The New and Improved "G80 Rumours Thread" *DailyTech specs at #802*

Well if its getting 12k, a intel core duo e6700 gets around 2500 cpu marks,

The 7950gx2 gets around 8k
12 k for the g80


graphics card scores of 5500 for the gx2
graphics card scores of 9500 for the g80

~70% increase

 

we're not that desperate are we? that tells us nothing. Could get 3000 '3dmarks' and tripple its performance in real world games.

 

Of course!

 

And La Inq is doing it again, give it one or two years and you can run Windows Vista SP4 (?) on your own Nvidia powered C(G)PU...

http://uk.theinquirer.net/?article=35216

 

Of course the r600 will end up faster, ATi knows what the competition is at, they can now play numbers games! Wow the Inq is going overtime with thier ability to get performance figures on a chip that probably doesn't have solid drivers yet I mean couldn't they state anything less obvious

 

think about 3.8G cpu mark at 06

 

Here are some fairly reliable numbers from the same source who was spot on with previous numbers (eg RV570, RV560, G71) before they got launched:

3DM05 / 3DM06
8800GTX 16909 / 11843
8800GTS 15233 / 10071
7950GX2 12931 / 8781
1950XTX 11866 / 7007
1950PRO 10044 / 5600

No specs of the system were given, but all the cards ran on the same system.

 

Those 06's are interesting.
If we take 3k off of them for the cpu score, it's a 25% gain for a 33% increase in ALUs (GTS->GTX).

Hmm.

If the chip has the same number of trannies as the sum on the GX2, and runs its ALUs twice as fast, and only gets a 50% boost over the GX2, I'd call that ... surprising. Still very nice, but I'm wondering about the efficiency of the system.

 

It's a single 575MHz core, against two 500MHz G71 cores.
How is a 75MHz difference running "twice as fast" ?
The core speed isn't directly comparable between two very distinct architectures.

The results are in line with expectations, but i would prefer to see real world games/benchmarks.

 

You're quite correct, my apologies. I meant ALU speed.
Of course, if the ALUs are doing all the processing, it isn't really clear to me what the core clock is actually clocking aside from internal cache access. ;^/ Thread dispatch? ?? I must be missing something there.

True enough. If heat and power generation/use scale at better than the diff in performance, it's a win.

Actually, it makes me wonder what kind of operations this architecture will really excel at.

 

How can you gauge efficiency without a reference value? Obviously the increase in transistor count wasnt directed solely at performance.

 

I don't recall what the drop for 4xAA in 1600 on a G71 in 3dfart06 is, but if the 7k/8xAA rumour under those settings should be real, I wouldn't be suprised if the drop for 4x samples is roughly around 20+%.

 

Just saw it, the German game magazine Gamestar claims to know the specs:
http://www.gamestar.de/news/hardware/geforce_8800_gtx_und_gts/1465523/geforce_8800_gtx_und_gts.html

Seems pretty much the same what has been suggested here, unfortunately, no source is given and "Unified shader (according to source: Unified shader)" sounds rather weird to me ...

 

the guy who has written that article should take more sleep:
http://www.gamestar.de/index.cfm?pid=29&pk=24

 

Well, he looks always like that, but the news he posts are usually accurate - much better than L'Inq for sure.

 

Don't get me wrong, I was just kidding..not really giving an opinion about his work.

 

http://uk.theinquirer.net/?article=35244



what do you think about it ???

 

Might be an issue with different clock domains not playing nicely when you start messing about with their speeds. This was supposed to be the problem with the 7900s that shipped at higher than reference speeds and the cause of a lot of returns.

Given that G80 is going to be very sophisticated and I think is likely to run different parts of the chip at different speeds, I can see that Nvidia might not think it's worth the hassle of endorsing overclocking when they know it just causes problems.

 

*Rolls Eyes at Inquirer*