Where are the leaks for some 7900 benchies!?

M

mrkram

Gawd
Joined
Sep 4, 2005
Messages
564
Man o man March 9th couldn't come any slower! Im sure someone out there has got a link brewing to show some early benchmarks for all of us waiting to get our pre march 9th fix............GO GO GO!
 
If the benchies are correct then the performance difference between these and my 7800gts overclock does not seem that much...I doubt that these will beat the x1900xtx...
I guess I will keep on enjoying my 7800gts a while longer...
 
1) I heard 7900 leaks are currently still fakes

2) what's the deal with using 3d mark? Why's that wrong for 7900?
 
Does the 7800GTX 512 and the 7900GTX use the same memory?GDDR3 1.2ns? :confused:
 
PsichoDM said:
2) what's the deal with using 3d mark? Why's that wrong for 7900?
Click to expand...

Becuase 3D-shitty-mark gives no real info on how the card will preform in games.
In 3D '06 a 6600GT will get a higher 3Dmark score than a X850XT-PE...due to SM3.0.
But test the 6600GT and X850XT-PE in real games...and the picture changes...

Terra - 3D-shitty-mark is useless....
 
I was gonna say the same thing :). Those are some pretty DAMN good numbers for 12x10!
 
3DMark isn't useless.

It's one indication of graphics performance. It shouldn't be taken as the SOLE indication, just the same way as a Quake 4 benchmark doesn't show how well a card will handle FEAR or Call of Duty 2. But as one single measure of performance, its no better or worse than any other single test.

And yes, the SM3.0 issue is anomalous for a few cards - but it has no bearing whatsoever on 7900GTX v 1900XTX comparisons does it?
 
I just wanna see [H] whip out their review on the card as I can see the comparison between the 7900GTX and the X1900XTX or XT in both performance at high res and image quality! March 9th can't come quick enough!
 
pibrahim said:
3DMark isn't useless.

It's one indication of graphics performance. It shouldn't be taken as the SOLE indication, just the same way as a Quake 4 benchmark doesn't show how well a card will handle FEAR or Call of Duty 2. But as one single measure of performance, its no better or worse than any other single test.

And yes, the SM3.0 issue is anomalous for a few cards - but it has no bearing whatsoever on 7900GTX v 1900XTX comparisons does it?
Click to expand...
3Dmark should be used to compare cards among the same manufacturers. The 6600GT vs X850XT theory doesn't hold here due to different manufacturers and technology differences. Now, 3Dmark a 6600GT vs the new 7900GT and watch the picture change radically.

And the 7900 vs X1900 arguments doesn't hold either since ATI cards improperly render a scene in 3Dmark06, thus affecting their score.
 
Sure, the cards might render scenes in different ways (I don't know the ins and outs of this). But, assuming that we don't buy into the idea that either nVidia or ATI slipped some cash FutureMark's way, why wouldn't Futuremark try their best to compensate for that?

Far Cry shows some differences in rendering (with the shadows, etc) but that doesn't stop people using it for benchmark purposes. If the argument is that there's differences in rendering methods, image quality, or anything like that then it applies to every game benchmark too, not just 3DMark.

And for what its worth, 3DMark seems to be doing just fine with the X1900XT/X benchmarks - it gives them a lead over the GTX 512 (http://www.firingsquad.com/hardware/ati_radeon_x1900_xt_preview/page6.asp), which is in keeping with the majority of games which also give the former the card the lead...
 
No offense, but I find those scores completely rediculous...

By those benchies, my 7800GTX 256MB at stock speeds (460MHz, 1.30GHz) just beat out the 7900GTX at stock speeds... which makes absolutely ZERO sense at all...

7800GTX 256MB -> 7900GTX 512MB
24pipes x2 ALU -> 24pipes x2 ALUs
110nm core -> 90nm core
460MHz core -> 650MHz core
1.3GHz RAM -> 1.6GHz RAM
8,200 3dmark05 stock -> 7,600 3dmark05 stock

(and I am running a similar setup to their "test" setup)... Its the same architecture, just smaller die.... These benchmarks are a hoax... No one would be upgrading or buying these cards as they are almost being outperformed by 7800GT's according to those benchmarks... wait till the 9th when we can see some real scores.

Edit: I didnt realize it said 1280x1024... but even so... That shouldnt degrade the score that much. Not to mention that it should have been run in the program's native res... that way people can more readily compare the scores... I still say these scores are a hoax. Just like the screenshots from another thread showing the 7900GT scores in 03 05 and 06.

Edit 2: Commander corrected me below.
 
Commander Suzdal said:
The 7800 series has x2 ALUs per pipe, just like the 7900. See Brent's [H] preview of the 7800.
Click to expand...

yea, i just realized that...fat fingers on a laptop at work... I miss my ergo keyboard at home :(
 
i bet the [H] staff is barely on page 8 of 23 on the review =P

edit: correction page 8/23 of the Evaluation I keep forgeting :D
 
BLiTzKRiEG said:
No offense, but I find those scores completely rediculous...

By those benchies, my 7800GTX 256MB at stock speeds (460MHz, 1.30GHz) just beat out the 7900GTX at stock speeds... which makes absolutely ZERO sense at all...

7800GTX 256MB -> 7900GTX 512MB
24pipes x2 ALU -> 24pipes x2 ALUs
110nm core -> 90nm core
460MHz core -> 650MHz core
1.3GHz RAM -> 1.6GHz RAM
8,200 3dmark05 stock -> 7,600 3dmark05 stock

(and I am running a similar setup to their "test" setup)... Its the same architecture, just smaller die.... These benchmarks are a hoax... No one would be upgrading or buying these cards as they are almost being outperformed by 7800GT's according to those benchmarks... wait till the 9th when we can see some real scores.

Edit: I didnt realize it said 1280x1024... but even so... That shouldnt degrade the score that much. Not to mention that it should have been run in the program's native res... that way people can more readily compare the scores... I still say these scores are a hoax. Just like the screenshots from another thread showing the 7900GT scores in 03 05 and 06.

Edit 2: Commander corrected me below.
Click to expand...

How about you try again but use 1280X1024 like the 7900GTX was tested on :p The res makes a larger difference than you think.
 
Budwise said:
How about you try again but use 1280X1024 like the 7900GTX was tested on :p The res makes a larger difference than you think.
Click to expand...

Lol! Damn those bastards! Throwing in a monkey wrench like that. In other news, my old camry goes from 0-60 in 3.3...



EDIT:Oops, sorry to confuse you all. I meant 3.3 as 3 times 3 = 9. Sorry for the confusion.
 
Well my 7800gts in sli score 7688 on 3dmark06 at the same resolution that the 7900gtx 512mb is tested at (to get 5200 for its score)...I would have think that for the price that is now more than twice the value of the 7800gts (2), the 7900gtx would score the same or higher...Oh well...I guess these babies will be used for a while more...
Although I love nvidia, I have to admit that I am disappointed in this launch...I doubt that the 7900gtx will clearly beat the x1900xtx (if it beats the x1900xtx at all)...
 
Meh, I don't care much about 3Dmarks, I just want to know how it will run Oblivion... well okay I'll settle for fear and black&white 2.
 
magnuspah said:
Well my 7800gts in sli score 7688 on 3dmark06 at the same resolution that the 7900gtx 512mb is tested at (to get 5200 for its score)...I would have think that for the price that is now more than twice the value of the 7800gts (2), the 7900gtx would score the same or higher...Oh well...I guess these babies will be used for a while more...
Although I love nvidia, I have to admit that I am disappointed in this launch...I doubt that the 7900gtx will clearly beat the x1900xtx (if it beats the x1900xtx at all)...
Click to expand...

You can't use 3dmark06's final score as a justification for comparing GPUs as 06 is a complete system benchmark... Overclocking the CPU, RAM and upping the system bandwidth have a significant impact on an 06 score (especially with you running a dual core opteron).

Use the strictly GPU comparisons of 3dmark05 or 3dmark03 for flat out speed comparisons, and just look at the game tests and shader model scores in 3dmark06...
 
thats why I am not putting any faith in these scores as they are not screenshots... and on a non OC'ed system/non OC GPU...

TBH, I finally saw the light in the idea that rather than upgrading to a 7900GTX... Now is the time to nab that second 7800GTX and go SLi... 7800GTX people are selling them like crazy... I'm gonna bump up to SLi here in the next 2 weeks and save myself some money doing it.
 
Yep, if you already own a 7800gt or a 7800gtx for about half the cost of upgrading (selling your 7800gt or gtx included) of upgrading, you can get sli and get a performance that rivals or beat a single 7900gtx overclocked (if the scores and some indications of only an 8-10% performance increase are right).
 
etjr said:
i bet the [H] staff is barely on page 8 of 23 on the review =P

edit: correction page 8/23 of the Evaluation I keep forgeting :D
Click to expand...

lol..lets hope theyre on 22/23
 
magnuspah said:
Yep, if you already own a 7800gt or a 7800gtx for about half the cost of upgrading (selling your 7800gt or gtx included) of upgrading, you can get sli and get a performance that rivals or beat a single 7900gtx overclocked (if the scores and some indications of only an 8-10% performance increase are right).
Click to expand...


Or you could sell your 7800 GTX cards(like me ;) ), make about $800 off of them and buy the new stuff. It's up to you. :)
 
I agree with everyone saying how crappy 3dmark is, but i need to know if the 7900gtx is much better than my x1900xtx so i can order one lol.
 
InorganicMatter said:
3Dmark should be used to compare cards among the same manufacturers. The 6600GT vs X850XT theory doesn't hold here due to different manufacturers and technology differences. Now, 3Dmark a 6600GT vs the new 7900GT and watch the picture change radically.

And the 7900 vs X1900 arguments doesn't hold either since ATI cards improperly render a scene in 3Dmark06, thus affecting their score.
Click to expand...

It doesn't matter if both cards are nvidia or not. Back when the Geforce 5 series came out, I 3dmarked an FX 5200 against a GF4 Ti 4200 and they got the same score due to the FX being a DX9 part, but I can assure you that the Ti 4200 was much much faster in real gaming. 3dmark is a totally garbage app that can be made to say whatever the driver writers want it to say.
 
 
Those look plenty legit to me! Awesome scores, but nothing mind blowing.
 
What's the difference between the scores in pics 11 and 13? I didn't see what was changed there...
 
pibrahim said:
What's the difference between the scores in pics 11 and 13? I didn't see what was changed there...
Click to expand...

I think the 11 is overclocked and the 13 is stock...even at overclock, the SM2 and SM3 scores are not that great...at least not better than 7800gts in sli....If these are true and those scores are overclocked scores, I guess, I will have to wait for the next gen..or at least until the game benchmarks come out to prove to me otherwise...
At least the 7900gtx seems to be able to compete with the x1900xtx...
For ref...my 7800gts sli overclocked scored 3227 in sm2 and 3211 in HDR/SM3...
 
You must log in or register to reply here.
Back
Top