GPU Selection by GPU Benchmark Scores...

gnu_B wrote on 1/25/2014, 8:21 PM

I have MSP 12 and a GTX 460 for my GPU and it works OK. I render to MP4 and I use the correct codec and settings.
Now I'm looking to getting something to better assist with rendering. It looks like the guys at videocardbenchmark.net have already done some of the work for me with their video card performance rankings.
Here are cards I am considering and their Benchmark Scores:
GTX 460 (benchmark score 2667) * my current card *
GTX 560Ti (Benchmark Score 3536)
GTX570 (Benchmark Score 4383)
GTX 580 (Benchmark Score 4965)
GTX 650 Ti (Benchmark Score 2700)
GTX 660 (Benchmark Score 4110)
GTX 670 (Benchmark Score 5382)

I'm interested in comments as to whether using GPU benchmark scores is a realistic way to assess expected GPU performance when rendering with MSP.

Also, if any of the cards on my list are confirmed problem candidates, please let me know.

Thanks.

-gnu_B

Comments

ScottishBattleAxe wrote on 1/29/2014, 7:54 AM

I'd like to know is my 780 Ti will be utilized by MS13, because MS12 doesn't use it (I did select settings to tell it to)

musicvid10 wrote on 1/29/2014, 10:00 AM

If you have a reasonably fast cpu, the differences will be so minor you really shouldn't bother. There is also a quality hit using gpu.
Use your gpu for gaming.

ScottishBattleAxe wrote on 1/29/2014, 2:17 PM

I do have a 4770K @ 4.2GHz, so it's pretty quick.. I was under the assumption that GPU encoding would be faster, without any change in quality. What would degrade the quality?

musicvid10 wrote on 1/29/2014, 3:39 PM

Hardware acceleration.
Whether you notice it or not is an individual matter.

MSmart wrote on 1/29/2014, 7:41 PM

Hey Scottish, welcome to the forum.

The Search utility on this forum is great. See link at top. Do a "gpu quality" search using the Vegas Pro forum and Last 2 Years to find many good posts on the subject.

Here's one such topic:

http://www.sonycreativesoftware.com/forums/ShowMessage.asp?ForumID=4&MessageID=877280

gnu_B wrote on 1/30/2014, 10:37 AM

My original question pertained to assessing the rendering speed potential of GPUs by comparing GPU Benchmark scores.

I do appreciate the thread's drift in the direction of GPU rendering vs quality, I have learned a lot from the brief conversation.

However, I am using MSP to render video for a regularly-occurring TV show that goes out in HD. Time constraints are such that faster rendering is critical to my editing, because only so much editing time is available to me to produce a show. If I opt for finer image quality and slower rendering, it means that some important stories will simply not get covered.

So, I would like to restate my question:
Are published GPU Benchmark scores a reliable way to assess performance for a particular video card?

Sincerely,

-gnu_B

musicvid10 wrote on 1/30/2014, 12:33 PM

You would have to compare rendering times for yourself, unless you can find benchmark scores SPECIFICALLY for a particular video encoder on an IDENTICAL system as yours.

Most GPU benchmarks published are for GAMING, which has nothing whatsoever to do with video encoding. Two different planets.

That being said, refer to my first response. It's probably simply not worth it on an able system. Would the difference between a 2% and a 5% advantage really be that important? A good CPU would be a much better investment.