OK, so I do not have the answer, but what I've come up with might explain some of the discrepancies in these posts.
I decided to run some experiments, based on the fact that I had been pleased with my perceived render time improvements in V6.
I have played with making a project in V4, rendering to a new track (applying alpha overlays, video filters and picture in picture) and then trying the same thing in V6.
My experiments so far have left me very confused. I need to document exactly what is going on, but basically I've seen the following:
Using a 15 second clip, and applying a PIP and a PNG title to the clip in V4, the render times were 36 seconds. In V6, the same project rendered in 27 seconds no matter how many rendering threads I selected in the options section.
YET, in earlier experiments, there was a major difference in rendering times for another clip with PIP when changing the rendering threads.
Applying a Chroma Blur filter to a clip in V6, and rendering to a new track, gave me wildly different results!
Rendering a 15 second clip with chroma blur using one thread yielded a 38 sec render time.
Without closing the application, and changing it to 2 threads I showed a 16 minute render time for the exact same thing!
Leaving the settings at 2 threads, closing and re-opening the application with the same .veg, now showed a render time of 6 minutes!
Changing back to 1 thread (withut closing the application) now shows a render time for about 33 minutes!
Changing nothing, and closing out and re-opening the app, now gives a render time of about 38 seconds again.
Something's lying! There appears to be no consistency to this. Looking at CPU utilization (I have a single P4 3G processor with hyperthreading) gives me lots of different information, but I need to process this when I'm more awake (it's 3AM).
Suffice to say, at this point field reports may mean nothing as to render times, improved or degraded, until this all gets sorted out.
From what I've seen so far, for me it would be best to set the threads to 1 and leave it there, especially with a hyperthreaded P4, and especially in the middle of working on a project
Hopefully I'll find some answers to pass along shortly, but for now I've got to finish what I'm working on before more "playtime".
Pass along what you discover based on this.
- Plas
I decided to run some experiments, based on the fact that I had been pleased with my perceived render time improvements in V6.
I have played with making a project in V4, rendering to a new track (applying alpha overlays, video filters and picture in picture) and then trying the same thing in V6.
My experiments so far have left me very confused. I need to document exactly what is going on, but basically I've seen the following:
Using a 15 second clip, and applying a PIP and a PNG title to the clip in V4, the render times were 36 seconds. In V6, the same project rendered in 27 seconds no matter how many rendering threads I selected in the options section.
YET, in earlier experiments, there was a major difference in rendering times for another clip with PIP when changing the rendering threads.
Applying a Chroma Blur filter to a clip in V6, and rendering to a new track, gave me wildly different results!
Rendering a 15 second clip with chroma blur using one thread yielded a 38 sec render time.
Without closing the application, and changing it to 2 threads I showed a 16 minute render time for the exact same thing!
Leaving the settings at 2 threads, closing and re-opening the application with the same .veg, now showed a render time of 6 minutes!
Changing back to 1 thread (withut closing the application) now shows a render time for about 33 minutes!
Changing nothing, and closing out and re-opening the app, now gives a render time of about 38 seconds again.
Something's lying! There appears to be no consistency to this. Looking at CPU utilization (I have a single P4 3G processor with hyperthreading) gives me lots of different information, but I need to process this when I'm more awake (it's 3AM).
Suffice to say, at this point field reports may mean nothing as to render times, improved or degraded, until this all gets sorted out.
From what I've seen so far, for me it would be best to set the threads to 1 and leave it there, especially with a hyperthreaded P4, and especially in the middle of working on a project
Hopefully I'll find some answers to pass along shortly, but for now I've got to finish what I'm working on before more "playtime".
Pass along what you discover based on this.
- Plas