Q100289: Understanding performance and export speeds between Mari 2 and Mari 3

Follow

SUMMARY

This article explains the differences in performance and export speeds across various versions of Mari, as well as some of the reasons behind them.

 

MORE INFORMATION


HOW TO TEST MARI EXPORT SPEEDS

To test the export speeds in Mari you can run the following simple script via Mari's Python console that will record the time it takes to export textures out of Mari:

import timeit

start = timeit.default_timer()

mari.geo.list()[0].channelList()[0].exportImagesFlattened('To_Save_Path/img_$CHANNEL.$UDIM.tif')

elapsed = timeit.default_timer() - start

print elapsed

NOTE: For the script above please replace the path in exportImagesFlattened with the location where you'd like your exported textures to be saved at. 

 

MARI 2 VS. MARI 3 PERFORMANCE AND EXPORT SPEEDS

Users who are comparing the export speeds between different Mari versions are likely to experience slower export times in Mari 3 compared to Mari 2.

With Mari 3, we introduced Colour Management which means that it now takes longer for the machine to compute during the baking and exporting process as it requires more data processing. However, we have been working to counter this with each new release and by running the test in each version we can see a clear improvement in export times between Mari 3.1v1 and Mari 3.2v1.  

In Mari 3.2v1, we also introduced the ability for the user to turn off Colour Management by navigating to File > Settings and unchecking 'Color Management enabled' . By forcing Mari to disable Colour Management, there is less demand on the data processes which should enforce similar performance in Mari 3.2v1 as in Mari 2.6v5.

Running the above export speeds script in Mari 3.2v1's Example Head project with Color Management turned on and off, will show the following time results:
 
Turned off:
1.97729602252

Turned on:
3.47527038408


FURTHER READING

If you are experiencing other performance issues there are further settings or options that can be turned off to improve Mari’s efficiency. A detailed article about additional causes to consider is found here: Q100253: Troubleshooting Mari performance issues

Another useful article for best project performance can be found here:  Q100188: Optimising your Mari projects for best performance

 
As we understand that performance issues can sometimes be directly dependant on your project and setup, if you are still experiencing issues after testing the advice in these articles please contact us via the Support Portal and let us know the steps you have taken so far. All initial information we require for investigation is outlined under Q100090: Reporting a Mari issue

For more information on how to open a Support ticket, please refer to the 'Using the Support Portal' article.

Was this article helpful?
0 out of 0 found this helpful

Comments