9800GTX pic and specs

[Power5]

Ummm, who is to say that Nvidia CAN launch the 9800GTX that they leaked months ago? Maybe they just don't have G100 (or whatever was supposed to be on this Card) in working order yet. So, 3870x2 is kicking their butt, so they release a revamped GTS with TriSLI to fill the gap till they can make the next Chip work. Everyone is saying that they don't have to release a much better card, what if they CANT!

[YukonTrooper]

Weak.

[Arkangyl]

Is it really that surprising? the 8800 GT and (new) GTS really should have been 9xxx line cards, except nVidia didn't want to kill of sales of it's old G8x parts by releasing it. Notice how the 8800 GTX & Ultra have been disappearing over the last few months?

G9x is very much just a slightly improved G8x on the newer process.

Anyone know if the next step is a G1x0 or if it's going for multiplicity with the G9x? Smaller die + less power seems to suggest GPU might be going in the same direction of the CPU, that there are limits to what a single die can do and to progress the answer is multi-core designs

[Kunaak]

256 bit?

what in the hell is up with that?

are they trying to de-evolve cards or something?
I thought newer, was suppose to get better... not worse.

[Jimmer411]

Sooooo.....it's an 8800GTS 512MB with a slight overclock and boasts Tri-SLI compatibility.

Thats the entire 9 series so far



So far it looks like both companys were trying to saturate the market with DX10 cards. Focusing on the midrange like they did and closing the huge gap that used to be there was a good idea for both camps. We win

[HKPolice]

No way! It'll have at least 192 shaders and 24 rops!

[Kunaak]

maybe they got it mixed up with the 9600GT.

I mean 256 bit, and 512 ram?

come on.... damn....
what happend?

[TopherTony]

I too think that nvidia is just waiting to see what ATI comes out with. After launching the 8800GTX, 8800 ULTRA, 800 GTS 320, 640, 512, 8800GT 256 AND 512, I doubt that the 9800 gtx and gx2 are the only cards nvidia has up their sleeves.

But I hope I am wrong and ATI can rally! Too much green as of late, I need some red to believe in.

[deathman20]

maybe they got it mixed up with the 9600GT.

I mean 256 bit, and 512 ram?

come on.... damn....
what happend?

ATI is from everything we heard going to use the 256-bit bus as well on the newer R700 cards. Faster memory = less need to up the bus bandwidth to get same results which = less money spent on the card for manufacturing.

[Swatrecon_]

No way! It'll have at least 192 shaders and 24 rops!

omg, that's what it should have

[Eastcoasthandle]

We can't call it crap until we see how it performs...

Anyway, they're merely doing it to combat ATi's efforts. If they weren't comfortable, they'd just push forward and drop a real new lineup instead of playing sit and wait, all the while giving them time to set up to drop on a smaller process size.

That doesn't make sense at all from a business prospective. If they were that comfortable they would just introduce their true next gen video card knowing they would beat whatever the R700 offered through sheer performance, price or both. This would mirror the G80 release. Beside, they already own Ageia and are planning on implementing the PhysX SDK using CUDA (assuming) through it's TWIMTB program. By dropping the true next gen ASAP with Physx (through CUDA) would show confidence in their arch design, captured market share and overall price/performance ratio of their next gen video card including PhysX. But we are not seeing this at all. IMO, it looks more of a reactionary move then proactive one.

Either:
A. They aren't ready
B. Truly concerned of what the R700 offers so they can counter it
C. Bluffing
D. worried about Intel and AMD/ATI as a whole

I guess we have to wait and see...

[Forsaken1]

A,B and D.Le'ts hope those are not the specs.

[flippin_waffles]

I think i'll pick C....

[Swatrecon_]

I think that ATI has a pretty bright future if Nv continues on like this. I really expected them to keep dominating, seeing that they had 2 years to work on a new card. 192 Shaders and 24 ROP's would kickass.

[turtle]

As has been mentioned on chiphell, the memory clock is going to be 2200mhz for the stock config, and by assumption that means it will use .8ns 2400mhz GDDR3 from Hynix...So I imagine we'll see overclocked parts from AIBs with the mem clock closer to 2400mhz effective.

The current G92/G94 chips use 1.0ns 2000mhz GDDR3 iirc.

While I don't like it any more than ya'll, this should be at least some good news for a architecture starved for bandwidth. A 20% increase on paper will make the part look good in reviews, that much can be assumed.

That all being said, it's a pretty weak way to go about things...I imagine anyone that has a 9600gt/3800 and on up will wait for RV770 and the upcoming 512-bit gen from nvidia.

[adamsleath]

ble, so its official 9800gtx is worse then a factory oced 8800ultra...

this is so boring

[takamishanoku]

Have to say was expecting this, but still very dissapointed (from an xtreme sense). However i still think with the new SP meaning that 128 will roughly be 150SP (old), this thing will fly with the faster mem.

Selfishly i'm glad i bought a 8800GTX 2 months ago

[ether.real]

That doesn't make sense at all from a business prospective. If they were that comfortable they would just introduce their true next gen video card knowing they would beat whatever the R700 offered through sheer performance, price or both. This would mirror the G80 release. Beside, they already own Ageia and are planning on implementing the PhysX SDK using CUDA (assuming) through it's TWIMTB program. By dropping the true next gen ASAP with Physx (through CUDA) would show confidence in their arch design, captured market share and overall price/performance ratio of their next gen video card including PhysX. But we are not seeing this at all. IMO, it looks more of a reactionary move then proactive one.

Either:
A. They aren't ready
B. Truly concerned of what the R700 offers so they can counter it
C. Bluffing
D. worried about Intel and AMD/ATI as a whole

I guess we have to wait and see...

In semiconductor manufacturing, you use the old process as long as you can, thanks to the bathtub curve fail rates that come with new processes. This is purely due to lack of competition, nothing else.

[gr8golf]

That doesn't make sense at all from a business prospective. If they were that comfortable they would just introduce their true next gen video card knowing they would beat whatever the R700 offered through sheer performance, price or both. This would mirror the G80 release. Beside, they already own Ageia and are planning on implementing the PhysX SDK using CUDA (assuming) through it's TWIMTB program. By dropping the true next gen ASAP with Physx (through CUDA) would show confidence in their arch design, captured market share and overall price/performance ratio of their next gen video card including PhysX. But we are not seeing this at all. IMO, it looks more of a reactionary move then proactive one.

Either:
A. They aren't ready
B. Truly concerned of what the R700 offers so they can counter it
C. Bluffing
D. worried about Intel and AMD/ATI as a whole

I guess we have to wait and see...

From a business perspective - nobody should ever hold back a superior product in hopes of *future* sales. The only possible motivation for doing so would be if the cost was substantially less for making the current products than the superior ones, or to clear out inventory.

[virtualrain]

I recall a story from about a year ago that the 8800GTX silicon actually had 160 shaders but they were only using 128 to improve yields. An engineer from nVidia actually confirmed this and indicated that the full 160 shaders might be utilized after a die shrink.

So my guess is that the card will have at least 160 shaders, and I'm also guessing there is no reason to go back to 512MB of memory and they will stick with 768MB (also supported by the pics of the card and how long it is).

[Zytek_Fan]

That doesn't make sense at all from a business prospective. If they were that comfortable they would just introduce their true next gen video card knowing they would beat whatever the R700 offered through sheer performance, price or both. This would mirror the G80 release. Beside, they already own Ageia and are planning on implementing the PhysX SDK using CUDA (assuming) through it's TWIMTB program. By dropping the true next gen ASAP with Physx (through CUDA) would show confidence in their arch design, captured market share and overall price/performance ratio of their next gen video card including PhysX. But we are not seeing this at all. IMO, it looks more of a reactionary move then proactive one.

Either:
A. They aren't ready
B. Truly concerned of what the R700 offers so they can counter it
C. Bluffing
D. worried about Intel and AMD/ATI as a whole

I guess we have to wait and see...

QFT!

GeForce 9 series is definitely a *yawn*

[Roger_D25]

maybe they got it mixed up with the 9600GT.

I mean 256 bit, and 512 ram?

come on.... damn....
what happend?

Ditto...

[axan]

bah dissapointing

[dengyong]

The pic doesn't fit the description.

[adamsleath]

So my guess is that the card will have at least 160 shaders, and I'm also guessing there is no reason to go back to 512MB of memory and they will stick with 768MB (also supported by the pics of the card and how long it is).

that's morelike it...