Network rendering - new 32 core nodes issues

[chrisjacks21]

Ooohh - sounds like there has been some developments whilst I have been away! Have emailed Vlado for the fix so will let you know if performance improves on our E5's...

[instinct]

for c/c++ cloc works nicely : http://cloc.sourceforge.net/

Also one for the marketing people. Running gource on your repositories can be pretty entertaining. e.g. Python: http://www.youtube.com/watch?v=aPk1B...eature=related

Regards,
Thorsten

[vance3d]

anything resolved on this? we have the slow down as well without proxies involved as well.

V Miller

[vlado]

Yes, the nightly builds include all that stuff already. If you don't have access to them, please email us to vray@chaosgroup.com

Best regards,
Vlado

[vance3d]

we tried the latest nightly and it didn't solve our issues. we are trying to trace the cause of the slow down but haven't found the problem yet.

V Miller

[vlado]

Ok then; first of all, make sure you also turn off the 3ds Max bitmap pager. If you continue to have issues, please email me a scene to vlado@chaosgroup.com

Best regards,
Vlado

[vance3d]

bitmap pager seems to have been the problem. the systems are rendering as fast as they should now. thanks for the pointer.

V Miller

[vance3d]

An update to our experiences with E5-2650's

After making sure that bitmap paging was turned off, our rendering times on these machines was reduced to times that were on par with what we expected from them. There seems to be an exception though when it comes to displacement. We are currently rendering a displaced surface that on our I7's will render in 35 minutes and on the E5's it takes an hour and a half. We have tried several things to diagnose the problem including changing memory settings in vray and sending all the maps to the machines so they don't draw them from the server while rendering. Are there scalability issues with displacement? The machines in question are dual 8 core with hyperthreading and 32g of ram.

Any thoughts?

V Miller

[vlado]

Are there scalability issues with displacement?

Not that I'm aware of, but then again, I thought the same about proxies It would be good to get me a scene for testing, so that I can check what's going on.

Best regards,
Vlado

[vance3d]

I'll put something together and send it to you.

thanks,

V Miller

[vlado]

Thanks for the file; the issue is not the displacement at all, but the storing of illumination samples in the sss2 map for the fastsss2 shader. We will have to work on optimizing this; I will let you know when we have something, hopefully in a couple of days.

Best regards,
Vlado

[vance3d]

Thanks vlado. I'm always impressed with how quickly you guys respond to things. I look forward to what you come up with.

V Miller

[Morbid Angel]

Thanks for the file; the issue is not the displacement at all, but the storing of illumination samples in the sss2 map for the fastsss2 shader. We will have to work on optimizing this; I will let you know when we have something, hopefully in a couple of days.

Best regards,
Vlado

Vlado can this be implemented into maya as well? as I believe it could resolve some of the issues we used to have.

[vlado]

Vlado can this be implemented into maya as well? as I believe it could resolve some of the issues we used to have.

It is part of the core V-Ray, so it will go into all V-Ray implementations. It's a bit tricky to get right and stable though, like with all MT stuff...

Best regards,
Vlado

[vance3d]

Any updates on the fast sss issue?

V Miller