• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

GeForce GTX 780 Ti benchmarks leak

Seems like they're really designing those new cards for high resolutions, huh? I can deal with 1080p, so I guess I won't have to pick the 880/980 when I upgrade again, the 870/970 will probably be enough.
 
Is there anything between 'quiet mode' and 'uber mode' for a 290X? If not, it seems to me that 'quiet mode' is just a clever misnomer, labelled so in order to make it look better if somebody chose to test against it.
 
Wait a minute, why is the R9 290X running in Quite Mode? That 8-12% loss is important. Also, I have a feeling that the R9 290X with custom coolers will perform better than how they are now (correct me on this if I am wrong).These benchmarks make no sense =/
 
Wait a minute, why is the R9 290X running in Quite Mode? That 8-12% loss is important. Also, I have a feeling that the R9 290X with custom coolers will perform better than how they are now (correct me on this if I am wrong).These benchmarks make no sense =/
Pls understand.
 
Wait a minute, why is the R9 290X running in Quite Mode? That 8-12% loss is important. Also, I have a feeling that the R9 290X with custom coolers will perform better than how they are now (correct me on this if I am wrong).These benchmarks make no sense =/

There really is little difference in performance between Quiet en Uber mode, and by default it is on Quiet mode since it otherwise gets quite loud. It will probably indeed get a lot better with custom coolers though.
 
Wait a minute, why is the R9 290X running in Quite Mode? That 8-12% loss is important. Also, I have a feeling that the R9 290X with custom coolers will perform better than how they are now (correct me on this if I am wrong).These benchmarks make no sense =/

I can see a 30% increase in at least two of the games on the chart shown.

Also, custom coolers don't allow for a card to "perform better", they allow the card to run at cooler temperatures and not sound like a Space Shuttle. This leaves potential for overclocking, however.

And seeing as my 780 benchmarked higher than some of the 290X owners running their card in Uber Mode, I doubt the benchmarks would look much different for it here.

sk3tch said:
R9 290X ("Uber" switch on, no OC), i5 2500K @ 4.6 GHz.

00000.png


And note that my 780 was running at 1245MHz but only 60% fan and 70C. Still quieter and cooler than the "Uber Mode" benches from sk3tch.
 
Haha, wow. What's next? Justifications for the $999 prices, wait we got those already! :D
Can you provide an argument as to why cards shouldn't be benchmarked at comparable noise levels?

Is there anything between 'quiet mode' and 'uber mode' for a 290X? If not, it seems to me that 'quiet mode' is just a clever misnomer, labelled so in order to make it look better if somebody chose to test against it.
"Quiet Mode" is still louder than a stock 780, FWIW.
 
Can you provide an argument as to why cards shouldn't be benchmarked at comparable noise levels?

"Quiet Mode" is still louder than a stock 780, FWIW.

Well aren't they comparing the performance here? If they wanted to compare the noise levels than yeah that's fine but this isn't really a good comparison.
 
Well aren't they comparing the powers here? If they wanted to compare the noise levels than yeah that's fine.

Yeah, it's silly to compare the potential power of the cards by benchmarking them based on noise levels. I don't think that's a solid metric by any means unless specifically comparing them based on noise.
 
Did you champion this logic during the 280/480 era?
No, I just laughed at NVidia. But you should have chosen the FX5800 era -- the original leaf blower. It was doubly hilarious set up against some of the best cards ATI ever made. I have no idea how NV ever expected to sell those.

Well aren't they comparing the performance here? If they wanted to compare the noise levels than yeah that's fine but this isn't really a good comparison.
I think for people considering a product it's useful to see how it performs within a noise envelope they actually want to use. Ideally, you'd benchmark all cards at all profiles, and provide noise data for each profile (like computerbase does). Everyone can then decide how to weigh the multi-objective optimization problem between noise and performance.
 
I DON'T KNOW WHAT TO BLOW MY MONEY ON ANYMORE
<- G-Sync :P

No, I just laughed at NVidia. But you should have chosen the FX5800 era -- the original leaf blower. It was doubly hilarious set up against some of the best cards ATI ever made. I have no idea how NV ever expected to sell those.
But thats more than 10 years ago, the examples I gave are relatively recent.
 
Can you provide an argument as to why cards shouldn't be benchmarked at comparable noise levels?

Comparing top GPUs which were created for best performance in noise level is like comparing fastest cars by their fuel consumption.

It is subject for comparison but most of people don't give a flying fuck about it same as Power consumption.
 
And seeing as my 780 benchmarked higher than some of the 290X owners running their card in Uber Mode, I doubt the benchmarks would look much different for it here.


And note that my 780 was running at 1245MHz but only 60% fan and 70C. Still quieter and cooler than the "Uber Mode" benches from sk3tch.

That particular benchmark favors NVIDIA so it is no surprise that your 780 performed better. Your 780 was heavily OC'd, as well. My 290X was not OC'd. Just "Uber" switch on...which just eliminates a lot of the throttle. You may have been quieter but you were not cooler. :) My cards have never gone over 70 C even in Crossfire.

Either way, it doesn't matter. The 780/Titan/290X trade blows a lot of the times...and when one is beat, it's not by this insane margin. The Ti may change that a bit - but at a $150 price premium over the 290X. If it were $699 for Ti 6GB, I'd be down...but until then it is wait and see.

There really is little difference in performance between Quiet en Uber mode, and by default it is on Quiet mode since it otherwise gets quite loud. It will probably indeed get a lot better with custom coolers though.

Yes, it does - quite a bit, actually. It throttles down a lot more when the fans are not enabled.

http://www.pcper.com/reviews/Graphics-Cards/AMD-Radeon-R9-290X-Hawaii-Configurable-GPU
 
Comparing top GPUs which were created for best performance in noise level is like comparing fastest cars by their fuel consumption.

It is subject for comparison but most of people don't give a flying fuck about it same as Power consumption.

Wouldn't it be more like comparing the exhaust note? Which people do try to change. Fuel consumption would be like comparing energy draw.
 
That particular benchmark favors NVIDIA so it is no surprise that your 780 performed better. Your 780 was heavily OC'd, as well. My 290X was not OC'd. Just "Uber" switch on...which just eliminates a lot of the throttle. You may have been quieter but you were not cooler. :) My cards have never gone over 70 C even in Crossfire.
Are you using watercooling? Most reviews have the 290X reaching 90°C+, both in Uber and Quiet mode.
 
Did you champion this logic during the 280/480 era?
I'm not sure why AMD cheerleaders keep pointing at the GTX 480 to prove that "power-hungry, loud and hot is totally fine and well accepted when Nvidia does it".
Everyone *loathed* the GTX 480 for it, myself included.
It was the time when the common opinion was "Nvidia really dropped the ball with this".
 
Are you using watercooling? Most reviews have the 290X reaching 90°C+, both in Uber and Quiet mode.

No, no. I just have a custom fan profile set - believe me, it sounds like a leaf blower (it's around 70%). The cooler may be "sub optimal" - but it gets the job done.

http://www.guru3d.com/articles_pages/radeon_r9_290x_review_benchmarks,12.html

I game with headphones, so I couldn't care less.

I'm not sure why AMD cheerleaders keep pointing at the GTX 480 to prove that "loud and hot is totally fine and well accepted when Nvidia does it".
Everyone *loathed* the GTX 480 for it, myself included.
It was the time when the common opinion was "Nvidia really dropped the ball with this".

We're at the end of the 28nm lifecycle - so everyone is beating everyone else with hotter and louder cards. So right now, no one can really use this. :)
 
I'm not sure why AMD cheerleaders keep pointing at the GTX 480 to prove that "power-hungry, loud and hot is totally fine and well accepted when Nvidia does it".
Everyone *loathed* the GTX 480 for it, myself included.
It was the time when the common opinion was "Nvidia really dropped the ball with this".
Power-hungry for a top enthusiast card is pointless. Loud and hot are valid complaints, no matter who does it.

The point of discussion here was that no one complained that the 280/480 need to be benchmarked at similar noise levels back then.
 
Power-hungry for a top enthusiast card is pointless. Loud and hot are valid complaints, no matter who does it.

The point of discussion here was that no one complained that the 280/480 need to be benchmarked at similar noise levels back then.

Well, times have changed. The push for more efficient AND powerful GPUs has been on - so AMD merely made their top-end part a bit more configurable by having the two BIOS on it. Previously (with the 7970) there was a switch, as well - but just two identical BIOS. All this shows is AMD's acceptance that enthusiasts want to tinker, whether it be quiet or loud or custom BIOS - they've set their cards up to accept the needs of enthusiasts. I'm not sure why that's a bad thing.

NVIDIA, in comparing their hardware to AMD, does a disservice to all by using "quiet" mode. No enthusiast buys a 290X and leaves that enabled.

At least they call it out in their graphs.
 
Power-hungry for a top enthusiast card is pointless. Loud and hot are valid complaints, no matter who does it.

The point of discussion here was that no one complained that the 280/480 need to be benchmarked at similar noise levels back then.

What forums were you frequenting?
on Guru3d there was a huge uproar about the 400 series heat and noise levels (its heat more than its noise though). 200 series was not as bad from what I remember.

I am not sure anyone would argue otherwise... the 290x is hot and loud and that is surely a detraction against it in comparison to the Nvidia offerings.
 
Damn, getting really hard to wait out for maxwell... I'm still good with what I got, but the impulsive side of me is making this tough for something new.
 
Well, times have changed. The push for more efficient AND powerful GPUs has been on - so AMD merely made their top-end part a bit more configurable by having the two BIOS on it. Previously (with the 7970) there was a switch, as well - but just two identical BIOS. All this shows is AMD's acceptance that enthusiasts want to tinker, whether it be quiet or loud or custom BIOS - they've set their cards up to accept the needs of enthusiasts. I'm not sure why that's a bad thing.
It's not a bad thing at all. You can do something similar in software on NV (and I think earlier AMD?) cards by simply setting the power target and fan profiles way up -- that's what e.g. computerbase do, and they compare both performance and noise at normal and "max" levels. (I know I keep mentioning computerbase, I'm not affiliated in any way with them, it's just that I currently consider their GPU reviews some of the most thorough and professional around -- it's too bad they don't offer official English translations)

As long as everything is correctly labeled, I think that's entirely reasonable. The only thing I consider damaging is some (less thorough/reputable) sites using unlabeled "Uber" numbers in their charts.
 
I'll drop a couple grand on 2 of these cards if the 6gb versions are $999. The 3gb versions would be nice at $650, but I'd feel like that was a step back from my Titans. The most I can get on them is 1202mhz core.
 
I'll drop a couple grand on 2 of these cards if the 6gb versions are $999. The 3gb versions would be nice at $650, but I'd feel like that was a step back from my Titans. The most I can get on them is 1202mhz core.

I'm hoping $800 for 6GB and $900 or $1000 for 12GB. No way it would be a $300 uptick to go from 3GB to 6GB.
 
What forums were you frequenting?
on Guru3d there was a huge uproar about the 400 series heat and noise levels (its heat more than its noise though). 200 series was not as bad from what I remember.

I am not sure anyone would argue otherwise... the 290x is hot and loud and that is surely a detraction against it in comparison to the Nvidia offerings.
Since you didnt notice the point of the discussion, let me repost it again;
The point of discussion here was that no one complained that the 280/480 need to be benchmarked at similar noise levels back then.

People are worn out of $500+ GPUs
I dont think so.
 
Can you provide an argument as to why cards shouldn't be benchmarked at comparable noise levels?
.

Because different people care differently about sound levels, but everybody universally cares about high performance?

If I'm a gamer that uses high quality headphones, I'm not gonna care how much sound my GPU makes. I think it's pretty disingenuous of Nvidia to benchmark against the 290X running in Quiet mode.

Will wait for the HardOCP review.
 
Since you didnt notice the point of the discussion, let me repost it again;



I dont think so.

I read your point of course. I was just mentioning that from my perspective I saw a lot of discussion around the noise and thermals when fermi released... namely on Guru3d. I was not a GAF member then. I think your point is unfounded and maybe just related to what you saw on GAF perhaps.
 
I'll drop a couple grand on 2 of these cards if the 6gb versions are $999. The 3gb versions would be nice at $650, but I'd feel like that was a step back from my Titans. The most I can get on them is 1202mhz core.

Why do that. The 880 is set of 6800 cuda cores and a Q2 Launch, That's 2.5x the speed of a Titan.
 
Why do that. The 880 is set of 6800 cuda cores and a Q2 Launch, That's 2.5x the speed of a Titan.

Any source for this? We've gone from 280 -> 480 -> 580 -> 680 -> 780 with around 15-25% (each generation) improvements in speed. Why would it all of a sudden be 250% faster?
 
Why do that. The 880 is set of 6800 cuda cores and a Q2 Launch, That's 2.5x the speed of a Titan.

Any source for this? We've gone from 280 -> 480 -> 580 -> 680 -> 780 with around 15-25% (each generation) improvements in speed. Why would it all of a sudden be 250% faster?

He has no source, just sounds like a rumor.

780Ti is just as disappointing as I thought. The real sites like Anand etc will highlight this.

I think the two best cards are going to be the 290 + 780. For more mainstream gamers, the 280x and 760 are ideal.
 
Top Bottom