Skip to content

[vulkan] addmm support non-vulkan inputs #39078

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
IvanKobzarev wants to merge 13 commits into from
Closed

[vulkan] addmm support non-vulkan inputs #39078

IvanKobzarev wants to merge 13 commits into from

Conversation

@IvanKobzarev
Copy link
Contributor

@IvanKobzarev IvanKobzarev commented May 27, 2020 •
edited
Loading

Stack from ghstack:

Adding support of non-vulkan inputs for addmm operator:
if it is not on vulkan - converting to it inside operator,

if we run torchscript pretrained model - weights of linear op will be on CPU, we need this to run mobilenetV2 on Vulkan backend

Differential Revision: D21962425

This was referenced May 27, 2020
Closed
Closed
Closed
@dr-ci
Copy link

dr-ci bot commented May 27, 2020 •
edited
Loading

💊 CI failures summary and remediations

As of commit f968a65 (more details on the Dr. CI page):

🚧 1 fixed upstream failure:

These were probably caused by upstream breakages that were already fixed.

Please rebase on the viable/strict branch (expand for instructions)

Since your merge base is older than viable/strict, run these commands:

git fetch https://github.com/pytorch/pytorch viable/strict
git rebase FETCH_HEAD

Check out the recency history of this "viable master" tracking branch.

ci.pytorch.org: 2 failed

This comment was automatically generated by Dr. CI (expand for details).Follow this link to opt-out of these comments for your Pull Requests.

Please report bugs/suggestions on the GitHub issue tracker or post in the (internal) Dr. CI Users group.

See how this bot performed.

This comment has been revised 45 times.

@IvanKobzarev IvanKobzarev mentioned this pull request May 27, 2020
Closed
@IvanKobzarev
drop batch norm stub as fuse_models Conv2-BatchNorm2 works on "[vulka…
f31db2c 
…n] addmm support non-vulkan inputs"


Adding support of non-vulkan inputs for addmm operator:
if it is not on vulkan - converting to it inside operator, 

if we run torchscript pretrained model - weights of linear op will be on CPU, we need this to run mobilenetV2 on Vulkan backend

[ghstack-poisoned]
This was referenced May 30, 2020
Closed
Closed
@IvanKobzarev
rebase on "[vulkan] addmm support non-vulkan inputs"
ad39f9b 
Adding support of non-vulkan inputs for addmm operator:
if it is not on vulkan - converting to it inside operator, 

if we run torchscript pretrained model - weights of linear op will be on CPU, we need this to run mobilenetV2 on Vulkan backend

[ghstack-poisoned]
@IvanKobzarev IvanKobzarev mentioned this pull request Jun 2, 2020
Closed
@IvanKobzarev
Update on "[vulkan] addmm support non-vulkan inputs"
7be72f1 
Adding support of non-vulkan inputs for addmm operator:
if it is not on vulkan - converting to it inside operator, 

if we run torchscript pretrained model - weights of linear op will be on CPU, we need this to run mobilenetV2 on Vulkan backend

[ghstack-poisoned]
AshkanAliabadi
AshkanAliabadi approved these changes Jun 10, 2020
View reviewed changes
aten/src/ATen/native/vulkan/VulkanAten.cpp Outdated
VulkanTensor t = vtensor_from_vulkan(self.is_vulkan() ? self : self.vulkan());
VulkanTensor m1 =
vtensor_from_vulkan(mat1.is_vulkan() ? mat1 : mat1.vulkan());
VulkanTensor m2 =
Copy link
Contributor

@AshkanAliabadi AshkanAliabadi Jun 10, 2020

Choose a reason for hiding this comment

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

const is preferred to protect against the inadvertent changes below.

@IvanKobzarev
Update on "[vulkan] addmm support non-vulkan inputs"
629425d 
Adding support of non-vulkan inputs for addmm operator:
if it is not on vulkan - converting to it inside operator, 

if we run torchscript pretrained model - weights of linear op will be on CPU, we need this to run mobilenetV2 on Vulkan backend

Differential Revision: [D21962425](https://our.internmc.facebook.com/intern/diff/D21962425)

[ghstack-poisoned]
@IvanKobzarev
Update on "[vulkan] addmm support non-vulkan inputs"
9a157d5 
Adding support of non-vulkan inputs for addmm operator:
if it is not on vulkan - converting to it inside operator, 

if we run torchscript pretrained model - weights of linear op will be on CPU, we need this to run mobilenetV2 on Vulkan backend

Differential Revision: [D21962425](https://our.internmc.facebook.com/intern/diff/D21962425)

[ghstack-poisoned]
@IvanKobzarev
Update on "[vulkan] addmm support non-vulkan inputs"
ba6520e 
Adding support of non-vulkan inputs for addmm operator:
if it is not on vulkan - converting to it inside operator, 

if we run torchscript pretrained model - weights of linear op will be on CPU, we need this to run mobilenetV2 on Vulkan backend

Differential Revision: [D21962425](https://our.internmc.facebook.com/intern/diff/D21962425)

[ghstack-poisoned]
@IvanKobzarev
Update on "[vulkan] addmm support non-vulkan inputs"
e635595 
Adding support of non-vulkan inputs for addmm operator:
if it is not on vulkan - converting to it inside operator, 

if we run torchscript pretrained model - weights of linear op will be on CPU, we need this to run mobilenetV2 on Vulkan backend

Differential Revision: [D21962425](https://our.internmc.facebook.com/intern/diff/D21962425)

[ghstack-poisoned]
@IvanKobzarev
Update on "[vulkan] addmm support non-vulkan inputs"
f12b527 
Adding support of non-vulkan inputs for addmm operator:
if it is not on vulkan - converting to it inside operator, 

if we run torchscript pretrained model - weights of linear op will be on CPU, we need this to run mobilenetV2 on Vulkan backend

Differential Revision: [D21962425](https://our.internmc.facebook.com/intern/diff/D21962425)

[ghstack-poisoned]
@IvanKobzarev
Update on "[vulkan] addmm support non-vulkan inputs"
8ad1d7c 
Adding support of non-vulkan inputs for addmm operator:
if it is not on vulkan - converting to it inside operator, 

if we run torchscript pretrained model - weights of linear op will be on CPU, we need this to run mobilenetV2 on Vulkan backend

Differential Revision: [D21962425](https://our.internmc.facebook.com/intern/diff/D21962425)

[ghstack-poisoned]
@IvanKobzarev
Update on "[vulkan] addmm support non-vulkan inputs"
eef569e 
Adding support of non-vulkan inputs for addmm operator:
if it is not on vulkan - converting to it inside operator, 

if we run torchscript pretrained model - weights of linear op will be on CPU, we need this to run mobilenetV2 on Vulkan backend

Differential Revision: [D21962425](https://our.internmc.facebook.com/intern/diff/D21962425)

[ghstack-poisoned]
@IvanKobzarev
Update on "[vulkan] addmm support non-vulkan inputs"
142e121 
Adding support of non-vulkan inputs for addmm operator:
if it is not on vulkan - converting to it inside operator, 

if we run torchscript pretrained model - weights of linear op will be on CPU, we need this to run mobilenetV2 on Vulkan backend

Differential Revision: [D21962425](https://our.internmc.facebook.com/intern/diff/D21962425)

[ghstack-poisoned]
@IvanKobzarev
Update on "[vulkan] addmm support non-vulkan inputs"
f968a65 
Adding support of non-vulkan inputs for addmm operator:
if it is not on vulkan - converting to it inside operator, 

if we run torchscript pretrained model - weights of linear op will be on CPU, we need this to run mobilenetV2 on Vulkan backend

Differential Revision: [D21962425](https://our.internmc.facebook.com/intern/diff/D21962425)

[ghstack-poisoned]
@facebook-github-bot
Copy link
Contributor

facebook-github-bot commented Jun 12, 2020

@IvanKobzarev merged this pull request in 71372b4.

@facebook-github-bot facebook-github-bot deleted the gh/IvanKobzarev/22/head branch June 15, 2020 14:15
xwang233 pushed a commit to xwang233/pytorch that referenced this pull request Jun 20, 2020
@IvanKobzarev @xwang233
[vulkan] addmm support non-vulkan inputs (pytorch#39078)
24e3ad8 
Summary:
Pull Request resolved: pytorch#39078

Adding support of non-vulkan inputs for addmm operator:
if it is not on vulkan - converting to it inside operator,

if we run torchscript pretrained model - weights of linear op will be on CPU, we need this to run mobilenetV2 on Vulkan backend

Test Plan: Imported from OSS

Differential Revision: D21962425

Pulled By: IvanKobzarev

fbshipit-source-id: 8222edd31dfb14b326d15e6fec5c8778783479df
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@dreiss dreiss Awaiting requested review from dreiss

1 more reviewer

@AshkanAliabadi AshkanAliabadi AshkanAliabadi approved these changes

Reviewers whose approvals may not affect merge requirements

Assignees

No one assigned

Labels

Merged

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

5 participants

@IvanKobzarev @facebook-github-bot @AshkanAliabadi @mruberry