bbigger
03-25-2005, 11:23 PM
After doing a network render on a 3-node farm I yielded some pretty nice results time-wise per node per packet size. When I decided to offload the render to just the 2 "headless" machines, the time per node per packet was about double.
Specs: The main machine is a dual 2400 AMD with a gig of RAM and the "headless" machines are single 1.8 Ghz P4's with a gig of RAM.
My question is this: Is it faster, better, or optimal to have the machine they are being queued from and where the files are being originated/stored to be running in the renderfarm or not? I'm just using a mapped drive on my main machine as the access point to draw and store data.
If there's a way to optimize a network render, any info would be appreciated.
I hope that made sense. Thanks!
bbigger
Specs: The main machine is a dual 2400 AMD with a gig of RAM and the "headless" machines are single 1.8 Ghz P4's with a gig of RAM.
My question is this: Is it faster, better, or optimal to have the machine they are being queued from and where the files are being originated/stored to be running in the renderfarm or not? I'm just using a mapped drive on my main machine as the access point to draw and store data.
If there's a way to optimize a network render, any info would be appreciated.
I hope that made sense. Thanks!
bbigger