Skip to content
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

[npu][hybrid] support offload #37224

Merged
merged 1 commit into from
Nov 17, 2021
Merged

[npu][hybrid] support offload #37224

merged 1 commit into from
Nov 17, 2021

Conversation

wangxicoding
Copy link
Contributor

@wangxicoding wangxicoding commented Nov 15, 2021 •
edited
Loading

PR types

Others

PR changes

Others

Describe

npu hybrid parallel support offload.

test

Ernie3.0，base模型，NPU单机8卡
baseline=2mp+2pp+2dp

模型配置 值
hidden_size 3072
num_attention_heads 48
num_hidden_layers 39
num_sharding_layers 36
branch_hidden_size 256
branch_num_attention_heads 4
显存
卡id baseline(MB) offload(MB) 节省显存(MB)
0 21954 12451 9503(43.3%)
1 21954 12451 9503(43.3%)
2 25559 14745 10814(42.3%)
3 25559 14745 10814(42.3%)
速度
baseline(tokens/s) offload(tokens/s) 降速
3196.6 3114.0 2.6%

@paddle-bot-old
Copy link

Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.

zhiqiu
zhiqiu approved these changes Nov 16, 2021
View reviewed changes
Copy link
Contributor

@zhiqiu zhiqiu left a comment

Choose a reason for hiding this comment

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

LGTM

@PaddlePaddle PaddlePaddle locked and limited conversation to collaborators Nov 16, 2021
@PaddlePaddle PaddlePaddle unlocked this conversation Nov 16, 2021
gongweibao
gongweibao approved these changes Nov 16, 2021
View reviewed changes
Copy link
Contributor

@gongweibao gongweibao left a comment

Choose a reason for hiding this comment

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

LGTM

@wangxicoding wangxicoding merged commit 762819a into PaddlePaddle:develop Nov 17, 2021
@wangxicoding wangxicoding deleted the npu_hybird_offload branch November 17, 2021 03:45
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@gongweibao gongweibao gongweibao approved these changes

@zhiqiu zhiqiu zhiqiu approved these changes

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@wangxicoding @zhiqiu @gongweibao