How can it be that the 8800 GTS becomes slower in the Oblivion bench than the 8600 GTS? These diagrams let me doubt on the quality of your reviews!

My thougts exactly.

I'm extremly unhappy with the results that the 8600 GTS yielded and thus I think that nVIDIA has made a big mistake..big time....Giving that there are only 32 stream processors and a very narrow memory inteface of 128-bits ,I'm hardly suprised..
So..I'll ask you guys a queastion for a much nVIDIA is gonna screw the mid-rage with GPUs sporting only 128-bits interface?!?!
This is just insane...!!! The midrange is sucked big time..I cannot justify a 199-229 USD price point of an 8600GTS giving the level of performance it provides..must likely I think its price should be around the 149 USD mark..
I'm an nVIDIA fanboy so I've not lost my fate on nVIDIA products..but this one..

I'm expecting a new more powerful GPU for the midrange in the form of 8800 GS consisting of 64 Stream Processors and a wider 256-bit bus..

i'm looking forward to the mobile version of these along with the santa rosa platform

this'll be perfect for mobile gaming without draining too much battery life

the desktop variant's price shud be dropped $50 to be competitive since it doesn't perform any better than the x1950 pro and 7900 gs

FOR XBIT-LABS REVIEW STAFF :

I think those benches of the 8800GTX and 8600 GTS in your site are not providing the full peformance picture (mainly for 8800GTX) giving that you are using a very old AMD FX-60 CPU that is limiting the 8800 GTX to show its full potential in benchmarks..To overcome the bottleneck I think Xbit-labs should move to a more powerfull CPU like an Intel C2E X6800/QX6800....I said that only becauese I think your reviews should be more accurate form performance perspectives..

It is right that the CPU might be the bottleneck in these tests - but how should it be possible that a GPU having only one third (32 - well clocked a little bit higher) performs better than the GPU having 96 parallel Units - especially in these high resolutions where the smaller amount of memory (256 MB) should be much more limiting than the 640 MBs of the 8800...I hope that it is a mistake because I just ordered the 8800 GTS for my system - depending on these results I should have ordered the 8600...=)

Bah, overclocking of the card only got 2 sentences in the whole article :/

I would be really interested in 8600GTS performance after overclocking, as a review at HardOCP suggests that it increases the cards performance significantly.
That could make it more interesting than the X1950Pro, which doesn't overclock very well AFAIK.


Anyway, fingers crossed for ATI's offering in the $200-$300 range being better.

Nice review anyway!! thanks

think abt it logically

if a person is gonna buy a mainstream card in the region of $200,i'm pretty sure he won't have a quad-core prcessor(right now anyway) ro even an FX-60

so for a mainstream rig the fx-60 is actually being a bit too kind
it does bottleneck the 8800 cards but its hardly an issue with the 8600 cards

it is better to buy R1950pro... and wait for GF 8800GS

Radeon X1950 pro is a King :)

Great article. You must be the only one amongst the reviews I have seen, to comment on the higher noise level of the 8600 GTS.

It seems the 8600GT could be a more reasonnable card from the price/performance standpoint. I hope you review it too.

great stuff i was just woundering what the proformance results would be in 1024 by 768 display resolutions

I hate green PCB.

Shortly taken, I scr*wed myself by ordering a Asus Nvidia EN8600GTS/HTDP 256MB DDR3 SDRAM (PCI-E) before viewing all reviews... I thought I bought a nice high end card for this kind of money...

excellent article, ty it help me a loot in deciding should i upgrade to 8600GTS or stay on ATI 1950pro :D

hehe

ty and keep up a good work :D