Skip to content

Conversation

@abc99lr
Copy link
Contributor

@abc99lr abc99lr commented Feb 7, 2023

This MR adds GUPS code to code-samples. Our version of GUPS is a GPU random access benchmark. It supports testing random access performance for both global memory and shared memory. The benchmark supports read, write, read_write, updates (with/without loops) access types.

Besides the CUDA code for benchmarking (gups.cu), we also provide a Makefile and a Python script (run.py) for running tests in batch.

The GUPS implementation in gups.cu was modified based on open-sourced code. The license of the code is maintained in LICENSE.gups.cu. For the license information of other files, please refer to LICENSE.

@nsakharnykh
Copy link
Contributor

tagging @sleeepyjack for review

@harrism
Copy link
Member

harrism commented Feb 22, 2023

Is this associated with a specific blog post?

@sleeepyjack
Copy link

Is this associated with a specific blog post?

This benchmark is used in the upcoming blog post about cuco. I sent you an invitation to the draft.

@nsakharnykh nsakharnykh merged commit 9b28573 into NVIDIA-developer-blog:master Mar 2, 2023
@harrism
Copy link
Member

harrism commented Mar 6, 2023

I didn't get a chance to submit my review but I was going to say that you need to end your files with blank lines. (If you look at the code, github warns about these).

@abc99lr
Copy link
Contributor Author

abc99lr commented Mar 6, 2023

I didn't get a chance to submit my review but I was going to say that you need to end your files with blank lines. (If you look at the code, github warns about these).

Hi Mark, could you take a look at #50? I added newlines and created #50 for that changes. Thanks

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants