Uh oh!
There was an error while loading. Please reload this page.
Revert "[SPARK-30642][SPARK-30659][SPARK-30660][SPARK-30662]" - #27487
Revert "[SPARK-30642][SPARK-30659][SPARK-30660][SPARK-30662]"#27487zhengruifeng wants to merge 4 commits into
Conversation
…ctors" This reverts commit d0c3e9f.
dongjoon-hyun
commented
Feb 7, 2020
Wow. It's a massive revert. If you have a discussion on the other PRs, could you add some pointers into the description? What is the performance regression?
|
zhengruifeng
commented
Feb 7, 2020
@dongjoon-hyun There is some test and discussion in #27374, I also update the description. |
dongjoon-hyun
commented
Feb 7, 2020
Thank you, @zhengruifeng . |
dongjoon-hyun
commented
Feb 7, 2020
Also, cc @huaxingao since she is also the author of the last PR. |
SparkQA
commented
Feb 7, 2020
Test build #118014 has finished for PR 27487 at commit
|
This reverts commit f59685a.
SparkQA
commented
Feb 7, 2020
Test build #118016 has finished for PR 27487 at commit
|
…vectors" This reverts commit 073ce12.
SparkQA
commented
Feb 7, 2020
Test build #118020 has finished for PR 27487 at commit
|
zhengruifeng
commented
Feb 7, 2020
retest this please |
SparkQA
commented
Feb 7, 2020
Test build #118022 has finished for PR 27487 at commit
|
This reverts commit 96d2727.
SparkQA
commented
Feb 7, 2020
Test build #118029 has finished for PR 27487 at commit
|
zhengruifeng
commented
Feb 7, 2020
zhengruifeng
commented
Feb 8, 2020
Merged to master |
srowen
commented
Feb 8, 2020
Thank you. I should have realized that sparse vectors could be an issue in the first place. That's my bad. |
zhengruifeng
commented
Feb 8, 2020
@srowen Never mind, I should have maken more perfermance tests. Maybe we should do such refactors only when we have enough time for testing. When working on these PRs, it maybe too close to the code freeze. |
viirya
commented
Feb 25, 2020
Is this to revert the PRs on branch-3.0 too? I cannot find it in branch-3.0 commits. |
### What changes were proposed in this pull request? Revert #27360#27396#27374#27389 ### Why are the changes needed? BLAS need more performace tests, specially on sparse datasets. Perfermance test of LogisticRegression (#27374) on sparse dataset shows that blockify vectors to matrices and use BLAS will cause performance regression. LinearSVC and LinearRegression were also updated in the same way as LogisticRegression, so we need to revert them to make sure no regression. ### Does this PR introduce any user-facing change? remove newly added param blockSize ### How was this patch tested? reverted testsuites Closes#27487 from zhengruifeng/revert_blockify_ii. Authored-by: zhengruifeng <ruifengz@foxmail.com> Signed-off-by: zhengruifeng <ruifengz@foxmail.com>
zhengruifeng
commented
Feb 25, 2020
@viirya Yes, this is to revert on branch-3.0 too. I am go to revert it on 3.0. Thanks for pointing it out! |
### What changes were proposed in this pull request? Revert apache#27360apache#27396apache#27374apache#27389 ### Why are the changes needed? BLAS need more performace tests, specially on sparse datasets. Perfermance test of LogisticRegression (apache#27374) on sparse dataset shows that blockify vectors to matrices and use BLAS will cause performance regression. LinearSVC and LinearRegression were also updated in the same way as LogisticRegression, so we need to revert them to make sure no regression. ### Does this PR introduce any user-facing change? remove newly added param blockSize ### How was this patch tested? reverted testsuites Closesapache#27487 from zhengruifeng/revert_blockify_ii. Authored-by: zhengruifeng <ruifengz@foxmail.com> Signed-off-by: zhengruifeng <ruifengz@foxmail.com>
What changes were proposed in this pull request?
Revert
#27360
#27396
#27374
#27389
Why are the changes needed?
BLAS need more performace tests, specially on sparse datasets.
Perfermance test of LogisticRegression (#27374) on sparse dataset shows that blockify vectors to matrices and use BLAS will cause performance regression.
LinearSVC and LinearRegression were also updated in the same way as LogisticRegression, so we need to revert them to make sure no regression.
Does this PR introduce any user-facing change?
remove newly added param blockSize
How was this patch tested?
reverted testsuites