8800GTS 640mb = 8800GT 512mb

[infrared]

Yeah, it's mostly a voltage limitation I think. I only gained 26mhz by removing the IHS and using a decent cooler, even though the temperatures plumeted by about 20c overall.

Just waiting to get the vmods done on my card to confirm this

 

[Grings]

A3 revision G80 GTS's can go over 700mhz, but a2 ones 675-680 is about the best you can get on stock cooling

 

[mab1376]

yeah mine peaks at 675MHz without a volt mod.

 

[Thinker_145]

I think that if the GT performs better than the GTS on a single card setup it would also follow that 2 GT's will perform better than 2 GTS 640MB...

Err I was talking about 24" LCD with AA and that's 12x10 with no AA.

And could you link to the full review and what is the shader clock on the GTS?I would also like to see what happens with 2xAA in crysis.

 

[AddSub]

Interestingly enough, in that HardwareCanucks review of 8800GTS 640MB SSC (the 112 SP one) in Dx10 tests, (G80)8800GTS beats (G92)8800GT in 9 out of 10 tests. I guess with equal amount of SP's (112 vs. 112), higher than standard clocks and wider memory bus, 8800GTS 640MB is not that bad.

 

[Thinker_145]

And that crysis benchmark posted simply cant be the highest tweaked DX9 settings.They mean XP very high and that performance is just not possible.

 

[DarkMatter]

And that crysis benchmark posted simply cant be the highest tweaked DX9 settings.They mean XP very high and that performance is just not possible.

Yep, it is.

 

[gOJDO]

A3 revision G80 GTS's can go over 700mhz, but a2 ones 675-680 is about the best you can get on stock cooling

Nope, almost all A2's are limited to 620MHz, while most of A3 are limited below 680MHz.

There was an article by X-bit about OC-ing an A2: http://www.xbitlabs.com/articles/video/display/gf8800gts320MB-roundup_9.html

Surprisingly, each of the six cards could overclock its main GPU frequency to the same point, to 621MHz. A frequency range of 621-648MHz seems to be the limit for the revision A2 G80 chip. This time neither card could work at a frequency of 648MHz for more than a couple of hours.

I had an Asus EN8800GTS 320MB A2 at that time and mine was also crippled at 621MHz. My friend had XFX with A3 chip, which was artifact free at 680MHz.

 

[Thinker_145]

Yep, it is.

No it's not.

That is easily the performance numbers of high settings.They are using the cheap version of very high or something.

 

[Xaser04]

No it's not.

That is easily the performance numbers of high settings.They are using the cheap version of very high or something.

I take it you havn't heard of the Very high settings hack for xp then? It basically gives you the very high settings from vista (albeit without the dx10 lighting tweaks) in XP but with better performance.

This is how I play crysis on my G92 GTS.

Also that sort of performance is quite possible given the low resolution and high clock on the GT. (Crysis is very shader hungry and the G92 GT has more shader horsepower than the G80 GTS thanks to more stream processors).

I get around 40fps average with my G92 GTS with the very high hack @ 1680x1050. Please note though shadows have to be dropped to medium quality as they absolutely cripple the performance (even noticed this on my old 8800GTX).

 

[Tatty_Two]

You can look at benchmarks running GPU's at stock speeds all you want but what really matters is how they perform when overcloked.

Some cards are intentionally underclocked to show better numbers in benchmarks for a more expensive product.Some cards are clocked a little higher at stock to show better numbers in benchmarks than your rivals card.

This site gave both the GTS and GT a beefy overclock and the results show that both cards are identical with no winner.One wins some and loses some.The other wins some and loses some.Both cards are equal in terms of performance.And the overclock given to both cards is certainly fair as to their OC capability.

Now unless someone else can prove otherwise it just seems to me that the whole world of PC hardware is just wrong about this issue.

http://xtreview.com/review219.htm

I trust myself more than a single review, I had a voltmodded Gainward 8800GTS 640MB clocked at 770mhz on the core and then I bought an Asus 8800GT 512MB G92 funnily enough clocked to 770mhz on the core, the difference here being that I didnt need to risk a voltmod to get the G92 to those speeds.

In every game I played at MY resolutions of 16xx x 10xx the 8800GT gave me more frames, in many, significantly more frames but to be fair, there were some where there was no visible difference on the eye.....is there a HUGE differfence, probably not but just as importantly, the G92 8800GT cost about 60% of what the 640MB G80 GTS cost at the time of the 8800GT's release.

 

[JRMBelgium]

http://forums.guru3d.com/showthread.php?t=243024

 

[DarkMatter]

No it's not.

That is easily the performance numbers of high settings.They are using the cheap version of very high or something.

Are you kidding? With those clocks and at 1280x1024 0xAA you could get 50+ in high settings. But that's only on XP. You can't compare it to Vista numbers. There's sometimes as much as 20% penalty on Vista.

It's worth noting that I get those numbers ONLY if I have AVG anti-virus closed. I don't know of other anti-viruses, but AVG dramatically reduces Crysis performance in my rig. I didn't notice this on other games, but since Crysis is the only game where I need as much power as I can get, it's the only one where I tried this.

 

[AddSub]

Wait, are we talking about the standard (G80)GTS or (G80)GTS SSC (one with 112 SP's like 8800GT instead of 96)? Because both in HardwareCanucks and FiringSquad reviews of the SSC card, the SSC card outperforms a 8800GT.

 

[Thinker_145]

http://forums.guru3d.com/showthread.php?t=243024

So it's a review from a poster in some forums who "upgraded" from a 640GTS to 8800GT.Ya i'll take the xtreview over that.And interesting that not a single game is running any sort of AA.Benchmarks without AA should be discounted.

 

[Thinker_145]

Wait, are we talking about the standard (G80)GTS or (G80)GTS SSC (one with 112 SP's like 8800GT instead of 96)? Because both in HardwareCanucks and FiringSquad reviews of the SSC card, the SSC card outperforms a 8800GT.

The SSC card is unquestionably better than 8800GT.The review I have posted is showing the standard 96SP GTS being on par with the GT.

 

[ktr]

Expreview did a G80 vs G92 roundup few days ago...good reading.

http://en.expreview.com/2008/04/02/g80-vs-g92-hi-end-geforce-gfx-cards-round-up/

according to them. the 8800GTS 640MB = 9600GT 512MB.

 

[Thinker_145]

Are you kidding? With those clocks and at 1280x1024 0xAA you could get 50+ in high settings. But that's only on XP. You can't compare it to Vista numbers. There's sometimes as much as 20% penalty on Vista.

It's worth noting that I get those numbers ONLY if I have AVG anti-virus closed. I don't know of other anti-viruses, but AVG dramatically reduces Crysis performance in my rig. I didn't notice this on other games, but since Crysis is the only game where I need as much power as I can get, it's the only one where I tried this.

You do know that there are different versions of this XP very high setting.There are some cheap versions where they leave out some not so noticible details.

I have the as close to DX10 very high tweak in XP and I just ran the GPU benchmark at 12x10 with 16xAF and I got 20FPS.There is no way he is running all DX9 very high settings.

 

[Thinker_145]

Expreview did a G80 vs G92 roundup few days ago...good reading.

http://en.expreview.com/2008/04/02/g80-vs-g92-hi-end-geforce-gfx-cards-round-up/

according to them. the 8800GTS 640MB = 9600GT 512MB.

Yes I have seen that and all the cards are at stock.......

 

[AddSub]

In Crysis, old G80 GTS (at least the 96 SP version) will perform worse due to lower shader clocks, at least in lower-to-mid res with no AA/AF around to spoil the fun for bandwidth starved G92. Same goes for Oblivion which is also heavily dependent on shader clocks. Just check out a few latest 9800GTX reviews, in many of them they use 8800GT and old 8800GTS cards.

But like I said, the 112 SP version of old GTS outperforms 8800GT in quite a few tests, as HardwareCanucks and FiringSquad reviews of the SSC card show.

 

[Tatty_Two]

Wait, are we talking about the standard (G80)GTS or (G80)GTS SSC (one with 112 SP's like 8800GT instead of 96)? Because both in HardwareCanucks and FiringSquad reviews of the SSC card, the SSC card outperforms a 8800GT.

As far as I was aware, we were talking of overclocked cards based on standard architecture.

 

[Tatty_Two]

So it's a review from a poster in some forums who "upgraded" from a 640GTS to 8800GT.Ya i'll take the xtreview over that.And interesting that not a single game is running any sort of AA.Benchmarks without AA should be discounted.

But as I posted, so did I and generally found the 8800GT to be a little better in my resolution, not by a great deal but as i said, thats not so important, the GT cost just 60% of the G80 640MB price so why would anyone wanny but the 640? EVERYONE judges a card by it's "bang for buck", as in performance divided by cost, in that respect the 8800GT was on it's release a fair bit ahead.

 

[CrackerJack]

boy, your just pissing in everybodys cheerios!