http://translate.google.com/transl [...] n&ie=UTF-8
 
what do you think of the result?
 
in my view people who said AMD have improve SO MUCH in drivers proof to be wrong.as the true DX10 benchmark shows.because from spec of the two card.the 9800GX2 is behind.or is it a architectural differences that make the result all "wrong"?

Yes, it's architectural inefficiency. 3870 is bottlenecked by low TMUs. A single 3870 is significantly slower than a single g92 8800gts, so it makes perfect sense.

See this though:
http://forums.legitreviews.com/about15370.html

From what I kno the architecture of the ATI cards have proven to be very good. Thus the results in 3Dmark and synthetic type tests. But When it comes to real gaming, a big fat no. Compared to top end cards, don't get me wrong.
 
Either way, thats just 1 website, and yes technically speaking now that they use DX 10 benchmarks the results might change, and prove that the Nvidias are the better synthetic cards as well.  
 
But with the new DX 10.1 I dunno what to say. I think the air will be cleared after the new line of cards come out.

Yeah, the 3870X2 seems to do well in Vantage, however, not so much in the real world.  However, to be fair, we have no DX10.1 benches or tests.  That is what the 2000 (although they can't do it) and 3000 series were designed for.  Hopefully that will start to change soon and we will be able to see cards like the 3000s and 4000s shine.

i think there will be a DX10.1 patch to the Vantage AFTER more games come out in DX10.1 form.but got now DX10 Vantage.
 
also that link you gave me.im not saying the result is fake.but isnt the result still abit "underachieving for such new gen card?i wonder what the score would be for the GT200.cant wait!!!lol
 
but still GDDR5 thats pretty impressive and that will whoop some ass if Nvidia come up with GDDR3 lol
 
that 4870 have 480SP where as 9800GX2 got 256SP thats about 53% in difference but in the benchmark shows about 70% gain thats very impressive.i guess its the GDDR5 made the difference even though the card is running on 256bit bandwidth.
 
i got a question why dont they just use the cheaper GDDR4/3 with 512bit to do the job rather then getting the expensive GDDR5 to offset the tight bus?

How do you know it's expensive? Maybe they have a streamlined factory floor for it.

 
something new will cost more isnt it?look at ddr2 to ddr3 now.

The cynic in me says:
"Ho, ho, ho, how far ahead would ATi cards be if Nvidia hadn't insisted that MS make DX10 more like DX9...."  
DX10 is basically (in my eyes) DX9.5...
DX10 was supposed to be revolutionary in the way it handled shaders (hence the vastly different make-up of the ATi cards in comparison to Nv). Nvidia couldn't get a decent card out that worked like that, so MS pandered to their every whim and bang DX10 was born, that easn't revolutionary, but evolutionary...
Judging by the (supposed) DX10.1 benefits in Assasins Creed, the sooner we get to DX10.1 (the real DX10) the better.  

Double post...

^ Exactly. DX9.5. The reasoning for ATI going with GDDR5 will be shown in this generation, if its utilized completely, its more than just faster. And as for the 256 bus, going 512, your pcb has to be very very complex, which puts the cost higher than the GDDR5, plus, using the newer ram reduces power usage as well. GDDR5 is the way to go for future cards/solutions. Cant wait to see their MCM working