However, I solved one mystery why users get such different rendering times and sometimes no speed advantage with HA! I did my own simple test: Test file was a 2:00min .MTS video composed of a few little clips without any effects applied. I rendered in H.264 AVC 1920x1080 16Mb/s both with and without HA and got 02:02min (no HA) & 01:03min (HA) which shows the GPU was working well. I took screenshots of Windows Resource Monitor (file attached). In the first case the CPU is working 100%, in the second a little less with the GPU halving rendering time!
However, the situation changed entirely when I applied heavy editing, in this case reversing every clip in the video. I then got rendering times of 06:30min (no HA) & 06:22min(HA)! This shows two things: editing can increase rendering times by several 100%! The fact that HA did not have any effect can be explained by looking at the CPU activity - it was intermittent in both cases and I also noticed a high disk activity. So the more effects are applied (especially ones that affect an entire clip) the HDD becomes the limiting factor, so much so that the GPU has literally no effect. It also means that a very fast HDD is equally important as the GPU - the system needs to be well balanced.
Before I spend big time on new hardware and software - does anybody know if PD 15 works well with the latest Nvidia GPUs using the Pascal architecture?
|
||
Filename | PD Rendering Performance.jpg |
![]() |
Description | Rendering performance with various factors |
|
Filesize |
687 Kbytes
|
|
Downloaded: | 37 time(s) |
MSI GT680 with Intel Core i7 2630QM @ 2.00GHz - Win 7 64bit - RAM 16Gb - Geforce GTX460M 1.5Gb (driver v337.88 ) - SSD 111Gb (for system & apps) - HDD 750Gb 7,200rpm 3Gb/s