• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

some fermi performance numbers

I' am not trying to start anything here but I have to agree...the difference in a majority of cases is about 7%, IDK if you overclock it a 5850 against a stock clocked 5870 the maybe I guess, but then again a 5870 can be overclocked also......funny how some people owning the 5850 have to bag on the 5870, as if the people who bought a 5870 are dumbasses or is it that somebody feels that they need to justify being a tightass not being willing to spend the extra $60-$100.....once again IDK!

http://www.xbitlabs.com/misc/pictur...radeon-hd5850/diagr/22_585vs587aa_big.png&1=1

To be clear the bolded comment is indeed my personal belief though I can neither confirm nor deny your other accusations. The stock 5870 can reach higher overclocks than the 5850 due to a higher default core voltage as I've stated earlier. If you have a nice 5850 core then you will be able to hit the same overclocks with a voltage bump to an equivalent 5870 level. Of course YMMV and since the 5850 is by nature a binned product it is certainly possible, even likely, that the average 5870 will clock higher. I can certainly see going 5870 if you don't overclock at all OR if your budget is not crucial (i.e. going for the 5870 won't force you to bump down a notch on the processor). In addition the beefier cooling of the 5870 will be better-equipped than the 5850 clock-for-clock, and quieter. The 5870 is ingenious on AMD's part given the infinitesimal performance return (per hardware cost) it offers for the modest premium.

But let's not derail.

If the difference between the GTX 470 and GTX 480 is at least greater than the difference between the 5850 and 5870 given respective maximum overclocking potential, I believe the 480 can be a relatively greater success (compared to the 5870, that is).
 
http://translate.google.com/transla...rce-GTX-470-enthuellt-946411.html&sl=de&tl=en

Preliminary benchmark summary (may change if NVIDIA gets better drivers, or final specs are at higher clocks, etc.): 5850 < GTX 470 < 5870 in all cases in 3DMark, the GTX 470 being closer to 5870 overall.

In Unigine, something weird happens. At 4x AA, the GTX 470 leads at 29fps, compared to 27fps and 22fps for 5870/5850 respectively. But at 8x AA, the 5870 leads at 23fps followed by the GTX 470 (20fps) and 5850 (19fps). Not sure what resolution they tested at but it appears to be 1920x1200 since that would line up well with the Chinese data leak earlier.

I suspect MINIMUM framerates are better with the GTX 470 and that's what matters more, but no good data on that so far.

No game benchies yet, I'd love to see what the GTX 470's minimum framerates are on AVP or the newest STALKER.
 
Last edited:
http://translate.google.com/transla...rce-GTX-470-enthuellt-946411.html&sl=de&tl=en

Preliminary benchmark summary (may change if NVIDIA gets better drivers, or final specs are at higher clocks, etc.): 5850 > GTX 470 > 5870 in all cases in 3DMark, the GTX 470 being closer to 5870 overall.

In Unigine, something weird happens. At 4x AA, the GTX 470 leads at 29fps, compared to 27fps and 22fps for 5870/5850 respectively. But at 8xx AA, the 5870 leads at 23fps followed by the GTX 470 (20fps) and 5850 (19fps). I assume this is 1920x1200 since that would match up with the recent Chinese benchmark leak. All fps cited are with DX11/Tessellation on.

I suspect MINIMUM framerates are better with the GTX 470 and that's what matters more, but no good data on that so far.

No recent comparative game benchies. I'd love to see what the GTX 470's minimum framerates are on AVP or the newest STALKER.

fermi will easily win every dx11 game benchamark
 

Yes, exactly what I was thinking about re: framerates. I have no particular brand loyalty (my previous card was NVIDIA, and I currently run a 5850), and from my more or less neutral point of view, Fermi's architecture looks like it will be better in high-stress situations, i.e., when fps drop like a rock.

So 58xx series cards should be very competitive from an average fps angle, but lose out to Fermi when comparing minimum fps.

NVIDIA's main problem, though, is a) price/performance and b) availability. Feature set is debatable (e.g., EyeFinity vs. CUDA/PhysX/3D).
 
Yes, exactly what I was thinking about re: framerates. I have no particular brand loyalty (my previous card was NVIDIA, and I currently run a 5850), and from my more or less neutral point of view, Fermi's architecture looks like it will be better in high-stress situations, i.e., when fps drop like a rock.

So 58xx series cards should be very competitive from an average fps angle, but lose out to Fermi when comparing minimum fps.

NVIDIA's main problem, though, is a) price/performance and b) availability. Feature set is debatable (e.g., EyeFinity vs. CUDA/PhysX/3D).

no, it just GTX 480 doing tessellation better than 5870...

which no massive tessellation in use, 5870 will be in par with GTX 480..

looks like somehow charlie have a really good source in the back...
wasn't going to believe that but this really does back his "assumption" up...:eek:
 
no, it just GTX 480 doing tessellation better than 5870...

which no massive tessellation in use, 5870 will be in par with GTX 480..

looks like somehow charlie have a really good source in the back...
wasn't going to believe that but this really does back his "assumption" up...:eek:

I'm sorry. Didn't charlie say that Fermi tesselation performance was pants because shaders had to do the work? (No dedicated hardware tesselation)
 
I'm sorry. Didn't charlie say that Fermi tesselation performance was pants because shaders had to do the work? (No dedicated hardware tesselation)

thought his last article did mention in Heaven benchmark does show impressive improvement over 5870 due to massive use of tessellation..
thats what I am referring to...
 
fermi will easily win every dx11 game benchamark

If the A2 benches are to go by then yea, only if it's tessellation heavy as hell and has tessellation heavy scenes in benchmarks and most levels where say [H] would make runthroughs for FPS measurements.
 
I'm sorry. Didn't charlie say that Fermi tesselation performance was pants because shaders had to do the work? (No dedicated hardware tesselation)

That might be partially or even totally right because the GTX 470 loses its lead when Unigine goes from 4x AA to 8x AA.

Or it could be due to some sort of memory issue. Since preliminary GTX 470 specs have memory at only 3200 Mhz for some reason (heat? power? cost?).

By the way, another interesting tidbit from that benchmark article in German: the GTX 480 may NOT have all 512 cores active! If the GTX 470 is at 448, then that implies that the GTX 480 will have only 480 cores. Assuming core is the bottleneck (e.g., assuming memory is not bottleneck), 480/448 = 1.07, which would make a 480-core GTX 480 only slightly faster than a GTX 470.
 
Last edited:
all you need is a good cooling and a volt mod from evga to make this beast really roar.

nvidia is clocking these cards at low speeds so it doesnt overheat in cases that barely get any air.
 
I bet the price will make you squeal....ati wont need to adjust any prices if these numbers are close. The nvidia part will cost much more, and barely perform better. ATI will be set for a while, getting above original msrp. but well below the overpriced nvidia offering of similar performance.
 
NVIDIA will be releasing a part that performs slower than the 5870 whilst costing $100 more?

FAIL. The GTX 470 will have to cost $399 at max in order to succeed.
 
Multiple sources now say that at best it barely beats the 5870 and some say below it. I think that is not right and might be a driver issue but I would be very surprised now if it is even 10 percent faster.
 
Multiple sources now say that at best it barely beats the 5870 and some say below it. I think that is not right and might be a driver issue but I would be very surprised now if it is even 10 percent faster.

those sources dont even have the drivers to even support this card, lol.
 
uh?

DRIVERS? then how to operate them without it ? :rolleyes:

I bet nVidia have a magic wand to make it work..

they dont even have the card, those asian benchies are bogus.

only nvidia has the card and the drivers as of now.
 
they dont even have the card, those asian benchies are bogus.

only nvidia has the card and the drivers as of now.

uh huh....

and somehow it shows exactly where it perform just like nVidia's graph..

what a coincident huh...
 
uh huh....

and somehow it shows exactly where it perform just like nVidia's graph..

what a coincident huh...

that was from nvida headquarters, there is a difference.

and the graph shows nothing about what you talking about.
 

I have to agree with dook on that one. Just posting that graph means nothing without context to the situation (which wasn't given). To most, that is some random ass graph with no information

I am hoping that the 400 series is kickass. If it falls this short, it will be unfortunate as ATI can sit idley by without being pressured to release a refresh. Another words, there will be very little progression in this release because if it competes on par with 5000 series, it will distance us from the 6000 series or 500 series :(.

It does seem that Nvidia is hang their hat completely on this gimmicky proprietary tech like 3D and Physx though (Yes they are gimmicky, 2 major titles utilize physx, and very few utilize 3D. When these things are readily implemented into many titles, they will no longer be a gimmick, which is what I hope happens). That is very unfortunate because it leads to the conclusion that performance wise, it will be unimpressive. Puting so many marbles into the 3D and Physx basket throws some red flags up concerning the performance of these cards :(. Hopefully this is not a marketing tactic by Nvidia to point discerning eyes away from the subpar performance (though it seems to have many fanboys convinced that this the polished turd that is Physx and 3D vision at this point is enough to indicate obvious superiority of Nvidia).

Heres to hoping that the 400 series is epically kick ass! Competing companys fighting for the worlds fastest video card titles means that us, we educated, non brand loyalist consumers, get some super fast, sexy performing videocards ASAP.
 
Last edited:
LOL!

But overall I agree that the numbers are just too poor. There is no way it wont beat a 5870 that would mean disaster for Nvidia. but all these claims of 20 percent over 30 percent over and the like look more and more bogus every day.

Too hot, Too slow, too late in my opinion.
 
No need for these cards, wait for the B stepping cards.

Those will be the gtx475 and the gtx485.

Should have better clocks and power characteristics.
 
LOL!

But overall I agree that the numbers are just too poor. There is no way it wont beat a 5870 that would mean disaster for Nvidia. but all these claims of 20 percent over 30 percent over and the like look more and more bogus every day.

Too hot, Too slow, too late in my opinion.

true, dx11 is over and dx12 is already here, yawn.

and lol if you actually believe those numbers from that asian site unless you believe that they went in the future with a time machine and brought back the final drivers that dont even exist right now.
 
Last edited:
Fun fact: The 5850 is within 5% of the 5870 (close to 2% actually) when both are at the same clocks. Since they have the same core but different voltages, many 5850s will be able to overclock just as far as 5870s given the same GPU vCore.


Nonetheless, like the 5870, the GTX 480 will be a genius move. It will be a smash hit with the die-hard pay-for-the-best enthusiasts. I am betting that the performance difference over the 470 will be small for the premium, but it would be nice to be pleasantly surprised.

How hard is it to do this overclock on the 5850? Do you need special cooling to maintain it, or just a good case?
 
true, dx11 is over and dx12 is already here, yawn.

and lol if you actually believe those numbers from that asian site unless you believe that they went in the future with a time machine and brought back the final drivers that dont even exist right now.

What?

Every single thing you post is pure biased speculation and you're accusing him of making up information!?

Next thing you are going to say is that the image quality of the 480 in the heaven benchmark is much superior than ATI even though they are rendering the exact same information. Oh wait, you already did say that, my bad :rolleyes:

anywho, trolls are funny to poke and feed. It makes for some entertaining, fictional fact sharing.
 
LOL!

But overall I agree that the numbers are just too poor. There is no way it wont beat a 5870 that would mean disaster for Nvidia. but all these claims of 20 percent over 30 percent over and the like look more and more bogus every day.

Too hot, Too slow, too late in my opinion.


Wow...now we're quoting Charlie Demerajin? Look I agree, the card is definitely a poor showing for nVidia. Unless they truly optimize some drivers here, it's going to be hit and miss equivelent to a 5870 stock. Nothing to write home about. But Charlie? LOL...
 
So many posts here with people having no idea what they are talking about. Why don't you wait until the actual card hits the market and then is reviewed. At the moment, nobody except Nvidia knows.
 
So many posts here with people having no idea what they are talking about. Why don't you wait until the actual card hits the market and then is reviewed. At the moment, nobody except Nvidia knows.

Where's the fun it that? How are they going to aggravate one another by posting "facts"?
 
Not hardly, Charlies 5% numbers were for the GTX480, not 470. And if you did around, you'l find where I said the GTX470 should be right around the 5870 only slightly faster.

Hmm from what I remember you kept spouting off how 5% was wrong, and it was like 30% faster.

Either way ill wait for real reviews
 
Back
Top