Unfortunately, since it's done on two different machines, it's not 100% correct, but we get the idea.

 

First, there is a FSXMark11 comparison.

2500K @ 4.7 + GTX680 resulted in 43.7FPS.

This is actually quite poor for a new crown king card, because...

 

A comparable system:

2600K with HT off @ 4.7 + GTX580 results in 47FPS, while other components are quite the same

 

At first I thought, nah, probably something wrong, there is another user having:

2500K @ 4.6 + GTX480 resulting in 43.1FPS

 

So wow, apparently this new card is actually worse than GTX580? Since these are all synthetic tests, they all do show one thing. I hope we can get more 680s flowing in the measure better.

 

Another test was also done in the IQ area. After lots of work of consolidating systems and setting everything to the same levels, John (JohnFromCO) was extremely patient to go with me though series of tests on my cloudtest to measure the FPS when using high levels of AA.

I finally concluded that both GTX580 and GTX680 perform quite the same when it comes to Antialiasing and FSX.

The two screenshots show vanilla FSX with REX setup and FPS measurement. Both are maintaining 33fps in the completely same setup. FSX.cfg is the same, REX settings are the same, resolution is the same... everything I could think of.

 

Conclusion:

If you have a 580, going 680 is possibly a downgrade. 580 is still rock solid with FSX.

 

A big thanks to John!!

 

Here are two screenshots, from each system:

 

Word Not Allowed

 

John

thanks so much guys for this

Word Not Allowed, keep in mind that the GTX 680 driver(s) are not mature, it will take a couple more to fine tune the card.

 

Also, this card is made for PCI-e 3.0, so until we can test it on a mobo + CPU fully using the 3.0 capability + mature drivers it's kinda unfair to compare both card, almost forgot....the 680 has less core (bad for FSX), the GTX 685 will be a better card against the GTX 580....I'll be waiting for your next testing.

 

Stay safe flying...

Alain, all that is known, thanks.

 

The intention of this test was to put a card under same conditions to test. Testing PCIE3 would be unfair. I wanted to know how much, if any, is this card better for FSX. Excluding PCIE3 variable.

I kinda expected those results.

Interesting stuff. I wonder why this is (excluding the drivers issue that Alain raises) When the 680 costs less than a small country I will buy one. I am going to run the cloud test tonight when I get in..

 

jake

The 680 is now closer in design philosophy to an ATI card than it is to an traditional NVidia card, we already know ATI cards perform badly compared to NV in FSX, so this news is exactly what I was expecting. Don't expect mature drivers to make much difference either.

Yeah, newer drivers never pushed the performance (in FSX) up high. Not talking about another games.

The 680 is now closer in design philosophy to an ATI card than it is to an traditional NVidia card, we already know ATI cards perform badly compared to NV in FSX, so this news is exactly what I was expecting. Don't expect mature drivers to make much difference either.

 

Very good point. I never understood how that affects performance in FSX or how much, but these 600 come with a lower shader clock (it doesn't run at 2x the core clock anymore) and a crap load of SP's (like 3 times as many as the 580) to make up for the lower freq. Then there's the ROP's count... I don't know. I'm going with a 580 in the end that I found 2nd hand for just 280€, so these 680 - AA tests are very much appreciated guys.

The 685 or whatever they brand it should be better in every aspect than Fermi, but if this design based on increasing the number of shader processors and lowering the clocks is a factor, it might still have subpar performance in FSX. I don't think that should be the case, cause after all a 512 SP 580 performs much better than a let's say 216 SP GTX260 in clouds + SGSS

Now 280€ for the 580 is one sweet deal Dario!

Now 280€ for the 580 is one sweet deal Dario!

 

It's a reference one though, we'll see how noisy and hot it runs once I crank up the clocks lol

Srdan,

 

Thanks for this first of all. But please note that the GTX 680 was tested with the i5 2500K and the GTX 580 was tested with the i7 2600K. NickN said that the extra cache of the 2600K is a little beneficial for FSX. Maybe it's that?

 

Also, do you know if there's a difference in smoothness with the GTX 680? Or does it microstutter more in FSX?

Smoothness:

No, there is no way to compare that, other than me buying the 680 and testing. All other tests are just a perception thing.

Considering that smoothness didn't get better over couple of last generations of cards, only that FSX was faster, I don't believe that's gonna be the case here.

 

Cache:

Possibly and probably. Still, the IQ test is pretty discouraging for the 200€ more expensive card.

But as many said, was quite expected.

It's a reference one though, we'll see how noisy and hot it runs once I crank up the clocks lol

 

Unless you are getting a dual fan 3rd party heat-sink for the 580, the blower style fan cooler runs hot on the 580. When I crank mine up to 900MHZ, my temps will end up in the high 70's while in heavy clouds... Although, if I want, I can crank up the fan and send my temps into the mid 50's.

Unless you are getting a dual fan 3rd party heat-sink for the 580, the blower style fan cooler runs hot on the 580. When I crank mine up to 900MHZ, my temps will end up in the high 70's while in heavy clouds... Although, if I want, I can crank up the fan and send my temps into the mid 50's.

 

Well, that's actually quite good. Coming from a 480 overclocked to 860MHz, spite of the Windforce 3 it wasn't unusual to see it in the mid 80's celsius in cloudy days, and that's with a custom fan profile that had them running at 90-100% speed.

The turbine Zotac 480 I had was absolutely awful. Woke up the neighbours at 820MHz when the fan was at full speed, haha

Hahaha, a blowdryer. Man, I do miss the noises sometimes :LMAO: