Recommended Posts

I'm running CA Premier x6, 64bit, v. 16.4.0.81 (7/21/14). I have 3+ year old Dell Studio XPS 9100, Intel I7-930 (2.8Ghz, 8MB cache), 12GB RAM (1333 DDR3). This is a quad core processor.  Video is AMD 5670 (1GB), 64bit. Win7, 64 bit.

When I initiate a Ray Trace, CPUs are buried @ 100%. But interestingly, only about 25-35% (3-4GB) of system memory is used. When I open Performance Monitor, I see that Ray Trace spawns only 9 threads, while CA itself has opened 20+ threads under the same Process ID (PID).

I know that the video card is fairly pedestrian, and that I can throw in a Radeon 6970, 2GB, 256-bit card for about $100. Likewise, I could go to a Samsung 840 SSD for about $200, and really  improve the overall performance of the computer.

But would that really help Ray Trace? I've been reading about how resource intensive Ray Trace is, but what I'm seeing in PerfMon has me reconsidering my upgrade plans. Is Ray Trace really so CPU-bound? Are there any config/parameter files that can be tweaked in order to allocate more threads to CA/Ray Trace?  PerfMon tells me my computer's resources like IO and memory are underutilized, except for CPU.

Are there any tuning parameters possible? Or should I be thinking exclusively about processor speed and cores?

 

Thanks for any thoughts.
 

Link to comment
Share on other sites

Ray tracing is one of several computational problems that is very cpu intensive. Generally speaking the time to do a ray trace is based primarily on the multiplication of number of pixels by number of light sources. There are other factor as well but they are normally drawfed by that cost.

 

To make a ray trace faster:

1) Get more and faster cores.

2) Reduce the number of lights.

3) Reduce the number of pixels in the resulting image.

 

An SSD may help with the startup time, and certainly helps with many other operations in Chief. But that is a very minor improvement to ray tracing.

 

Unfortunately at this time we don't take advantage of the processing power of the video card for ray tracing. But it would certainly help with other tasks.

 

For the case you were looking at it doesn't look like more memory will help with your raytrace, but it might help with file caching or if you run across a really large model.

  • Upvote 2
Link to comment
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
 Share