[HOU-4224] Heavy Instancer RAM Usage Compared to Karma

[Nevoreko]

Hi,

After getting many RAM issues with instancers we started investigating and found this behavior, where loading 2 instancers with the same prototype reference will use double the amount of RAM, it doesn't matter how it's setup, we tried to have shared prototype, instanceable prototype, it always get loaded twice in RAM.

In this example scene we got some multiple heavy instancers as a test, and the RAM usage reads as such:
Karma CPU - 17.5Gb
V-Ray - 43Gb

https://we.tl/t-jVtjUg5yNz

Would there be any way to get on par with Karma on your side or some workaround on our side?

V-Ray 6.2.05.22651
Houdini 20.0.590

Thanks,
Marvin

[viktor_angelov]

Hello Nevoreko the matter has been logged into our system [ HOU-4224 ] We will reach out once a fix has been implemented.

[paolo_selva1]

Hi, I'm also interested in any update about this matter, if anything, thank you!

In my previous company, we used a new API call that SideFX (Mark) put in place in their internal OpenUSD, to be able to achieve this and we used it with success in our delegate, in case this helps:

https://github.com/PixarAnimationStu...nUSD/pull/1745

[viktor_angelov]

Hello Nevoreko paolo_selva1 I am glad to let you know that the matter has been done. The Multiple GeomInstancer plugins instances the same prototype now which decreases the amount of RAM used.

You can download any of the latest nightly builds and test the implementation: https://nightlies.chaos.com/main#/vray6houdini/nightly