Compare GPUs using standard numerical benchmarks in MATLAB.
Aktualisiert 25. Jul 2023

Lizenz anzeigen

Anmerkung des Herausgebers: This file was selected as MATLAB Central Pick of the Week

GPUBENCH times different MATLAB GPU tasks and estimates the peak performance of your GPU in floating-point operations per second (FLOP/s). It produces a detailed HTML report showing how your GPU's performance compares to pre-stored performance results from a range of other GPUs.
Note that this tool is designed for comparing GPU hardware. It does not compare GPU performance across different MATLAB releases.
Requires MATLAB R2016b or above and a GPU with CUDA Compute Capability 3.5 or higher.

Zitieren als

MathWorks Parallel Computing Toolbox Team (2024). GPUBench (, MATLAB Central File Exchange. Abgerufen .

Kompatibilität der MATLAB-Version
Erstellt mit R2023b
Kompatibel mit R2016b und späteren Versionen
Windows macOS Linux

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!
Version Veröffentlicht Versionshinweise

- Fix bug where lack of CPU memory could prevent running on GPU.
- Fix bug where single output syntax errored out.


Fix string issues with MATLAB version R2016b.


- Update stored data to have one JSON file per device
- Add data-file links to the summary report to make sharing results easier
- Add line styles to report plots better differentiate results when more than 8 devices are plotted

Update reference data for R2018a to R2020b.
Update report style.
Make summary table sortable.
Remove the warning in report when the data was generated using a different release to the client's.

Update report style.
Remove warning about old pre-stored results.

Update R2017b data file
Fix a problem with the data location when running the app in R2014b
Add data files for R2017a

* Improve compatibility with MATLAB R2014b
* Update gpuBench with data for R2014b,...,R2016a
* Clean gpuBench code by using gputimeit and by removing dead code
* Make gpuBench robust when running MATLAB with -nodesktop or -nojvm

Fix recursion problems when using the MATLAB App version.

* Add datafile for R2013b

* Add results for R2013a (including K20!)

* Suppressed warnings about results being skipped
* Now includes a set of pre-stored host-PC data so that you get a rough CPU/GPU comparison when just viewing the report
* Reduced largest size used for MTIMES to avoid out of memory

* Add an "app" version for use with R2012b and above
* Updated data-files for R2012a and R2012b

Try to prevent timeout being hit on very slow GPUs that happen to be driving the display as well.

Add data for C2075