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

Use perf_analyzer without GPU/CUDA #1

Open
ShuaiShao93 opened this issue Mar 13, 2024 · 3 comments
Open

Use perf_analyzer without GPU/CUDA #1

ShuaiShao93 opened this issue Mar 13, 2024 · 3 comments

Comments

@ShuaiShao93
Copy link

Is your feature request related to a problem? Please describe.
We are building a perf analyzer workflow on CI, and we plan to use perf_analyzer as client. The client worker won't have GPU/CUDA. But I tried both pip install tritonclient and the NGC image, both failed by complaining there is no cuda/gpu.

Describe the solution you'd like
I don't get why perf_analyzer requires GPU. It should be able to run perfectly on machines that has no GPU.
@lkomali
Copy link
Contributor

lkomali commented Mar 13, 2024

Hi, can you clarify your setup?
What version of Triton are you using?
Are you running Triton from NGC container?
Can you clarify if the machine you are using do not have any GPU/CUDA installed and running models only on CPU?

@ShuaiShao93
Copy link
Author

What version of Triton are you using? Are you running Triton from NGC container?

I tried perf_analyzer in both nvcr.io/nvidia/tritonserver:24.02-py3-sdk and pip3 tritonclient 2.43.0. Both need GPU/CUDA.

Can you clarify if the machine you are using do not have any GPU/CUDA installed and running models only on CPU?

The triton server and perf_analyzer are run on separate machines. The perf_analyzer one doesn't have GPU/CUDA installed.

@lkomali lkomali transferred this issue from triton-inference-server/server Mar 13, 2024
@debermudez
Copy link
Contributor

@matthewkotila i thought we removed this requirement?

fpetrini15 pushed a commit that referenced this issue Jun 26, 2024
@tgerdesnv
Support for fixed number of requests (#633)
3ce873c 
* first pass. Hardcoded values

* Working for concurrency (hardcoded whenever count windows is used for now)

* working for req rate as well

* Add CLI. Add/fix unit tests

* Remove hack. Restore all normal functionality

* Refactor thread config into one class. Add more testing

* Rename arg to request-count

* Fix request rate bug

* Update info print

* fix corner case

* move fixme to a story tag

* add assert to avoid corner case

* rename variables

* self review #1

* copyright changes

* add doxygen to functions

* Don't allow sweeping over multiple concurrency or request rate with request-count
fpetrini15 pushed a commit that referenced this issue Jul 29, 2024
@tgerdesnv
Support for fixed number of requests (#633)
54ec281 
* first pass. Hardcoded values

* Working for concurrency (hardcoded whenever count windows is used for now)

* working for req rate as well

* Add CLI. Add/fix unit tests

* Remove hack. Restore all normal functionality

* Refactor thread config into one class. Add more testing

* Rename arg to request-count

* Fix request rate bug

* Update info print

* fix corner case

* move fixme to a story tag

* add assert to avoid corner case

* rename variables

* self review #1

* copyright changes

* add doxygen to functions

* Don't allow sweeping over multiple concurrency or request rate with request-count
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
3 participants
@debermudez @ShuaiShao93 @lkomali