08-31-2008 09:27 PM
Let, me explain how our studio infra is configured: We have
1. 47 numbers of Quad core HP 4600 workstations with FX1700 graphics card with 4 GB memory running on 64 bit WCCS,
2. 3 Number of 2 CPU Zeon HP 6600 workstations with FX1700 graphics card with 4 GB memory running on 64 bit WCCS,
3. 48 port foundry 1GB switch, 2 48 port unmanaged switch,
4. 15 machines of P4 (32 bit ) with 2 gb memory as secondary render farm running on Linux
5. Dell NX 1950 active-active cluster with MD 3000 12 TB storage,
6. Domain server running on DEL R300,
7. TL 2000 Tape library from DELL and Symantic Tape Backup software
Our animation pipline is on Maya 2008 - 64 bit application.
We are one of the first to implement WCCS for animation pipeline. We use this to make use of the unused processors for rendering, which is a back end process while the atrist are working in Maya.
We face virtual memory problem often. I I believe reading an articial were VM is related to PRocessor. I also believe WCCS is a processor based model and not a memory based model.
Can you help us daigonise the bottlenecks, using any of your tools hence we can optimise the performance. Example: what is the max Virtual Page We can go to get better performance. In aspect of core we have more than 200 core. We plan to add in the next 6-9 months another 1000 core.