8800gtx preview

[Elfear]

Originally posted by: Matt2

We all know there will be speed penalties when using dual cards in Crossfire/SLI. What I mean by that is that you'll never get 2x single card performance.

Take into consideration that these benches show G80 >90% in Hl2 and Quake 4, I doubt that Crossfired X1900XTXs are 90% faster than a single X1900XTX. probably more like 60-70% depending on resolution and settings (If you get Crossfire working properly).

Every generation we also see GPUs become more efficient. While you're X1900s might get slowdowns, it is very possible that G80 can handle higher resolutions without as much overhead.

I agree with you wholeheartedly. I'm just basing my opinion off of other rumors we've heard as well us the benchmarks that sparked this threrad. As others have noted, the results for the X1950XTX look off compared to the results obtained by other reviewers. That could of course be the result of using a different timedemo, but it still makes me somewhat leary. Going off other results, which say "30% faster than X1950XTX" and "50-70% faster than 7900GTX", it would appear like the 8800GTX <= to Crossfired X1900XTs. It's all up in the air at this point, I just don't see the 8800GTX being faster than a pair of X1900XT's at 720/840.

That being said, don't think for a minute that I don't hope the 8800GTX will spank my setup. I'm not trying to downplay the 8800GTX so that I can feel better about my own rig. I'd much rather have one card than two if the performance is the same, and I plan on switching if the card lives up to the hype.

 

[Matt2]

Ever since the 30% in 3dmark06, I thought that number was a bit low. Every game benchmark, true or not shows a lot higher than 30%.

Plus I've never seen DT lie out of their teeth. They did say they ran the benches themselves. I'd be more suspicious if those numbers came from the Inq or somewhere less credible.

 

[Narse]

Originally posted by: PC Surgeon
Maybe its just me, but that benchmark for HL2 looks a little low on the AMD-ATi side. I could be mistaken, but I'm pretty sure I get better than 80FPS @ the same settings. oh well, even the best make mistakes .

at 1920x1200 4xAA and High quality AF I get 97fps average with my x1950xtx, those scores for the 1950 are way way low.

 

[Matt2]

Originally posted by: Narse

Originally posted by: PC Surgeon
Maybe its just me, but that benchmark for HL2 looks a little low on the AMD-ATi side. I could be mistaken, but I'm pretty sure I get better than 80FPS @ the same settings. oh well, even the best make mistakes .

at 1920x1600 4xAA and High quality AF I get 97fps average with my x1950xtx, those scores for the 1950 are way way low.

they're probably using a timedemo for consistant results. There's no way you can know that you would get 97fps avg in the exact same timedemo as they used.

 

[Corporate Thug]

Originally posted by: Cookie Monster

Originally posted by: Corporate Thug

Originally posted by: josh6079
Those scores look really degrading to the X1950XTX.

Here Anandtech scored ~20 more than they did with the same card at a higher resolution with the same amount of AA in Quake 4.

i think those scores are very suspicious. a 1950XTX only puts up 34 frames in quake 4 @ 1600x1200 w/4x AA? i dont think so...

DT =/= AT.

Remember, different sites use different methods, settings and timedemos.

They could for one haved used 16xHQ AF along with TRAA/AAA as well in a different timedemo.

But it looks like G80 is on 80nm.

Link

While ATI prepares to roll out five 80nm grahics processor units (GPUs) in November, Nvidia plans to speed up its 80nm-production schedule of by rolling out revised versions of its G72 and G73 chips to compete with the rival, according to a Chinese-language Commercial Times report.

The revised versions of G73 (codenamed G73-B1) and G72, along with the highly anticipated G80, will be targeted to compete with ATI solutions from mid-range to high-end market segments, the paper noted.

This could explain the lower power consumption.

wouldnt they state that they were using 16xAF?

 

[ArchAngel777]

Originally posted by: Narse

Originally posted by: PC Surgeon
Maybe its just me, but that benchmark for HL2 looks a little low on the AMD-ATi side. I could be mistaken, but I'm pretty sure I get better than 80FPS @ the same settings. oh well, even the best make mistakes .

at 1920x1600 4xAA and High quality AF I get 97fps average with my x1950xtx, those scores for the 1950 are way way low.

That is a very odd resolution. I have never seen it used before.

 

[Narse]

Originally posted by: Matt2

Originally posted by: Narse

Originally posted by: PC Surgeon
Maybe its just me, but that benchmark for HL2 looks a little low on the AMD-ATi side. I could be mistaken, but I'm pretty sure I get better than 80FPS @ the same settings. oh well, even the best make mistakes .

at 1920x1600 4xAA and High quality AF I get 97fps average with my x1950xtx, those scores for the 1950 are way way low.

they're probably using a timedemo for consistant results. There's no way you can know that you would get 97fps avg in the exact same timedemo as they used.

I can see that, I wish they would state what demo, I am using the one built into lost coast.

 

[Narse]

Originally posted by: ArchAngel777

Originally posted by: Narse

Originally posted by: PC Surgeon
Maybe its just me, but that benchmark for HL2 looks a little low on the AMD-ATi side. I could be mistaken, but I'm pretty sure I get better than 80FPS @ the same settings. oh well, even the best make mistakes .

at 1920x1600 4xAA and High quality AF I get 97fps average with my x1950xtx, those scores for the 1950 are way way low.

That is a very odd resolution. I have never seen it used before.

Thas cause I am an idiot and should proof read my posts, its 1920x1200

 

[ArchAngel777]

Originally posted by: Narse

Originally posted by: ArchAngel777

Originally posted by: Narse

Originally posted by: PC Surgeon
Maybe its just me, but that benchmark for HL2 looks a little low on the AMD-ATi side. I could be mistaken, but I'm pretty sure I get better than 80FPS @ the same settings. oh well, even the best make mistakes .

at 1920x1600 4xAA and High quality AF I get 97fps average with my x1950xtx, those scores for the 1950 are way way low.

That is a very odd resolution. I have never seen it used before.

Thas cause I am an idiot and should proof read my posts, its 1920x1200

No biggy, I wasn't sure these days because I so many people use custom resolutions. I thought maybe there was some 5:4 panel released that supposed that insane resolution.

Edit ** Grrr.. fixing the quotes...

 

[Dethfrumbelo]

Yeah, it all depends on what demo they're using and how heavy the shading, poly count, AI, and physics are in that particular run. So long as they're consistent.

 

[Smartazz]

This is definatly the kind of performance I expected from the 8800GTX. The power consumption is way lower than I thought it would be.

 

[Wreckage]

Originally posted by: Elfear

Originally posted by: Centurin
If you want the new Ati card, you'll be waiting more than a month or two. More like 4 or 5.

Latest rumor has R600 coming out the last week of January.

Really? Inq keeps saying February or March. We won't know for sure until next year I guess.

 

[Nelsieus]

Wow, I must say, performance was quite impressive, and seeing the power-consumption was a big relief, too. Things are about to get exciting (can't wait until Wednesday!).

Also, before the ATI aquisition took place, R600 was scheduled for an early 2007 release (Jan). But right now I wouldn't be surprised if it's closer to March or April before R600 finally sees the light of day.

Remember when G70 released and some people wanted to wait for R520 instead. It ended up being an unexpected 5 month wait, and in the end, the performance was nearly identical. Same thing with G71. After R580 launched, some people wanted to wait for G71 thinking it was going to be a 32 piped monster. Even though it came out just a few months later, performance was, again, nearly identical.

So hopefully people take that into consideration when they ask themselves if they should wait or not.

Nelsieus

 

[Rommel44]

Originally posted by: Wreckage

Originally posted by: Elfear

Originally posted by: Centurin
If you want the new Ati card, you'll be waiting more than a month or two. More like 4 or 5.

Latest rumor has R600 coming out the last week of January.

Really? Inq keeps saying February or March. We won't know for sure until next year I guess.

I wouldn´t mind waiting another month or two BUT only if it means 65nm proces and so higher frequency and lower power requirements.

 

[Munky]

The whole waiting thing also depends a lot on the price and availability of the 8800gtx. If this goes like the 7800gtx launch, then people will be less likely to wait. OTOH, if this turns out to be a rerun of the 7800gtx-512 debacle, then there will be a much more compelling reason to wait.

 

[Nelsieus]

Originally posted by: Rommel44

Originally posted by: Wreckage

Originally posted by: Elfear

Originally posted by: Centurin
If you want the new Ati card, you'll be waiting more than a month or two. More like 4 or 5.

Latest rumor has R600 coming out the last week of January.

Really? Inq keeps saying February or March. We won't know for sure until next year I guess.

I wouldn´t mind waiting another month or two BUT only if it means 65nm proces and so higher frequency and lower power requirements.

65nm is highly unlikely. If they went with 65nm, it would be closer to the middle of next year before it released.

I think 65nm got thrown around as a hopeful rumor, but I'd be surprised if it were true (at this point).

Also, R600 will have a few more things going against it in the thermal / power area, so I wouldn't be surprised if it ended up consuming more compared to Geforce 8, just like R520/R580 did compared to G70/G71.

But then again, ATI has been hush-hush lately on R600. Either their waiting to surprise us and catch nVidia off guard, or trying to hide something. I'll be optimistic and choose the former.

Nelsieus

 

[Nelsieus]

Originally posted by: munky
The whole waiting thing also depends a lot on the price and availability of the 8800gtx. If this goes like the 7800gtx launch, then people will be less likely to wait. OTOH, if this turns out to be a rerun of the 7800gtx-512 debacle, then there will be a much more compelling reason to wait.

Agreed.

Nelsieus

 

[Skott]

I'm glad they mentioned power and amperage specs but they didnt mention actual length which I was hoping to see.

 

[Childs]

Originally posted by: josh6079
Those scores look really degrading to the X1950XTX.

Here Anandtech scored ~20 more than they did with the same card at a higher resolution with the same amount of AA in Quake 4.

You have to remember AT's review used X6800 (2.93Ghz), while DT used a QX6700 (I think 2.66Ghz). I think its reasonable to assume that a 8800GTX in a X6800 would be even faster.

 

[Cabages]

Will this case be able to fit this?

Is the GS version any smaller?

 

[TanisHalfElven]

Originally posted by: munky

Originally posted by: nitromullet

Originally posted by: munky
In case nobody noticed, the rumored power consumption specs always turn out to be blown out of proportion compared to the actual power consumption. It's good to see the g80 not using ridiculuusly a lot of power, even my 430 watt psu would easily handle it.

I agree, but why the dual 6-pin molex plugs if it only draws 4% more juice than an X1950XT?

I read somewhere that the second one is optional, like on the 6800u. I just cant remember where I saw it...

But even so, with a quad core cpu and the g80 drawing ~350 watts at load for the whole system is not too bad at all.

*edited for the grammer polise

the spelling police could also use a tip off.

 

[TanisHalfElven]

i just have to add boooooooooooooooooooo for DT to use old games like Q4 or HL2. we need benches for fear AND mainly for oblivion.

 

[KeithTalent]

Originally posted by: tanishalfelven

Originally posted by: munky

Originally posted by: nitromullet

Originally posted by: munky
In case nobody noticed, the rumored power consumption specs always turn out to be blown out of proportion compared to the actual power consumption. It's good to see the g80 not using ridiculuusly a lot of power, even my 430 watt psu would easily handle it.

I agree, but why the dual 6-pin molex plugs if it only draws 4% more juice than an X1950XT?

I read somewhere that the second one is optional, like on the 6800u. I just cant remember where I saw it...

But even so, with a quad core cpu and the g80 drawing ~350 watts at load for the whole system is not too bad at all.

*edited for the grammer polise

the spelling police could also use a tip off.

LMAO :laugh:

 

[Eagle26]

does the 8800GTS require 2 pci-e connectors????

 

[Eagle26]

does the 8800GTS require 2 pci-e connectors?????