R520 - 8600 3dmark05

[trinibwoy]

http://www.hkepc.com/bbs/viewthread.php?tid=466843&extra=&page=3

Originally posted by: HKEPC Admin
10vs24ps 650/1400,05 about 8600 <---- I can comfire this mark

That's kinda low given the much higher clocks and more vertex shaders in the spec he claims got that score. My GTX at 480/1300 gets > 8600 already.

 

[Powermoloch]

no pics? no eveedense? no satisfaction...

 

[dug777]

Originally posted by: Powermoloch
no pics? no eveedense? no satisfaction...

everyone believes everything else they've said about the r520 so far..or so it seems

 

[Powermoloch]

Originally posted by: dug777

Originally posted by: Powermoloch
no pics? no eveedense? no satisfaction...

everyone believes everything else they've said about the r520 so far..or so it seems

apparently :Q ~!

 

[biostud]

I can confirm 8600 too.

 

[trinibwoy]

Originally posted by: Powermoloch
no pics? no eveedense? no satisfaction...

True. But the difference here is the source. HKEPC is the same site that produced this - http://www.hkepc.com/hwdb/r520firstlook-e.htm - so they do have the hardware. And given their location, they usually have access to hardware long before their western counterparts.

So when the admin claims something, you might want to give it more heed than the usual bs floating around. In any case, it's something to discuss.

 

[Powermoloch]

Originally posted by: trinibwoy

Originally posted by: Powermoloch
no pics? no eveedense? no satisfaction...

True. But the difference here is the source. HKEPC is the same site that produced this - http://www.hkepc.com/hwdb/r520firstlook-e.htm - so they do have the hardware. And given their location, they usually have access to hardware long before their western counterparts.

So when the admin claims something, you might want to give it more heed than the usual bs floating around. In any case, it's something to discuss.

I already seen that pic already a while back...i'll just wait for anandtech to give the more reliable results

 

[trinibwoy]

Originally posted by: Powermoloch
I already seen that pic already a while back...i'll just wait for anandtech to give the more reliable results

Whatever makes you comfortable man Just posted it up for those who venture outside of the AT forums for their info.

 

[ddogg]

well if its true its nothing all that great considering all the hype. looks like this card should run neck-n-neck with the 7800GTX(maybe slightly faster).

 

[Powermoloch]

Originally posted by: trinibwoy

Originally posted by: Powermoloch
I already seen that pic already a while back...i'll just wait for anandtech to give the more reliable results

Whatever makes you comfortable man Just posted it up for those who venture outside of the AT forums for their info.

:thumbsup:

 

[Auric]

What's the scuttlebutt? Is anyone actually expecting significantly better performance than NVIDIA's offering?

 

[Corporate Thug]

maybe its worth the wait after all? :evil:

 

[ayabe]

I thought lastweek they said they weren't going to release any benches, now they are? WTF?

Since I can't read anything on that page these numbers could be anything. Could be for the 9800 Pro Chairman Mao Edition for all we know.

 

[Corporate Thug]

Originally posted by: ayabe
I thought lastweek they said they weren't going to release any benches, now they are? WTF?

Since I can't read anything on that page these numbers could be anything. Could be for the 9800 Pro Chairman Mao Edition for all we know.

i thought i had the only 9800 pro chairman mao edition....

 

[Janooo]

Just a quote from the HKEPC article: "As the card maker doesn't wish to reveal the R520 performance yet, but we can confirm that R520 at 24pp or 32pp will not be lower than that of the GeForce 7800GTX. Even though the 32pp and 24pp R520 mass production will face some difficulties, but the 16 pipelines R520 using the 90nm core will be clocked high to maintain a certain level of competitiveness. Apart from employing TSMC to make the 90nm R520 cores, UMC will be making them as well in the near future to make up the shortage problem and we can see more of the 32pp and 24pp R520 versions."

They have card in their hands. Why not to believe them?

 

[Avalon]

If one were to judge by those numbers, then this round will be about as close as last round, except for the fact that ATI will be up to date on their SM abilities.

Then again, it is 3dmark. I'll wait until I can see some high res benches.

 

[ayabe]

Originally posted by: Janooo
Just a quote from the HKEPC article: "As the card maker doesn't wish to reveal the R520 performance yet, but we can confirm that R520 at 24pp or 32pp will not be lower than that of the GeForce 7800GTX. Even though the 32pp and 24pp R520 mass production will face some difficulties, but the 16 pipelines R520 using the 90nm core will be clocked high to maintain a certain level of competitiveness. Apart from employing TSMC to make the 90nm R520 cores, UMC will be making them as well in the near future to make up the shortage problem and we can see more of the 32pp and 24pp R520 versions."

They have card in their hands. Why not to believe them?

So ATI gave their blessing then? Somehow, I think not. This doesn't confirm or deny anything for me, the waiting game continues.

 

[Munky]

Well, we've already seen that the 7800gtx, while only about 1000 points ahead of the x850 in 3dmock05, usually beats it by a big margin in real life games. So, using that same argument, one could say that the 24pipe r520xt, while only 1000 points ahead of the gtx, may beat by just as much in real life games. And maybe you need a FX-59 and a 3000x2000 display resolution to see the difference because, well, it's clearly bottlenecked by the cpu

 

[trinibwoy]

Originally posted by: munky
Well, we've already seen that the 7800gtx, while only about 1000 points ahead of the x850 in 3dmock05, usually beats it by a big margin in real life games. So, using that same argument, one could say that the 24pipe r520xt, while only 1000 points ahead of the gtx, may beat by just as much in real life games. And maybe you need a FX-59 and a 3000x2000 display resolution to see the difference because, well, it's clearly bottlenecked by the cpu

No your logic is flawed - the reason for the PE being close, and the R520 being ahead is probably the same - higher vertex performance. So you can't just flip flop the cards in the comparison. If anything, this indicates that the GTX will be closer to the R520 in non vertex fetch limited situations (real games) than the 1000 discrepancy here would indicate.

And 3dmark05 is not CPU bottlenecked at < 10,000 - else SLI configs would show no increase. My GTX goes from 7700 with a core of 432 to 8600 with a core of 486 - 12.5% increase on the core results in ~ 12% increase in score - so it is still very much GPU bottlenecked. This is on a 2.7Ghz Venice.

 

[Regs]

Originally posted by: trinibwoy
If anything, this indicates that the GTX will be closer to the R520 in non vertex fetch limited situations (real games) than the 1000 discrepancy here would indicate.

Of course, it also depends on how they program the real games as well. I'm sure games like Half Life will take full advantage over what ATi has to offer.

 

[trinibwoy]

Originally posted by: Regs

Originally posted by: trinibwoy
If anything, this indicates that the GTX will be closer to the R520 in non vertex fetch limited situations (real games) than the 1000 discrepancy here would indicate.

Of course, it also depends on how they program the real games as well. I'm sure games like Half Life will take full advantage over what ATi has to offer.

Well, I would expect the usual suspects - HL2, Doom3, EQII, Riddick etc etc to remain with the relevant camp. I'm interested in seeing how UEIII / F.E.A.R will do on the next generation parts.

 

[Munky]

Originally posted by: trinibwoy

Originally posted by: munky
Well, we've already seen that the 7800gtx, while only about 1000 points ahead of the x850 in 3dmock05, usually beats it by a big margin in real life games. So, using that same argument, one could say that the 24pipe r520xt, while only 1000 points ahead of the gtx, may beat by just as much in real life games. And maybe you need a FX-59 and a 3000x2000 display resolution to see the difference because, well, it's clearly bottlenecked by the cpu

No your logic is flawed - the reason for the PE being close, and the R520 being ahead is probably the same - higher vertex performance. So you can't just flip flop the cards in the comparison. If anything, this indicates that the GTX will be closer to the R520 in non vertex fetch limited situations (real games) than the 1000 discrepancy here would indicate.

And 3dmark05 is not CPU bottlenecked at < 10,000 - else SLI configs would show no increase. My GTX goes from 7700 with a core of 432 to 8600 with a core of 486 - 12.5% increase on the core results in ~ 12% increase in score - so it is still very much GPU bottlenecked. This is on a 2.7Ghz Venice.

Well, in case you couldnt tell, I was being sarcastic and took a few potshots at all those "OMG the 7800gtx is only 5 fps faster - must be the cpu, and you need 20x15 res to let it really stretch it's legs" posts. I know that 3dmock05 is still gpu limited, and I also suspect that a lot of it's score depends on pure vertex performance, as well as excessive use of shaders.

As far as it's relevance to real world games, you're implying that the Ati cards have a higher vertex performance, but it doesnt translate into an equivalently higher performance in real games, no? In part, that has been true ever since the r300 cards, but we dont know how that will relate to the r520. It would probably be true if it had 10vs and 16ps, since real games put more stress on the pixel shaders. But what if it has 24ps, or 16 extreme-ps (whatever that means)? Then it could turn out a lot faster in real games than a 7800gtx.

 

[trinibwoy]

Originally posted by: munky
Well, in case you couldnt tell, I was being sarcastic and took a few potshots at all those "OMG the 7800gtx is only 5 fps faster - must be the cpu, and you need 20x15 res to let it really stretch it's legs" posts. I know that 3dmock05 is still gpu limited, and I also suspect that a lot of it's score depends on pure vertex performance, as well as excessive use of shaders.

Ah, you're right - missed the italics and smiley

Originally posted by: munky
As far as it's relevance to real world games, you're implying that the Ati cards have a higher vertex performance, but it doesnt translate into an equivalently higher performance in real games, no? In part, that has been true ever since the r300 cards, but we dont know how that will relate to the r520. It would probably be true if it had 10vs and 16ps, since real games put more stress on the pixel shaders. But what if it has 24ps, or 16 extreme-ps (whatever that means)? Then it could turn out a lot faster in real games than a 7800gtx.

Yep, that's what I was getting at. But I'm sticking with 650x16. No way in hell would a 650x24 piped card score only 8600 in 3dmark05. A 650x24 piped R420 would score more than that.

 

[xtknight]

Originally posted by: trinibwoy
And 3dmark05 is not CPU bottlenecked at < 10,000 - else SLI configs would show no increase. My GTX goes from 7700 with a core of 432 to 8600 with a core of 486 - 12.5% increase on the core results in ~ 12% increase in score - so it is still very much GPU bottlenecked. This is on a 2.7Ghz Venice.

I couldn't agree more. The term "cpu bottleneck" is so overrated it deserves a page on Maddox.

Yep, that's what I was getting at. But I'm sticking with 650x16. No way in hell would a 650x24 piped card score only 8600 in 3dmark05. A 650x24 piped R420 would score more than that.

Unless they did something stupid like put only 8 ROPs (right term?) on it. Let's hope not. We don't have any idea of IPC or efficiency either.

 

[Hacp]

OMFG you can play 3dmark????