GK104 fake performance benchmark slides

X

xoleras

2[H]4U
Joined
Oct 11, 2011
Messages
3,551
600x437px-LL-4ede5d73_kepler4.png


Interesting because 8x MSAA is not an option in bf3.

Another fake slide, another day. I want real benchmarks.
 
does it need another thread :p
but yea I'm not believin NUTTIN till the official word or better previews are out
 
Uhm, you can force 8xAA with the nvidia control panel.

And even if it is 4xAA, it still performs quite nicely. Infact, those numbers would put it very slightly behind my 480 SLI.
 
Override AA does not work in bf3 with the last set of drivers I used before selling 580s, that was 285.79? I think. Further, using override AA in a dx11 title that includes all of the IQ options that bf3 is the height of stupidity, and very unlikely.

I look forward to the [H] review of these cards. If the rumors are true the card will certainly be impressive but this source doesn't look legitimate at all.
 
xoleras said:
[*img]http://cdn.overclock.net/4/4e/600x437px-LL-4ede5d73_kepler4.png[/img]

Interesting because 8x MSAA is not an option in bf3.

Another fake slide, another day. I want real benchmarks.
Click to expand...

xoleras said:
Same here. Performance rumored will be impressive if true but i'm sick of the fake slides.
Click to expand...



xoleras said:
Override AA does not work in bf3 with the last set of drivers I used before selling 580s, that was 285.79? I think. Further, using override AA in a dx11 title that includes all of the IQ options that bf3 is the height of stupidity, and very unlikely.

I look forward to the [H] review of these cards. If the rumors are true the card will certainly be impressive but this source doesn't look legitimate at all.
Click to expand...

Other reviews have shown 8x AA in BF3. HKE PC is a long-standing, legitimate Chinese review site, and they clearly have the card as well as testing of it done.

P.S. The only thing wrong with that slide is the drivers were older per HWC reviewer, so the performance will be higher on GTX680 than it shows once launch-day reviews hit.
 
theres another picture with the bars fixed. its like 1fps or less slower lol
 
xoleras said:
Same here. Performance rumored will be impressive if true but i'm sick of the fake slides.
Click to expand...

How exactly is it impressive though? It appears merely even to the 7970, and it's late. We've been used to Nvidia having a 15-20% advantage on AMD the last generations, so I'd say the rumored performance is actually rather disappointing.
 
xoleras said:
Override AA does not work in bf3 with the last set of drivers I used before selling 580s, that was 285.79? I think. Further, using override AA in a dx11 title that includes all of the IQ options that bf3 is the height of stupidity, and very unlikely.

I look forward to the [H] review of these cards. If the rumors are true the card will certainly be impressive but this source doesn't look legitimate at all.
Click to expand...

285 is the stone age. I've been on 295 since jan.

and forcing AA just shows performance. Height of stupidity for utility, yes. But for performance testing, not at all.
 
Raap said:
How exactly is it impressive though? It appears merely even to the 7970, and it's late. We've been used to Nvidia having a 15-20% advantage on AMD the last generations, so I'd say the rumored performance is actually rather disappointing.
Click to expand...

If it can match or beat the 7970 while also using less power, that would be quite a change from previous generations where Nvidia needed a bigger, more power-hungry chip to beat them. I think 10% faster at stock speeds is still possible, so they'd be right where they normally are.
 
What ever happened to the performance that could match 3 GTX 580 from the sameratine video at GDC? If this has 3 times as many rendering pipes as the 580s then why is its performance so low? I mean a 200% increase in the number of pipes and yet were seeing only a 25% performance increase here. Not to mention the that their road map stated that kepler was supposed to be 2x faster than Firmi, and then the next gen was supposed to be 8x faster than firmi.

http://www.maximumpc.com/article/home/nvidia_unveils_gpu_roadmap

I know this is for the CUDA performance scale, but even still the CUDA performance should also reflect on the GFX performance should it not?

Edit: Ok after reading more into the article the CUDA performance is supposed to be 16 faster with Maxwell and the GFX performance is supposed to be 3x faster on Maxwell not Kepler, but still the 200% pipe line increase seems underwhelming on this chart
 
Last edited:
genetech said:
What ever happened to the performance that could match 3 GTX 580 from the sameratine video at GDC? If this has 3 times as many rendering pipes as the 580s then why is its performance so low? I mean a 200% increase in the number of pipes and yet were seeing only a 25% performance increase here. Not to mention the that their road map stated that kepler was supposed to be 2x faster than Firmi, and then the next gen was supposed to be 8x faster than firmi.

http://www.maximumpc.com/article/home/nvidia_unveils_gpu_roadmap

I know this is for the CUDA performance scale, but even still the CUDA performance should also reflect on the GFX performance should it not?

Edit: Ok after reading more into the article the CUDA performance is supposed to be 16 faster with Maxwell and the GFX performance is supposed to be 3x faster on Maxwell not Kepler, but still the 200% pipe line increase seems underwhelming on this chart
Click to expand...

Try doing some research. These CUDA cores aren't the same as Fermi cores. It's a different architecture.

People should start doing some research instead of doubting things without fully understanding the context of the technology being discussed.
 
genetech said:
What ever happened to the performance that could match 3 GTX 580 from the sameratine video at GDC? If this has 3 times as many rendering pipes as the 580s then why is its performance so low? I mean a 200% increase in the number of pipes and yet were seeing only a 25% performance increase here. Not to mention the that their road map stated that kepler was supposed to be 2x faster than Firmi, and then the next gen was supposed to be 8x faster than firmi.

http://www.maximumpc.com/article/home/nvidia_unveils_gpu_roadmap

I know this is for the CUDA performance scale, but even still the CUDA performance should also reflect on the GFX performance should it not?

Edit: Ok after reading more into the article the CUDA performance is supposed to be 16 faster with Maxwell and the GFX performance is supposed to be 3x faster on Maxwell not Kepler, but still the 200% pipe line increase seems underwhelming on this chart
Click to expand...

nvidia wants you to think that the 680 has 3x the performance of the 580.

1. the demo was run the msaa on the 580's and fxaa (or the new fxaa version) on the 680.

2. the code has been optimized to run more efficient.
 
Why do people post fake slides? What's the point? to sabotage the product prior to release or create hype?
 
Forceman said:
If it can match or beat the 7970 while also using less power, that would be quite a change from previous generations where Nvidia needed a bigger, more power-hungry chip to beat them. I think 10% faster at stock speeds is still possible, so they'd be right where they normally are.
Click to expand...

Im not biting off on the power usage numbers just yet, while I don't believe these will be as power hungry as the Fermi cards are. I have a feeling the reason we are seeing this come up a lot is due to the "Wow" factor AMD has been getting from their zero core, so to get better hits they include that other than just faked performance slides.

I'll wait to see if they finally one up AMD on power consumption.
 
If you were sick of them, you wouldn't bother posting and linking a fake. Trollin' for views here, buddy.
 
matrix2000x2 said:
Why do people post fake slides? What's the point? to sabotage the product prior to release or create hype?
Click to expand...

People (we'll call them group 1) like to start shit and/or point the lime light at themselves. It's why Facebook is so popular.

Also, we (apparently) have too many bored house wives on this forum that like to gossip about fantasy truths and marketing spin. This statement is only reinforced when they start bickering amongst each other about said fictional/fantasy facts. The house wives only end up feeding the narcissists in group 1.

It's been said before, but... None of this means dick squat until legit reviews come out (personally waiting for the [H] review myself).
 
You must log in or register to reply here.
Back
Top