfrom
IntelLabs:issue6749
Mar 2, 2021
Merged

Fix scheduler bug where it rounds to 0 divisions for a chunk.#6760
sklam merged 4 commits into
numba:masterfrom
IntelLabs:issue6749

Conversation

@DrTodd13

@DrTodd13 DrTodd13 commented Feb 25, 2021

Copy link
Copy Markdown
Contributor

Resolves #6749

stuartarchibald
stuartarchibald reviewed Feb 25, 2021
View reviewed changes
Comment thread numba/np/ufunc/gufunc_scheduler.cpp Outdated
@DrTodd13 @stuartarchibald
Update numba/np/ufunc/gufunc_scheduler.cpp
7ec4753 
Co-authored-by: stuartarchibald <stuartarchibald@users.noreply.github.com>
stuartarchibald
stuartarchibald reviewed Mar 1, 2021
View reviewed changes
Comment thread numba/tests/test_parfors.py Outdated
@stuartarchibald stuartarchibald added Effort - short Short size effort needed 4 - Waiting on author Waiting for author to respond to review labels Mar 1, 2021
@stuartarchibald stuartarchibald added this to the Numba 0.54 RC milestone Mar 1, 2021
@stuartarchibald

stuartarchibald commented Mar 1, 2021

Copy link
Copy Markdown
Contributor

Manually testing against the sample code in #6749... mainline does this:

num_dim = 6
ranges = ([0, 1], [0, 1], [0, 1], [0, 1], [0, 1], [0, 14], )
num_threads = 4
sched[0] = 1 0 1 0 1 0 1 0 -787390336 271947308 -463249520 271765979 
sched[1] = -769903936 -167047680 307827184 -463126000 0 316800560 -332696768 -787390336 317704688 272245250 0 317704688 
sched[2] = 0 0 -332320592 -770377632 -787390336 316800560 307827184 318603472 -787390336 272247143 -787390336 -463126000 
sched[3] = -770377632 318604000 -812855408 272305680 307819792 315285840 307819864 318603832 0 271783384 32 1241058768

this patch does:

num_dim = 6
ranges = ([0, 1], [0, 1], [0, 1], [0, 1], [0, 1], [0, 14], )
num_threads = 4
sched[0] = 0 0 0 0 0 0 1 1 1 1 0 6
sched[1] = 0 0 0 0 1 0 1 1 1 1 1 6
sched[2] = 0 0 0 0 0 7 1 1 1 1 0 14
sched[3] = 0 0 0 0 1 7 1 1 1 1 1 14

which I think confirms it fixes the issue.

@DrTodd13 @stuartarchibald
Update numba/tests/test_parfors.py
03101ac 
Co-authored-by: stuartarchibald <stuartarchibald@users.noreply.github.com>
@DrTodd13 DrTodd13 added 4 - Waiting on reviewer Waiting for reviewer to respond to author and removed 4 - Waiting on author Waiting for author to respond to review labels Mar 1, 2021
stuartarchibald
stuartarchibald approved these changes Mar 2, 2021
View reviewed changes

@stuartarchibald stuartarchibald left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for the patch!

@stuartarchibald stuartarchibald added 4 - Waiting on CI Review etc done, waiting for CI to finish 5 - Ready to merge Review and testing done, is ready to merge and removed 4 - Waiting on reviewer Waiting for reviewer to respond to author 4 - Waiting on CI Review etc done, waiting for CI to finish labels Mar 2, 2021
@sklam
sklam merged commit 00c15b7 into numba:master Mar 2, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@stuartarchibald stuartarchibald stuartarchibald approved these changes
@esc esc Awaiting requested review from esc
@sklam sklam Awaiting requested review from sklam

Assignees

No one assigned

Labels

5 - Ready to merge Review and testing done, is ready to merge Effort - short Short size effort needed

Projects

None yet

Milestone

Numba 0.54 RC

Development

Successfully merging this pull request may close these issues.

Parallel accelerator yields incorrect results for high dimensional arrays

3 participants

@DrTodd13 @stuartarchibald @sklam