Hi,
We recently noticed something strange during an animation project, where we have a corridor which the camera is flying through very, very slowly. For some reason, neighboring frame render times were drastically different when they should really not be. For example, frame 107 and frame 108 looks pretty much identical but the frame times are 5 hrs and 8 hrs respectively. I would have expected maybe a 10 minute difference between frames, but not 3 hours. And also, our predictions were telling us that the frame times should be around 3 hrs, not 5 or 8 hrs. So I started investigating...
Please see attached log and attached CPU/RAM graph for one of the frames in this sequence: vrayrender.zip
We're using Maya 2015 Extension 1 + Service Pack 6 with V-Ray 3.30.02 stable build 26678. Timestamps are synced between render log and graphs, and the issue can be seen at 2016/Mar/17 09:40:34.
As you can see from the V-Ray render log, the render started at 06:06:11 and is 98% completed by 09:40:34 that same day. This gives us around 3h 40 minutes of render time so far. But then something strange happens. The next time V-Ray outputs a progress message, 100%, the time is 14:08:05. That means that it took almost 5 additional hours to go from 98% to 100%.
When I took a look at the CPU history for that particular render, I noticed CPU levels were quite low. The machine in question has 12 cores. And the CPU level happens to look just like as if one single core was struggling for a long, long time with something. This leads me to believe that there was one stray bucket left still rendering at 98% and it didn't complete until 5 hours later.
Now I'm trying to figure out what could be so difficult to calculate, and so different from nearly identical frames. Have you guys got any intelligent guesses?
Could we perhaps help each other out and try and figure out if this could be a bug or if it could be the result of bad setup on our part?
Either way - there are A LOT of hours to save here, if we can find the root cause.
Please let me know what you think. If there was a render element which would act as a "render time heat map", I'm guessing it could be easier to find which part of the image is causing this...
Cheers,
Fredrik
EDIT 1: I can send you the scene, but I don't want to post it onto the forum. Please give me an email address I can send it to.
EDIT 2: Forgot to mention Maya and V-Ray version and I added it in at the beginning of the post...
We recently noticed something strange during an animation project, where we have a corridor which the camera is flying through very, very slowly. For some reason, neighboring frame render times were drastically different when they should really not be. For example, frame 107 and frame 108 looks pretty much identical but the frame times are 5 hrs and 8 hrs respectively. I would have expected maybe a 10 minute difference between frames, but not 3 hours. And also, our predictions were telling us that the frame times should be around 3 hrs, not 5 or 8 hrs. So I started investigating...
Please see attached log and attached CPU/RAM graph for one of the frames in this sequence: vrayrender.zip
We're using Maya 2015 Extension 1 + Service Pack 6 with V-Ray 3.30.02 stable build 26678. Timestamps are synced between render log and graphs, and the issue can be seen at 2016/Mar/17 09:40:34.
As you can see from the V-Ray render log, the render started at 06:06:11 and is 98% completed by 09:40:34 that same day. This gives us around 3h 40 minutes of render time so far. But then something strange happens. The next time V-Ray outputs a progress message, 100%, the time is 14:08:05. That means that it took almost 5 additional hours to go from 98% to 100%.
When I took a look at the CPU history for that particular render, I noticed CPU levels were quite low. The machine in question has 12 cores. And the CPU level happens to look just like as if one single core was struggling for a long, long time with something. This leads me to believe that there was one stray bucket left still rendering at 98% and it didn't complete until 5 hours later.
Now I'm trying to figure out what could be so difficult to calculate, and so different from nearly identical frames. Have you guys got any intelligent guesses?
Could we perhaps help each other out and try and figure out if this could be a bug or if it could be the result of bad setup on our part?
Either way - there are A LOT of hours to save here, if we can find the root cause.
Please let me know what you think. If there was a render element which would act as a "render time heat map", I'm guessing it could be easier to find which part of the image is causing this...
Cheers,
Fredrik
EDIT 1: I can send you the scene, but I don't want to post it onto the forum. Please give me an email address I can send it to.
EDIT 2: Forgot to mention Maya and V-Ray version and I added it in at the beginning of the post...
Comment