-
Only 4% of 64GB RAM being used?
Hi,
We have a new machine.
Case: “E-Series” E200IBE Midi Case (Supermicro® CSE-732D4F-903B) – 900W PSU (No Hotswap)
Mainboard: Supermicro® X9DAi Mainboard (16x DIMM Slots)
Graphics Card(s): NVIDIA GeForce GTX 780 – 3GB (PCI-E 3.0) – (2304 GPU Cores)
Hard Drive (1): PNY 480GB Previal 3k Endurance Series – SATA-III – SSD
Hard Drive (2): Intel® “Dale Crest – 530”- 240GB – SATA3 – SSD (AE SCRATCH EXCLUSIVE)
CPU(s): 2x Intel® Xeon® Processor E5-2650v2 (8x 2.60GHz / 20MB) [w/ HT]
Memory: 64GB DDR-3 1866MHz – ECC Registered (PC3-15000) RAMIt is slower than we thought it would be. Compariable to our iMac 2012, 32GB RAM, i7.
We have 29 cores running at 2GB (tried 3GB and less cores, but same result). We have enabled the GPU (although had a warning about it being untested and possibly unreliable).
RAM preview is a smidgen quicker, but the render is quite slow and says that AE is only using 4% of the RAM.
Any ideas?
Olly Lawer
-
4 Replies
-
Open Task Manager (Ctrl+Shift+Esc) and click the Performance tab. Please take a screen shot of this while rendering.
Walter Soyka
Principal & Designer at Keen Live
Motion Graphics, Widescreen Events, Presentation Design, and Consulting
RenderBreak Blog – What I’m thinking when my workstation’s thinking
Creative Cow Forum Host: Live & Stage Events -
Watch this. There are some tips in there for optimizing your multiprocessor settings. It will take some testing, but ideally, when you multiprocess, you only want AE to use as many physical cores as you have installed. In your case: 16. So set the allocated RAM until you’re only using 16 CPUS (there might be a few, hence the testing) and then start testing to see which is the fastest render.
https://www.youtube.com/watch?feature=player_embedded&v=a4D9ScFcM3M
Jason Jantzen
vimeo.com/jasonj -
Thank you. this was very useful! I always thought it was a good idea to use the virtual cores!! Have now allocated 16 of the 32 cores to other applications. In fact, I bumped the RAM up per core to 4GB so it only uses 14 cores as this surely should be quicker. Alas, it still isn’t that quick…
Here is a screenshot as requested.
Olly Lawer
-
This is probably useful too.
First pic: 14 ‘real’ cores at 4GB per core:
Second pic: 29 cores at 2GB per code:
Not much difference really.
Olly Lawer
Reply to this Discussion! Login or Sign Up