CyberLink Community Forum
where the experts meet
| Advanced Search >
PD17 & Nvidia drivers
Reply to this topic
optodata
Senior Contributor Private Message Location: California, USA Joined: Sep 16, 2011 16:04 Messages: 3859 Online
[Post New]
Quote If you really want to compare encode technologies, one needs a basic transcode, not a full edited project, just too many more variables potentially at play.

Jeff

No one is stopping you from doing a full test. I look forward to your results!
Reply
Warry [Avatar]
Member Private Message Location: The Netherlands Joined: Oct 13, 2014 11:42 Messages: 134 Offline
[Post New]
Quote Warry, That potentially is a very misleading statement. The encode performance of a particular project is very much tied to the features incorporated in that project and hence not a good reflection on various encoding technologies as standalones.

Thanks Jeff for your reaction, and I can of course to a large extend agree with what you write.
What I try to keep in mind is that in this forum we are discussing these technologies whilst using PowerDirector that executes previews and productions with these technologies. We have seen that ALL the parameters,like CPU, GPU, diskdrives, drivers, settings, the PD version, and also the specs of the source video material, the changes we apply to that material, etc. etc. have a certain influence on the speed, quality and swiftness of the software. The Optodata tests results differ per driver!
A full test to compare the encoding and decoding capabilities of the technologies would be very interesting. The outcome may likely be (as you rightfully write) a gain of 50% to transform a video from A to B. But what if we apply effects and all the other things we need to get the video we want? And what is the advice to people who want to know what kind of computer they need for video editing?
My conclusion would be, that GPUs make a difference for video editing. That depending on the equipment used and the "effects" applied, in general there might be a speed gain of about 10-20% as the Optodata tests show, and there may be peaks as high as 50% as your tests show.
Hope that this makes sense.
Reply
JL_JL [Avatar]
Senior Contributor Private Message Location: Arizona, USA Joined: Oct 01, 2006 20:01 Messages: 3819 Offline
[Post New]
Quote No one is stopping you from doing a full test. I look forward to your results!
Optodata, I've got no need to do a test right now, but surely will as I've done in the past when CL releases something of interest or a real curiosity need arises. Someone claiming 7-8x by changing Nvidia drivers in PD is not a need, as I said, 100% impossible statement from driver features perspective.

The fact it's passed on and a test conducted to disprove, let’s just say, is interesting approach for advancing PD editing experiences.

Quote But what if we apply effects and all the other things we need to get the video we want? And what is the advice to people who want to know what kind of computer they need for video editing?

Warry, That’s exactly why it’s important to understand what the pieces do vs the whole. If one learns the effect of each part, they can then make their own informed choice on what’s important for their editing needs. For example, someone that wants to stabilize every clip better spend every dollar on a CPU. His GPU will do nothing to remove that stabilize pinch point in timeline editing or production. Conversely, someone that drops their video in the timeline and simply converts to HEVC because they feel a quality boost would benefit greatly from GPU, not just 10-20 or even 50%. See here: https://forum.cyberlink.com/forum/posts/list/52497.page#post_box_288574 a 10x change, nothing out of the ordinary here, basic CPU encoding on a lower end CPU to NVENC of a GTX960. No CPU around would match that performance in PD.

Don't get me wrong, nothing wrong with optodata tests, it represented events seen for his project. My point was simply that it's potentially a very misleading statement to generalize from that test that there might be a speed gain of about 10-20% or peaks of 50%, it really depends on intended use.

Jeff
Reply
Warry [Avatar]
Member Private Message Location: The Netherlands Joined: Oct 13, 2014 11:42 Messages: 134 Offline
[Post New]
OK. Thanks Jeff, "it really depends on intended use", I think that we are in agreement. In the meanwhile, Harmen, who started this discussion has finished his project. I guess that the performance and speed of production will keep us all busy from time to time.
Take care!
Reply
Reply to this topic
Powered by JForum 2.1.8 © JForum Team