GPU Benchmarking... 580 > 680?

Because the chart you linked to is for Price Performance. Which is to say; performance for the dollar value. Note that the 680 costs considerably more :)
 
No, they can list by price performance, the chart is by PERFORMANCE, not by price vs performance....

It defaults to the G3D Mark scores.

Otherwise the Quadro 4000 would be at the bottom of the list, priced at 6000 dollars
 
GPGPU is horrible on GTX 680....

But gaming wise it's the opposite.
 
No idea how the cards compare in Passmark, but it obviously likes the 580 more than the 680, and the 570 more than the 590, so who knows. It's definitely wrong on a number of counts if going by gaming performance.
 
this might explain more:

45193.png


45166.png


http://www.anandtech.com/show/5699/nvidia-geforce-gtx-680-review/17
 
Raap said:
No idea how the cards compare in Passmark, but it obviously likes the 580 more than the 680, and the 570 more than the 590, so who knows. It's definitely wrong on a number of counts if going by gaming performance.
Click to expand...

This is dollar to performance, not absolute performance. Kind of like the best bang for your buck.
 
sdlvx said:
This is dollar to performance, not absolute performance. Kind of like the best bang for your buck.
Click to expand...

It's not. If you notice there's a link there to view it as price/performance, and that is a completely different result across the board. This is their pure performance chart.

My personal opinion would just be because a passmark bench really means nothing and gives stupid results sometimes.
 
Kainzo said:
No, they can list by price performance, the chart is by PERFORMANCE, not by price vs performance....

It defaults to the G3D Mark scores.

Otherwise the Quadro 4000 would be at the bottom of the list, priced at 6000 dollars
Click to expand...

Yep, my bad :(

I read the link at page top for Price Performance and figured it had to be on that since the scores were goofy. As others had indicated, probably has more to do with passmark not being a particularly good measure of performance in reality.
 
cortexodus said:
...probably has more to do with passmark not being a particularly good measure of performance in reality.
Click to expand...
Depends on what you're using it for in reality. If you're doing AES encryption, it looks like PassMark is pretty much spot-on. It's certainly not a gaming benchmark, though.
 
LuminaryJanitor said:
Depends on what you're using it for in reality. If you're doing AES encryption, it looks like PassMark is pretty much spot-on. It's certainly not a gaming benchmark, though.
Click to expand...

+1. The test they use are artificial and do not perfect translate into gaming performance. The best way to compare GPU for gaming performance is FPS.

http://www.anandtech.com/bench/Product/555?vs=517

Here you will see the 680gtx out performs the 580gtx. Sometimes there performance is close at 1080p, but at higher resolutions the 680gtx pulls away.
 
Blkout said:
Uh yea...literally a ton of people, given the average weight of a human is about 120-180lbs, I can buy that.
Click to expand...

You must be european. Americans blow that range out of the water.

Passmark is flawed as it doesn't support SLI and the tests are very synthetic.
 
doug_7506 said:
+1. The test they use are artificial and do not perfect translate into gaming performance. The best way to compare GPU for gaming performance is FPS.

http://www.anandtech.com/bench/Product/555?vs=517

Here you will see the 680gtx out performs the 580gtx. Sometimes there performance is close at 1080p, but at higher resolutions the 680gtx pulls away.
Click to expand...
This is very helpful information. Comparisons based on actual fps in games are always better than artificial benchmarks. I can't play a benchmark.

It looks like I'll be stuck with shortscreen (1080 lines instead of the 1200 I currently enjoy on a CRT) if I want a 120 Hz LCD anytime soon. So, no real reason to get a 680 instead of a 580. The $499+ price of the 680 is a factor too.
 
With a synthetic benchmark you get synthetic results. That's all there is to it. I'd take real world performance over synthetic any day.
 
If running a synthetic test and getting a lower score at the end makes you sad

Get a GTX 580

If you want the best gaming experience around

Get a GTX 680

/thread ;)
 
Objekt said:
This is very helpful information. Comparisons based on actual fps in games are always better than artificial benchmarks. I can't play a benchmark.

It looks like I'll be stuck with shortscreen (1080 lines instead of the 1200 I currently enjoy on a CRT) if I want a 120 Hz LCD anytime soon. So, no real reason to get a 680 instead of a 580. The $499+ price of the 680 is a factor too.
Click to expand...

Pretty much so. Out of the games they tested, unless you have a 120hz monitor, you would not notice a difference between the 580 and 680 at 1680x1050. Do note that their test list is missing some big titles though.
 
Oh man, that sucks... I play that passmark game like 12 hours a day. Oh wait...
 
Doug makes a valid point, why upgrade to a 680 from 580 if there is NO performance increase for your monitor/setup?
 
Last edited:
Jorona said:
You must be european. Americans blow that range out of the water.

Passmark is flawed as it doesn't support SLI and the tests are very synthetic.
Click to expand...

I'm American. It was just an average, not sure why you felt the need to be so technical about it. You missed the point completely if the average weight of a human was your focus.
 
So does this mean companies should focus on the weight of a person for sales research?

Everyone run.... this baby is derailing!
 
That's the thing... from the FPS bencharmks at 1920x1080 - the 580 vs 680 are about the same....
 
Kainzo said:
That's the thing... from the FPS bencharmks at 1920x1080 - the 580 vs 680 are about the same....
Click to expand...

I haven't seen the FPS at 1080 between the 680 and 580 for BF3. If BF3 is what you are going to play a lot. You might want to try to find some. But if BF3 holds like the others, it shouldn't be much of a difference.
 
You must log in or register to reply here.
Back
Top