CompuServe Thread

#Interesting results

3 messages in this thread
#157888From: Chris JohnstonMar 8, 1995 10:19 PM
I've been rendering a big animation at work and have finally got network rendering to mostly work. I got an interesting result that I thought that I would share: 486DX2-50, 32 MB RAM, 22MB swap file renders in 15 min. Pentium 66, 16 MB RAM, 42MB swap file renders in about 45 min. These are on adjacent frames, so the rendering times should be very similar. I've also got a question to anybody that knows: does a spotlight that is set to cast shadows actually calculate a shadow map when its intensity has been turned down to 0? Chris – Chris Johnston, Written 02-Mar-1995 @ 23:16:14 Using: Windows NavCIS PRO 1.5
#157923From: Michael J SheldonMar 9, 1995 1:07 AM
>> 486DX2-50, 32 MB RAM, 22MB swap file renders in 15 min. Pentium 66, 16 MB RAM, 42MB swap file renders in about 45 min. These are on adjacent frames, so the rendering times should be very similar. << I'd be shocked if they were. The swap file size tells the story. 3ds is using the disk as memory, once you do this, your processor speed means very little. The governing factor becomes your DISK speed. Put equal memory in both systems, and the Pentium should edge out the 486. Put equal memory in both, then eliminate the need for the swap file, and you should see a big difference. – Michael J Sheldon "Falcon", msheldon@crl.com – When we are planning for posterity, we ought to remember that virtue is not hereditary. Thomas Paine, Common Sense [1776]
#158309From: Chris JohnstonMar 10, 1995 9:55 PM
Michael: I'm sorry that I left the impression that I expected that the rendering times on the two machines would be the same. What I was getting at was that there was very little difference between the two frames in question so, **everything else being equal *** there should be very little rendering time difference between them based on the content of the frame. I was assuming that the faster rendering time for the 486 was due to the larger swap file necessary on the Pentium. The approximate factor of 2 in the swap file size swamped out the difference in the speed of the processors. I find it interesting that in some cases the Pentium shows faster rendering times even with a considerably bigger swap file. For that matter, the rendering time varies from one frame to another, usually getting faster for subsequent frames as long as the scene stays relatively the same. My guess on this is that the renderer manages to reuse a fair amount of the work that it does on the first frame, so the whole process speeds up as you go. I would love to max out the memory on the machines that I'm using – we just don't have the budget for it right now. (animation is only one of my responsibilities, and not necessarily the major one) On most of my previous projects I have been able to simplify the model enough to avoid swapping to disk. This one is **way** to big to simplify that much so I have to live with the swapping. Chris – Chris Johnston, Written 04-Mar-1995 @ 22:42:50 Using: Windows NavCIS PRO 1.5