Skip to content

Move THCTensor_(lognormal) to ATen #21299

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.

Sign up for GitHub

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

Jump to bottom
Closed
syed-ahmed wants to merge 4 commits into from
Closed

Move THCTensor_(lognormal) to ATen #21299

syed-ahmed wants to merge 4 commits into from

Conversation

@syed-ahmed
Copy link
Collaborator

@syed-ahmed syed-ahmed commented Jun 3, 2019 •
edited by ezyang
Loading

Stack from ghstack:

Resubmit of #20624

Effective Bandwidth Benchmark

Float Type

Before:

log_normal, size, elements 65536 forward 4.84466552734375e-06 bandwidth (GB/s) 54.1098242015748
log_normal, size, elements 131072 forward 5.0425529479980465e-06 bandwidth (GB/s) 103.97273075895983
log_normal, size, elements 262144 forward 7.326602935791016e-06 bandwidth (GB/s) 143.11898832098927
log_normal, size, elements 524288 forward 1.1749267578125e-05 bandwidth (GB/s) 178.49214736623378
log_normal, size, elements 1048576 forward 2.05230712890625e-05 bandwidth (GB/s) 204.37019103643124
log_normal, size, elements 2097152 forward 3.9284229278564456e-05 bandwidth (GB/s) 213.53627534643442
log_normal, size, elements 4194304 forward 7.281541824340821e-05 bandwidth (GB/s) 230.40746595613766
log_normal, size, elements 8388608 forward 0.00013544559478759766 bandwidth (GB/s) 247.7336531514311
log_normal, size, elements 16777216 forward 0.0002670741081237793 bandwidth (GB/s) 251.27431659866272
log_normal, size, elements 33554432 forward 0.0005250406265258789 bandwidth (GB/s) 255.63303336753222

After:

log_normal, size, elements 65536 forward 5.47647476196289e-06 bandwidth (GB/s) 47.86728897588159
log_normal, size, elements 131072 forward 6.859302520751953e-06 bandwidth (GB/s) 76.43459351936045
log_normal, size, elements 262144 forward 7.7056884765625e-06 bandwidth (GB/s) 136.07817175445544
log_normal, size, elements 524288 forward 8.029937744140625e-06 bandwidth (GB/s) 261.1666574289786
log_normal, size, elements 1048576 forward 1.1892318725585938e-05 bandwidth (GB/s) 352.6901773138733
log_normal, size, elements 2097152 forward 1.9683837890625e-05 bandwidth (GB/s) 426.1672975875969
log_normal, size, elements 4194304 forward 3.241539001464844e-05 bandwidth (GB/s) 517.5694629130921
log_normal, size, elements 8388608 forward 5.803346633911133e-05 bandwidth (GB/s) 578.1910700272298
log_normal, size, elements 16777216 forward 0.00011091709136962891 bandwidth (GB/s) 605.0362768381755
log_normal, size, elements 33554432 forward 0.00021491527557373046 bandwidth (GB/s) 624.5146029834174

Double Type

Before:

log_normal, size, elements 65536 forward 5.793571472167969e-06 bandwidth (GB/s) 45.247392089547326
log_normal, size, elements 131072 forward 8.199214935302735e-06 bandwidth (GB/s) 63.943682918057576
log_normal, size, elements 262144 forward 1.3582706451416015e-05 bandwidth (GB/s) 77.19934195373004
log_normal, size, elements 524288 forward 2.3326873779296876e-05 bandwidth (GB/s) 89.90283137988553
log_normal, size, elements 1048576 forward 4.379749298095703e-05 bandwidth (GB/s) 95.76584673062604
log_normal, size, elements 2097152 forward 8.105754852294922e-05 bandwidth (GB/s) 103.48953493979646
log_normal, size, elements 4194304 forward 0.0001421213150024414 bandwidth (GB/s) 118.04855590951854
log_normal, size, elements 8388608 forward 0.00027796506881713865 bandwidth (GB/s) 120.71456367804988
log_normal, size, elements 16777216 forward 0.0005494546890258789 bandwidth (GB/s) 122.13721229493271
log_normal, size, elements 33554432 forward 0.0010767412185668946 bandwidth (GB/s) 124.65179718729368

After:

log_normal, size, elements 65536 forward 5.91278076171875e-06 bandwidth (GB/s) 44.33514628129032
log_normal, size, elements 131072 forward 7.789134979248047e-06 bandwidth (GB/s) 67.31017005056627
log_normal, size, elements 262144 forward 9.219646453857422e-06 bandwidth (GB/s) 113.73277763392811
log_normal, size, elements 524288 forward 1.5113353729248047e-05 bandwidth (GB/s) 138.7615242500079
log_normal, size, elements 1048576 forward 2.7089118957519532e-05 bandwidth (GB/s) 154.83353321964444
log_normal, size, elements 2097152 forward 4.64177131652832e-05 bandwidth (GB/s) 180.71997580169503
log_normal, size, elements 4194304 forward 8.719682693481446e-05 bandwidth (GB/s) 192.40626740399748
log_normal, size, elements 8388608 forward 0.0001693272590637207 bandwidth (GB/s) 198.16320293339717
log_normal, size, elements 16777216 forward 0.00033437252044677735 bandwidth (GB/s) 200.70089464986953
log_normal, size, elements 33554432 forward 0.0006206154823303223 bandwidth (GB/s) 216.26551676737367

Differential Revision: D15632930

@pytorchbot pytorchbot added module: cuda Related to torch.cuda, and CUDA support in general module: internals Related to internal abstractions in c10 and ATen module: operators labels Jun 3, 2019
@syed-ahmed syed-ahmed mentioned this pull request Jun 3, 2019
Closed
This was referenced Jun 3, 2019
Closed
Closed
Closed
@syed-ahmed
Update on "Move THCTensor_(exponential) to ATen"
e4d94c7 
Move THCTensor_(lognormal) to ATen

gh-metadata: pytorch pytorch 21299 gh/syed-ahmed/11/head
@syed-ahmed syed-ahmed changed the title Move THCTensor_(exponential) to ATen Move THCTensor_(lognormal) to ATen Jun 3, 2019
@syed-ahmed
Update on "Move THCTensor_(lognormal) to ATen"
ede6e76 
Move THCTensor_(lognormal) to ATen

gh-metadata: pytorch pytorch 21299 gh/syed-ahmed/11/head
@syed-ahmed syed-ahmed requested a review from ezyang June 4, 2019 05:53
@syed-ahmed syed-ahmed mentioned this pull request Jun 4, 2019
Closed
@syed-ahmed
Update on "Move THCTensor_(lognormal) to ATen"
e5ba146 
Move THCTensor_(lognormal) to ATen

gh-metadata: pytorch pytorch 21299 gh/syed-ahmed/11/head
@zou3519 zou3519 deleted the gh/syed-ahmed/11/head branch June 5, 2019 02:16
@facebook-github-bot
Copy link
Contributor

facebook-github-bot commented Jun 5, 2019

@ezyang merged this pull request in c82bf8e.

zdevito pushed a commit to zdevito/ATen that referenced this pull request Jun 5, 2019
@syed-ahmed @facebook-github-bot
Move THCTensor_(lognormal) to ATen (#21299)
af9ff95 
Summary:
Pull Request resolved: pytorch/pytorch#21299
ghimport-source-id: 2c63f289f02087f023feda8bff6b90ed49737889

Reviewed By: jerryzh168

Differential Revision: D15632930

Pulled By: ezyang

fbshipit-source-id: 85c17cdca486b46942c5b500e4fd4d95bb5657f9
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@ezyang ezyang ezyang approved these changes

Assignees

No one assigned

Labels

Merged module: cuda Related to torch.cuda, and CUDA support in general module: internals Related to internal abstractions in c10 and ATen open source

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

6 participants

@syed-ahmed @facebook-github-bot @ezyang @pytorchbot @mruberry