I have a dual Xeon, 18 cores on each, with Hyperthreading that's 36 cores x2 = 72 cores. Sadly TG4 is failing to take account of the second processor. This is a known issue with a few other DCC apps (the Foundry recently fixed the issue with Modo and it was an issue for V-ray too).
Any chance this could get fixed for those of us with beefy machines like this?
wow, nice machine!
In this thread something similar is discussed
http://www.planetside.co.uk/forums/index.php/topic,21616.15.html#msg217429
Here is a quote from Oshyan near the end of the thread:
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
"TG3 (and 4) are not ideal at multithreading, but we have made steady progress on the issue since TG2 and things are very good now for average machines (<32 threads). And both can make reasonable use of hyperthreads (logical rather than physical cores), although at very high core/thread counts (>32), it is best to only use physical cores as the overhead for use of hyperthreads becomes more than the modest performance gain that hyperthreading usually brings (~20%).
- Oshyan
« Last Edit: June 10, 2016, 06:28:39 AM by Oshyan »"
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
cheers, Klaus
Damn that sucks. You'd thnk the devs would notice it's 2016, multithreading has been around for a while.
Quote from: Matt Painter on July 02, 2016, 07:24:13 PM
Damn that sucks. You'd thnk the devs would notice it's 2016, multithreading has been around for a while.
Hmmm...
Maya and 3Ds Max as an example. Those posts are 1 year or so old. It might be harder then you think maybe?
http://forums.autodesk.com/t5/maya-general/maya-using-only-50-cpu-with-36-cores-72-threads/td-p/5484044
Yes, this was a known issue with a lot of apps, though Cinema 4D seemed okay right away, no fixes required. My understanding is that it's not an issue now with most apps.
Efficient multithreading is a pretty significant challenge, especially as core counts have really ramped up over time. As I said in that thread, using "real" (physical) cores and avoiding use of HT (easiest to disable it in the BIOS, though I know this makes it unavailable for all apps, which is a disadvantage) should give decent scaling even past 32 threads. I suggest you test that (HT disabled in BIOS) to see how well it works. If it's still not taking good advantage of your system then let us know.
Regardless we continue to try to optimize the renderer over time and it should get steadily more efficient on high-thread systems in the future.
- Oshyan