Skip to content

add channel last 3d support for maxpool3d on CPU#97775

Closed
CaoE wants to merge 28 commits intogh/CaoE/9/basefrom
gh/CaoE/9/head
Closed

add channel last 3d support for maxpool3d on CPU#97775
CaoE wants to merge 28 commits intogh/CaoE/9/basefrom
gh/CaoE/9/head

Conversation

@CaoE
Copy link
Collaborator

@CaoE CaoE commented Mar 28, 2023 •
edited
Loading

Stack from ghstack (oldest at bottom):

Testing

Single socket (28 cores):

shape fp32 forward / ms bf16 forward / ms fp32 backward / ms bf16 backward / ms
size: (1, 56, 264, 264), kernel: 3, stride: 1, mem_format: contig 3.959584 5.493402 0.557232 0.568485
size: (1, 56, 264, 264), kernel: 3, stride: 1, mem_format: CL 0.815511 1.351261 5.710506 10.57506
size: (32, 32, 100, 100), kernel: 3, stride: 1, mem_format: contig 10.63426 15.28637 2.67656 1.71365
size: (32, 32, 100, 100), kernel: 3, stride: 1, mem_format: CL 2.63570 2.05532 2.55452 2.33923
size: (4, 19, 10, 16, 16), kernel: 3, stride: 1, mem_format: contig 0.375469 0.479748 0.066364 0.065155
size: (4, 19, 10, 16, 16), kernel: 3, stride: 1, mem_format: CL3d 0.112197 0.112326 0.111697 0.145364

Single core:

shape fp32 forward / ms bf16 forward / ms fp32 backward / ms bf16 backward / ms
size: (1, 56, 264, 264), kernel: 3, stride: 1, mem_format: contig 92.16582 128.6513 6.684325 12.21541
size: (1, 56, 264, 264), kernel: 3, stride: 1, mem_format: CL 10.14318 29.80297 7.350142 11.25323
size: (32, 32, 100, 100), kernel: 3, stride: 1, mem_format: contig 238.55453 331.89967 19.694657 32.78853
size: (32, 32, 100, 100), kernel: 3, stride: 1, mem_format: CL 30.17079 32.75628 22.44543 30.17796
size: (4, 19, 10, 16, 16), kernel: 3, stride: 1, mem_format: contig 7.474389 9.937217 0.236015 0.434229
size: (4, 19, 10, 16, 16), kernel: 3, stride: 1, mem_format: CL3d 2.318954 2.469444 0.262125 0.401361

cc @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10

@CaoE CaoE mentioned this pull request Mar 28, 2023
Closed
@pytorch-bot
Copy link

pytorch-bot bot commented Mar 28, 2023 •
edited
Loading

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/97775

Note: Links to docs will display an error until the docs builds have been completed.

❌ 2 New Failures, 1 Unrelated Failure

As of commit b62a746 with merge base c68d0a7 (image):

NEW FAILURES - The following jobs have failed:

UNSTABLE - The following job failed but was likely due to flakiness present on trunk and has been marked as unstable:

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@pytorch-bot pytorch-bot bot added the release notes: nn release notes category label Mar 28, 2023
@github-actions github-actions bot added the module: cpu CPU specific problem (e.g., perf, algorithm) label Mar 28, 2023
@CaoE CaoE marked this pull request as draft March 28, 2023 13:48
CaoE added a commit that referenced this pull request Mar 28, 2023
@CaoE
add channel last 3d support for maxpool3d (#55)
e02a3f5 
ghstack-source-id: 5ee6dce
Pull Request resolved: #97775
mingfeima
mingfeima requested changes Mar 29, 2023
View reviewed changes
@CaoE
Update on "add channel last 3d support for maxpool3d (#55)"
bb4ab47 
cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10

[ghstack-poisoned]
@CaoE
Update on "add channel last 3d support for maxpool3d (#55)"
200de10 
cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10

[ghstack-poisoned]
@CaoE CaoE changed the title add channel last 3d support for maxpool3d (#55) add channel last 3d support for maxpool3d Mar 29, 2023
@CaoE
Update on "add channel last 3d support for maxpool3d"
c978563 
cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10

[ghstack-poisoned]
CaoE added a commit that referenced this pull request Mar 29, 2023
@CaoE
add channel last 3d support for maxpool3d
1ec2548 
ghstack-source-id: 4e2c476
Pull Request resolved: #97775
@CaoE
Update on "add channel last 3d support for maxpool3d"
7655993 
cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10

[ghstack-poisoned]
CaoE added a commit that referenced this pull request Mar 30, 2023
@CaoE
add channel last 3d support for maxpool3d
6c87f4f 
ghstack-source-id: 28f26c7
Pull Request resolved: #97775
@CaoE
Update on "add channel last 3d support for maxpool3d"
e9cea99 
cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10

[ghstack-poisoned]
CaoE added a commit that referenced this pull request Mar 30, 2023
@CaoE
add channel last 3d support for maxpool3d
7fc59ce 
ghstack-source-id: 48db671
Pull Request resolved: #97775
@CaoE
Update on "add channel last 3d support for maxpool3d"
7d39d4e 
cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10

[ghstack-poisoned]
CaoE added a commit that referenced this pull request Apr 5, 2023
@CaoE
add channel last 3d support for maxpool3d
20beda9 
ghstack-source-id: 1842703
Pull Request resolved: #97775
@CaoE CaoE requested a review from mingfeima April 7, 2023 02:12
@CaoE CaoE added the ciflow/trunk Trigger trunk jobs on your pull request label Apr 7, 2023
@CaoE
Update on "add channel last 3d support for maxpool3d"
ba16df8 
cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10

[ghstack-poisoned]
@CaoE CaoE mentioned this pull request Apr 11, 2023
Closed
@CaoE CaoE changed the title add channel last 3d support for maxpool3d add channel last 3d support for maxpool3d on CPU Apr 11, 2023
@CaoE
Update on "add channel last 3d support for maxpool3d on CPU"
496dcd1 
cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10

[ghstack-poisoned]
@CaoE CaoE requested a review from jgong5 April 13, 2023 06:39
@CaoE
Update on "add channel last 3d support for maxpool3d on CPU"
bd73046 
cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10

[ghstack-poisoned]
@CaoE CaoE marked this pull request as ready for review July 12, 2023 08:51
@CaoE
Update on "add channel last 3d support for maxpool3d on CPU"
b41848e 

### Testing
Single socket (28 cores):

shape | fp32 forward / ms | bf16 forward / ms | fp32 backward / ms  | bf16 backward / ms
-- | -- | -- | -- | --
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: contig | 3.959584 | 5.493402 | 0.557232 | 0.568485
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: CL | 0.815511 | 1.351261 | 5.710506 | 10.57506
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: contig  | 10.63426 | 15.28637 | 2.67656 | 1.71365
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: CL | 2.63570 | 2.05532 | 2.55452 | 2.33923
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: contig | 0.375469 | 0.479748 | 0.066364 | 0.065155
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: CL3d | 0.112197 | 0.112326 | 0.111697 | 0.145364



Single core:


shape | fp32 forward / ms | bf16 forward / ms | fp32 backward / ms | bf16 backward / ms
-- | -- | -- | -- | --
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: contig | 92.16582 | 128.6513 | 6.684325 | 12.21541
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: CL | 10.14318 | 29.80297 | 7.350142 | 11.25323
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: contig | 238.55453 | 331.89967 | 19.694657 | 32.78853
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: CL | 30.17079 | 32.75628 | 22.44543 | 30.17796
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: contig | 7.474389 | 9.937217 | 0.236015 | 0.434229
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: CL3d | 2.318954 | 2.469444 | 0.262125 | 0.401361

cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10

[ghstack-poisoned]
@CaoE CaoE mentioned this pull request Jul 24, 2023
Closed
CaoE added 2 commits July 26, 2023 19:34
@CaoE
Update on "add channel last 3d support for maxpool3d on CPU"
b703245 

### Testing
Single socket (28 cores):

shape | fp32 forward / ms | bf16 forward / ms | fp32 backward / ms  | bf16 backward / ms
-- | -- | -- | -- | --
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: contig | 3.959584 | 5.493402 | 0.557232 | 0.568485
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: CL | 0.815511 | 1.351261 | 5.710506 | 10.57506
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: contig  | 10.63426 | 15.28637 | 2.67656 | 1.71365
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: CL | 2.63570 | 2.05532 | 2.55452 | 2.33923
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: contig | 0.375469 | 0.479748 | 0.066364 | 0.065155
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: CL3d | 0.112197 | 0.112326 | 0.111697 | 0.145364



Single core:


shape | fp32 forward / ms | bf16 forward / ms | fp32 backward / ms | bf16 backward / ms
-- | -- | -- | -- | --
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: contig | 92.16582 | 128.6513 | 6.684325 | 12.21541
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: CL | 10.14318 | 29.80297 | 7.350142 | 11.25323
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: contig | 238.55453 | 331.89967 | 19.694657 | 32.78853
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: CL | 30.17079 | 32.75628 | 22.44543 | 30.17796
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: contig | 7.474389 | 9.937217 | 0.236015 | 0.434229
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: CL3d | 2.318954 | 2.469444 | 0.262125 | 0.401361

cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10

[ghstack-poisoned]
@CaoE
Update on "add channel last 3d support for maxpool3d on CPU"
21e063f 

### Testing
Single socket (28 cores):

shape | fp32 forward / ms | bf16 forward / ms | fp32 backward / ms  | bf16 backward / ms
-- | -- | -- | -- | --
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: contig | 3.959584 | 5.493402 | 0.557232 | 0.568485
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: CL | 0.815511 | 1.351261 | 5.710506 | 10.57506
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: contig  | 10.63426 | 15.28637 | 2.67656 | 1.71365
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: CL | 2.63570 | 2.05532 | 2.55452 | 2.33923
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: contig | 0.375469 | 0.479748 | 0.066364 | 0.065155
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: CL3d | 0.112197 | 0.112326 | 0.111697 | 0.145364



Single core:


shape | fp32 forward / ms | bf16 forward / ms | fp32 backward / ms | bf16 backward / ms
-- | -- | -- | -- | --
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: contig | 92.16582 | 128.6513 | 6.684325 | 12.21541
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: CL | 10.14318 | 29.80297 | 7.350142 | 11.25323
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: contig | 238.55453 | 331.89967 | 19.694657 | 32.78853
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: CL | 30.17079 | 32.75628 | 22.44543 | 30.17796
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: contig | 7.474389 | 9.937217 | 0.236015 | 0.434229
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: CL3d | 2.318954 | 2.469444 | 0.262125 | 0.401361

cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10

[ghstack-poisoned]
@CaoE
Copy link
Collaborator Author

CaoE commented Aug 11, 2023

@mikaylagawarecki Could you please review this PR ? Thank you.

@CaoE CaoE requested a review from mikaylagawarecki August 11, 2023 01:23
@mikaylagawarecki
Copy link
Contributor

mikaylagawarecki commented Aug 11, 2023

@CaoE Sure, will take a look later today!

mikaylagawarecki
mikaylagawarecki reviewed Aug 11, 2023
View reviewed changes
Copy link
Contributor

@mikaylagawarecki mikaylagawarecki left a comment •
edited
Loading

Choose a reason for hiding this comment

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

Hey @CaoE as a broader comment

Following up from our previous discussion, is #106104 ready for review? Happy to review it if it is (and would prefer that we use that testing before landing the rest of these PRs :)

Separately, we must be wary that the test_memory_format tests for ModuleInfos has been skipped rather than xfailed for quite a few ops, as we add channels_last_3d support for more of them, could we unskip these tests as well please e.g. https://github.com/pytorch/pytorch/blob/main/torch/testing/_internal/common_modules.py#L2555

@mikaylagawarecki mikaylagawarecki mentioned this pull request Aug 17, 2023
Closed
@CaoE
Update on "add channel last 3d support for maxpool3d on CPU"
cf9943d 

### Testing
Single socket (28 cores):

shape | fp32 forward / ms | bf16 forward / ms | fp32 backward / ms  | bf16 backward / ms
-- | -- | -- | -- | --
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: contig | 3.959584 | 5.493402 | 0.557232 | 0.568485
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: CL | 0.815511 | 1.351261 | 5.710506 | 10.57506
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: contig  | 10.63426 | 15.28637 | 2.67656 | 1.71365
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: CL | 2.63570 | 2.05532 | 2.55452 | 2.33923
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: contig | 0.375469 | 0.479748 | 0.066364 | 0.065155
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: CL3d | 0.112197 | 0.112326 | 0.111697 | 0.145364



Single core:


shape | fp32 forward / ms | bf16 forward / ms | fp32 backward / ms | bf16 backward / ms
-- | -- | -- | -- | --
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: contig | 92.16582 | 128.6513 | 6.684325 | 12.21541
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: CL | 10.14318 | 29.80297 | 7.350142 | 11.25323
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: contig | 238.55453 | 331.89967 | 19.694657 | 32.78853
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: CL | 30.17079 | 32.75628 | 22.44543 | 30.17796
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: contig | 7.474389 | 9.937217 | 0.236015 | 0.434229
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: CL3d | 2.318954 | 2.469444 | 0.262125 | 0.401361

cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10

[ghstack-poisoned]
@CaoE CaoE requested a review from wschin as a code owner August 18, 2023 05:55
@CaoE
Copy link
Collaborator Author

CaoE commented Aug 25, 2023

@mikaylagawarecki Could you please review this PR ? #106104 is ready for review. Thank you.

@CaoE
Update on "add channel last 3d support for maxpool3d on CPU"
7471bc4 

### Testing
Single socket (28 cores):

shape | fp32 forward / ms | bf16 forward / ms | fp32 backward / ms  | bf16 backward / ms
-- | -- | -- | -- | --
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: contig | 3.959584 | 5.493402 | 0.557232 | 0.568485
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: CL | 0.815511 | 1.351261 | 5.710506 | 10.57506
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: contig  | 10.63426 | 15.28637 | 2.67656 | 1.71365
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: CL | 2.63570 | 2.05532 | 2.55452 | 2.33923
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: contig | 0.375469 | 0.479748 | 0.066364 | 0.065155
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: CL3d | 0.112197 | 0.112326 | 0.111697 | 0.145364



Single core:


shape | fp32 forward / ms | bf16 forward / ms | fp32 backward / ms | bf16 backward / ms
-- | -- | -- | -- | --
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: contig | 92.16582 | 128.6513 | 6.684325 | 12.21541
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: CL | 10.14318 | 29.80297 | 7.350142 | 11.25323
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: contig | 238.55453 | 331.89967 | 19.694657 | 32.78853
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: CL | 30.17079 | 32.75628 | 22.44543 | 30.17796
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: contig | 7.474389 | 9.937217 | 0.236015 | 0.434229
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: CL3d | 2.318954 | 2.469444 | 0.262125 | 0.401361

cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10

[ghstack-poisoned]
@CaoE
Copy link
Collaborator Author

CaoE commented Aug 25, 2023 •
edited
Loading

@mikaylagawarecki Can we merge this PR before 2.1 branch cut if possible ?

@CaoE
Update on "add channel last 3d support for maxpool3d on CPU"
b62a746 

### Testing
Single socket (28 cores):

shape | fp32 forward / ms | bf16 forward / ms | fp32 backward / ms  | bf16 backward / ms
-- | -- | -- | -- | --
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: contig | 3.959584 | 5.493402 | 0.557232 | 0.568485
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: CL | 0.815511 | 1.351261 | 5.710506 | 10.57506
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: contig  | 10.63426 | 15.28637 | 2.67656 | 1.71365
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: CL | 2.63570 | 2.05532 | 2.55452 | 2.33923
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: contig | 0.375469 | 0.479748 | 0.066364 | 0.065155
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: CL3d | 0.112197 | 0.112326 | 0.111697 | 0.145364



Single core:


shape | fp32 forward / ms | bf16 forward / ms | fp32 backward / ms | bf16 backward / ms
-- | -- | -- | -- | --
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: contig | 92.16582 | 128.6513 | 6.684325 | 12.21541
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: CL | 10.14318 | 29.80297 | 7.350142 | 11.25323
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: contig | 238.55453 | 331.89967 | 19.694657 | 32.78853
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: CL | 30.17079 | 32.75628 | 22.44543 | 30.17796
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: contig | 7.474389 | 9.937217 | 0.236015 | 0.434229
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: CL3d | 2.318954 | 2.469444 | 0.262125 | 0.401361

cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10

[ghstack-poisoned]
mikaylagawarecki
mikaylagawarecki approved these changes Aug 25, 2023
View reviewed changes
Copy link
Contributor

@mikaylagawarecki mikaylagawarecki left a comment

Choose a reason for hiding this comment

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

Thanks for using the ModuleInfo testing

@mikaylagawarecki
Copy link
Contributor

mikaylagawarecki commented Aug 26, 2023

@pytorchbot merge -f "macos failures are unrelated"

@pytorchmergebot
Copy link
Collaborator

pytorchmergebot commented Aug 26, 2023

Merge started

Your change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Please use -f as last resort and instead consider -i/--ignore-current to continue the merge ignoring current failures. This will allow currently pending tests to finish and report signal before the merge.

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

Advanced Debugging
Check the merge workflow status
here

voznesenskym pushed a commit that referenced this pull request Aug 27, 2023
@CaoE @voznesenskym
add channel last 3d support for maxpool3d on CPU (#97775)
aa327e8 
### Testing
Single socket (28 cores):

shape | fp32 forward / ms | bf16 forward / ms | fp32 backward / ms  | bf16 backward / ms
-- | -- | -- | -- | --
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: contig | 3.959584 | 5.493402 | 0.557232 | 0.568485
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: CL | 0.815511 | 1.351261 | 5.710506 | 10.57506
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: contig  | 10.63426 | 15.28637 | 2.67656 | 1.71365
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: CL | 2.63570 | 2.05532 | 2.55452 | 2.33923
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: contig | 0.375469 | 0.479748 | 0.066364 | 0.065155
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: CL3d | 0.112197 | 0.112326 | 0.111697 | 0.145364

Single core:

shape | fp32 forward / ms | bf16 forward / ms | fp32 backward / ms | bf16 backward / ms
-- | -- | -- | -- | --
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: contig | 92.16582 | 128.6513 | 6.684325 | 12.21541
size: (1, 56, 264, 264), kernel: 3,   stride: 1, mem_format: CL | 10.14318 | 29.80297 | 7.350142 | 11.25323
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: contig | 238.55453 | 331.89967 | 19.694657 | 32.78853
size: (32, 32, 100, 100), kernel: 3,   stride: 1, mem_format: CL | 30.17079 | 32.75628 | 22.44543 | 30.17796
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: contig | 7.474389 | 9.937217 | 0.236015 | 0.434229
size: (4, 19, 10, 16, 16), kernel:   3, stride: 1, mem_format: CL3d | 2.318954 | 2.469444 | 0.262125 | 0.401361

Pull Request resolved: #97775
Approved by: https://github.com/jgong5, https://github.com/mikaylagawarecki
@facebook-github-bot facebook-github-bot deleted the gh/CaoE/9/head branch August 29, 2023 14:16
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@jgong5 jgong5 jgong5 approved these changes

@mikaylagawarecki mikaylagawarecki mikaylagawarecki approved these changes

@mingfeima mingfeima Awaiting requested review from mingfeima

@mruberry mruberry Awaiting requested review from mruberry

@ngimel ngimel Awaiting requested review from ngimel

@Skylion007 Skylion007 Awaiting requested review from Skylion007

@BowenBao BowenBao Awaiting requested review from BowenBao

@abock abock Awaiting requested review from abock

@thiagocrepaldi thiagocrepaldi Awaiting requested review from thiagocrepaldi

@wschin wschin Awaiting requested review from wschin

Assignees

No one assigned

Labels

ciflow/periodic Trigger jobs ran periodically on master (periodic.yml) on the PR ciflow/trunk Trigger trunk jobs on your pull request Merged module: cpu CPU specific problem (e.g., perf, algorithm) open source release notes: nn release notes category

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

6 participants

@CaoE @mikaylagawarecki @pytorchmergebot @jgong5 @mingfeima @pytorchbot