01Aug Need to decrease your rendering times in CS6?
I’ve been using a Nvidia GTX 285 2GB card for rendering since CS5, but it’s always interesting to see what newer, more powerful cards can do. For larger projects it can sometimes take the GTX 285 over an hour to render a complex timeline, but how much would it be worth to have that same timeline rendered in 15 min?
This chart represents several hundred user submitted test results from commonly supported and hacked graphics cards (see the full set of test results here). As you can see from the rendering speed chart, my GTX 285 is about 4 times slower then the top rated GTX 680. The surprising part of the chart is how low on the list the Quadro 4000 falls. It seems like a $650 graphics card like to Quadro should out perform the $450 GTX 680, even the GTX 580 which runs around $270 seems to provide better overall performance.
Gains in playback performance are a little less clear. It seems as though the GTX 570 gives the best results here, probably due to the 320-bit memory bus v.s. the 256-bit bus used on the GTX 680. I would guess the number of cores has the strongest effect on rendering time, and the memory bus has the strongest effect on playback, but I’d be interested to know if there are other reasons. No matter what card you choose, it appears all of them will help make timeline playback better.
Just searching ebay, it seems as though the GTX 570 would be the best bang for your buck. With used prices well under $200 it has a rendering speed that’s twice as fast as my GTX 285 and the boost in playback performance would probably be more noticeable when editing. But if the most important item on your list is rendering speeds the GTX 680 looks like the clear winner.
If you’d like to see how well your current graphics card performs, check out the Benchmark instructions for the files and timelines used for the test. I’ll have to wait and see how well my GTX 285 handles the Catleap IPS Display if it chocks I might be in the market for a used GTX 570 2.5GB GPU in the near future.
August 1st, 2012 at 12:32 pm
DeeJay, thanks for the article
I take it you are on a PC?
It’s trickier for people like me on a MacPro.
August 1st, 2012 at 1:40 pm
I believe the GTX 285, and GTX 570 are both reported as working on Mac towers. You’ll have to use the gpusniffer hack for the GTX 570 to work on a Macpro and some people have gone as far as flashing the 570 with a new firmware. But it can be done if you’re willing to put the risk/work.
August 1st, 2012 at 2:59 pm
Yes, I have read a lot about it. There is an offical Mac version of the GTX 285 (is that what you have?). You can get a PC GTX 470 and a PC GTX 570 in 10.7.4 with no EFI bootscreen. There are details on Netkas.org.
The ebay seller MacVidCards sells cards that will work with his instructions, even sells a 470 to work in 10.6.8. He is putting a custom ROM on them that allow it to run at PCI 2.0 speeds, otherwise a PC card (say a GTX 570) in a MacPro with hacks will run at PCI 1.0 speeds he says.
http://stores.ebay.com/macvidcards?_trksid=p4340.l2563
He Dave Pirelli on creative cow and Rominator on Netkas. He leave a lot of good tips, and offers premium hacked cards at premium prices.
A stock PC GTX 470 story:
http://forums.creativecow.net/thread/277/9638
I’d say: USE AT YOUR OWN RISK
August 1st, 2012 at 6:03 pm
I’m a PC user so I haven’t had to deal with any of that, but people I know have told me stories of the hassle they’ve had with graphics cards. It sounds worse then running linux used to be. A few of them have switched over to hackintosh systems with custom drivers and claim that they’ve worked out most of the bugs. I think I’ll stick to PC, less money more options.
August 2nd, 2012 at 8:10 am
Also I would worry about over taxing the power supply in a MacPro
August 2nd, 2012 at 1:46 pm
Can the MacPro’s power supply be upgraded?
October 15th, 2012 at 11:30 am
I’m not sure, but I think (like me) you might’ve interpreted the second chart wrong. On http://ppbm5.com/MPE%20Charts.php the subtitle of the second chart explains what the gain is of using MPE with the particular card over software rendering only. It seems like you interpreted it as render time vs. playback smoothness during editing. Basically the second chart doesn’t really mean very much since everyone will probably use the MPE hardware acceleration anyway. It’s all about the first chart: how fast does it render with MPE enabled. There aren’t really any figures on how it performs during editing playback, but my guess is that it’s closely related to render time.
October 23rd, 2012 at 8:41 pm
Thanks for the heads up!