It definitely decreased my render time with the Rendertest.. It may be some other anomaly that I haven't accounted for .. I said 6% earlier but I rechecked my my figures and it is more like 20 .. Not insignificant.

Perhaps I can get the same or better results by turning preview off ..

I will experiment with some other source media and see what the results are.

I try not to spend too much time ' testing ' for test sake. One can really get bogged down with that ..

It seems logical to raise Preview Ram and get "better" performance. On the other hand, I've never read a post that noted improved performance by raising that number...not a one.

On the other hand, I've seen a good number of posts where the cause for problems of various kinds were the result of having the number too high.

Edward, if you read this, will you clarify this? Under what cirucumstances is there improvement in preview by raising the Preview Ram?

That's kind of an elusive number. When rendering, it's honestly better to have the number low. However, when editing, it CAN be useful to have it higher. So... exactly what does that number do?

Take a look at your menu options - specifically Tools - Build Dynamic RAM Preview. This menu option will let you render a section of video to RAM to increase preview quality/speed. Once rendered to RAM, it will play at full resolution/frame rate. That can be very helpful when seeing how a transition or effect will really look. So, how long of a clip can you RAM render?

That's where that RAM number comes in handy. The bigger the number, the longer the segment that can be rendered to RAM. So if you're trying to preview a 10 second segment and it will only RAM render 5 seconds worth, the number needs to be twice as big. There are times when I will up that number to 1 Gig. But... I ALWAYS bring it back down to a lower value when I'm finished with that previewing.

I believe this number is also used some during rendering. I haven't tested in the current version but a few versions back I rendered a series of stills with NO movement. With the number set to a small value, the sequence rendered extremely quickly. Basically, it rendered the first frame and used that for each subsequent frame until the frame actually changed. Setting it to ZERO slowed the render down tremendously as it seemed to render EVERY frame instead of caching the already rendered frame when there were no changes.

So... does that help?

Your answer shows I knew nothing about this setting...well, generally speaking leaving at default value seems a safe bet based on what you say...

Jeff, On V10 you can specify how you want to process renders for "Sony AVC", which I use frequently. In the render window choose "Custom" and at the bottom of the screen select "Encode Mode" and then select , Automatic, GPU, or CPU only.

Hey Jeff,

I have seen the error of my ways, and I'm grateful that you inspired me to take a closer look.

I determined last night that the Rendertest put together by John Cline over at the Sony Forums seems to do best, when rendering the 1080i generated media to .m2t when preview RAM is set to ~1g .. Anything higher does not help . About 20% faster than the default 350mb on my rig.
With hindsight, I have no idea why, it is not useful, and even detrimental for my purposes..

However, when I experimented with rendering .avi and .m2t to .mpg, increasing the preview RAM beyond the default, slowed things down noticeably, maybe 10 - 20 % ..

I was under the mistaken impression that ' preview RAM ' had something to do with the preview displayed while rendering, so thanks Edward for helping to clear this up..


Further experimenting showed that even closing the preview window had no effect on my rendering times.

It looks like, that for my usual work flow, setting preview RAM to ' 0 ', might even be my best bet ...

Excellent, it is amazing how many people that setting causes issue for!

I really can't recommend going all the way to zero but small values have certainly worked for me. I only raise the number when I need to increase the length of RAM preview. Then I lower it back down to a small number. However, if zero works for you... works for me.

Thanks Roger, good information to know.