System speed


Treehouse2339
 Share

Recommended Posts

19 minutes ago, Treehouse2339 said:

In a perfect world how long should the first pass take?

 

About 1/2 a second would be prefect for me but personal computing power is not there yet.  The best you can get for fast raytracing will be the absolute latest in Dual Xeon processors.  $$$$.

 

Just a word of caution though as these do not usually have the highest clock speed it will cause other processes in Chief to lag.  So I should rephrase and say a dedicated raytrace machine should have Dual Xeon processors.

  • Upvote 1
Link to comment
Share on other sites

Someone should post a plan with a fixed view and raytaced scene and then we can all report back on how long it took to render 10 passes on our systems. 

 

Post screenshot of completed number of passes which also shows the Raytrace in question. 

 

Wouldn't this be the best bench mark comparison?

 

Of course post what system you used if it’s different from your profile description. 

 

...and be honest. 

Link to comment
Share on other sites

The ONLY way you can do anywhere near a scientific bench test is for all testers to use the exact, same .plan file and then you would only know about the tested systems, it would tell you only about the systems tested. That might be worth the time to fiddle with it.

 

DJP

 

  • Upvote 1
Link to comment
Share on other sites

Ok so that’s the challenge then. 

 

Somenody post a simple “test” plan with a few lights. Save a “test” view and set the “test” raytrace settings and hit raytrace. 

 

The raytrace setting should be set to number of passes, maybe 10, only, so it’s not too long of an excercise. 

  • Upvote 1
Link to comment
Share on other sites

I also believe we tried something like this a year or two ago and only a few responded, I believe it was based upon the Grandview plan in the CA sample gallery. One thing with Raytracing is that is fairly easy to predict the results. Raytracing is 100% CPU based and the per time pass is directly proportional to the CPU frequency and the number of cores. Double the cores and your per pass time is halved, increase the CPU frequency and your per pass time will be reduced accordingly.

 

If you are PBRing then it's a bit more complex as both the CPU and GPU is involved however from what I can see it is the GPU that is by far the most important in the equation.

Link to comment
Share on other sites

5 hours ago, TheKitchenAbode said:

If you are PBRing then it's a bit more complex as both the CPU and GPU is involved however from what I can see it is the GPU that is by far the most important in the equation.

This is good to know as I am probably going to need a new machine some time this year. Thanks Graham, and so many others, who contribute so much to this forum.

Link to comment
Share on other sites

Given the flexibility of PBR and the community's enthusiasm for it (me at least), I would encourage the OP to consider maximizing his investment in a graphics card.  I'm assuming RT will eventually give way to PBR, assuming again that PBR improves over time.

 

I acquired a GTX 1070ti with PBR and path walkthroughs in mind.  I find most of my walkthroughs push the GTX1070ti to around 50% while the CPU is at 95%.  It does depend on the complexity of the walkthrough scenes and parameters.  On the other hand, we don't know precisely where CA will go next: hand over more functions to the video card (see this post), optimize and expand CPU parallel processing?

 

The answer for now is: determine how you use CA and see which hardware pressure point needs some relief.  If for you the answer is RT, then multicore CPU with high frequency is the way.

 

P.S. I'm surprised one of the regulars here didn't ask you to create a signature with your configuration: you would have gotten more precise answers I think.

  • 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