If ATI Won a Round, Would Anyone Notice?

[cozwin]

yep these scores are pure bull, first why using diff processor speeds :O :?, and 2nd i get 11k with my 8800gts 320mb might be even higher now and thats with a overclocked e6300

[fornowagain]

http://www.computerbase.de/artikel/h...hnitt_3dmark06

dunno if these results are common, but computerbase usually is a pretty credibale source.

*edit* ah well, just noticed the testsystem was running windows xp. maybe not that compareable to the results posted in this thread, as they were achieved on vista.

http://www.bit-tech.net/hardware/200...a_superclock/2

On Vista.

[Eastcoasthandle]

http://www.bit-tech.net/hardware/200...a_superclock/2

On Vista.

Actaully those were under XP. For example they clearly state a scheduled DX 10 patch for COH. Another example is the Supreme Commander retial game using patch 3220 which is clearly for XP and lower OS. Oblivion is using patch 1.2 which is for XP. And most of the other games benched (at that time) were running at frame rates only found under XP, not Vista...which is a dead give away.

[Chewbenator]

Well I think one thing is clear. The 2900 seems to be beating the GTS when it comes to dual GPU performance.

[n91htmare]

NO, the only thing that's clear is that PC Mag does not know didly squat about reviewing computers. PC Mag is the mainstream of reviews like your Best Buy is the mainstream of computer. Everyone goes there to buy a computer but no one there knows anything about computers. Just like that, people go to PC Mag to look at reviews but no one there knows how to even do benching.

[DilTech]

Actaully those were under XP. For example they clearly state a scheduled DX 10 patch for COH. Another example is the Supreme Commander retial game using patch 3220 which is clearly not for XP and lower OS. Oblivion is using patch 1.2 which is for XP. And most of the other games benched (at that time) were running at frame rates only found under XP, not Vista...which is a dead give away.

Vista.

http://www.bit-tech.net/hardware/200...a_superclock/2

Intel Core 2 Extreme X6800 (operating at 2.93GHz -- 11x266MHz); Asus Striker Extreme motherboard (nForce 680i SLI); 2 x 1GB Corsair XMS2-8500C5 (operating in dual channel at DDR2-800 with 4-4-4-12-1T timings); Seagate Barracuda 7200.9 200GB SATA hard drive; Enermax Galaxy DXX 1000W PSU; Windows Vista Ultimate x86; Nvidia nForce standalone drivers version 15.00 WHQL.

Another dead give-away is that 158.18 wasn't made for windows XP, XP had 158.19. Take a look at guru3d's driver archive yourself.

[Eastcoasthandle]

NO, the only thing that's clear is that PC Mag does not know didly squat about reviewing computers. PC Mag is the mainstream of reviews like your Best Buy is the mainstream of computer. Everyone goes there to buy a computer but no one there knows anything about computers. Just like that, people go to PC Mag to look at reviews but no one there knows how to even do benching.

Actually, Chewbenator is correct when it comes to Vista. Unless there are other r600 vs gts 640 SLI reviews using Vista that we can look at?

[californian7856]

NO, the only thing that's clear is that PC Mag does not know didly squat about reviewing computers. PC Mag is the mainstream of reviews like your Best Buy is the mainstream of computer. Everyone goes there to buy a computer but no one there knows anything about computers. Just like that, people go to PC Mag to look at reviews but no one there knows how to even do benching.

i see the ATI fans still think that 2900XT > anything on the market

and nv fans still think that the test was done without turning on SLI

but make no mistake about it had they use 158.45 the result would have been different

[Eastcoasthandle]

here is a vista review
As you can see, even in 3 games it's performing better in Vista then GTS 320.

[Phlash]

[DilTech]

What's there to drool over? That's essentially CF vs Single card GTX.

[Eastcoasthandle]

What's there to drool over? That's essentially CF vs Single card GTX.

Not in the Vista, 3 game review that i posted
Also, (unless there is other reviews that suggest otherwise) the GTX SLI benchmark in Vista is pretty much on par what it can do in Vista.


Source

If the above is any indication the GTX is not scoring any where near the same as it does in XP. Therefore, (again unless there are other reviews using SLI vs R600 in Vista) there is no way to assume nor expect XP performance in Vista via SLI. Using similar gpu/cpu clocks and hardware.

[Phlash]

What's there to drool over? That's essentially CF vs Single card GTX.

Whoops, I just fell for the trap then

[nitteo]

Those numbers are right for the since the OC was only up to 3.2,thats QX6700+SLI GTX, I have Ultras, but close enough:

REMEMBER THIS IS STOCK CLOCKS ON BOTH CPU+GPUs

Bench 1: 1st Single Ultra A = 11,930

Bench 2: 2nd Single Ultra B = 12,013

Bench 3: Both on SLI = 12,981

Bench 4: XP QUAD OCD to 3600mhz + GPUs OCD = 18,303

Bottom line, the STOCK CPU severely bottlenecks the GPUs. This is just a "sneaky" way of manipulating the data.

[Eastcoasthandle]

Those numbers are right for the STOCK QX6700+SLI GTX, I have Ultras, but close enough:

REMEMBER THIS IS STOCK CLOCKS ON BOTH CPU+GPUs

Bench 1: 1st Single Ultra A = 11,930

Bench 2: 2nd Single Ultra B = 12,013

Bench 3: Both on SLI = 12,981

In vista above

Bottom line, the STOCK CPU severely bottlenecks the GPUs. This is just a "sneaky" way of manipulating the data.

How is it sneaky We are talking about the difference between Vista vs XP. There is no specific guideline suggesting that you have to be at a certain OC Also, results do vary from one system to another...

[DilTech]

Not in the Vista, 3 game review that i posted
Also, (unless there is other reviews that suggest otherwise) the GTX SLI benchmark in Vista is pretty much on par what it can do in Vista.


Source

If the above is any indication the GTX is not scoring any where near the same as it does in XP. Therefore, (again unless there are other reviews using SLI vs R600 in Vista) there is no way to assume nor expect XP performance in Vista via SLI. Using similar gpu/cpu clocks and hardware.

Care to explain how the dell, with a significantly slower cpu speed, scored higher in vista in 06 then the Falcon? Allow me to explain it, drivers. You see, PCMagazine doesn't say what drivers are used... Judging by it being that much lower though, I think it's pretty safe to say they're running MUCH older drivers.

To the guy who posted his sli numbers, try it again with the new 162's and see what you get.

[Eastcoasthandle]

Care to explain how the dell, with a significantly slower cpu speed, scored consistantly higher in vista then the Falcon? Allow me to explain it, drivers. You see, PCMagazine doesn't say what drivers are used...

To the guy who posted his sli numbers, try it again with the new 162's and see what you get.

There are plenty of variables not known into why the score is different.
-Bios settings
-Type of ram used
-Vista setup
-Motherboard used
-drivers and how they react to that particular PC setup
-the CPU could be THROTTLING
-etc

All can play a factor in why it's true that "results will vary from one PC to another". But let us not get off track here. Someone already posted his results and found them to be similar in Vista as to what's posted in the article making the "driver problem" moot argument. Also note, that drivers for Vista (regardless of what you believe) will not create XP like results. But as you can see, the Nvidia drivers are doing what it's suppose to do in XP. Currently, there are no forceware drivers that allow for XP type performance gains in Vista (again, using similar gpu/cpu, etc testbed). This also lends to the comments of how strong G80 is in XP vs vista. However, I reserve my opinion until we see more G80 reviews in Vista.

[DilTech]

There are plenty of variables not known into why the score is different.
-Bios settings
-Type of ram used
-Vista setup
-Motherboard used
-drivers and how they react to that particular PC setup
-etc

All can play a factor in why it's true that "results will vary from one PC to another". But let us not get off track here. Someone already posted his results and found them to be similar in Vista as to what's posted in the article making the "driver problem" moot argument. Also note, that drivers for Vista (regardless of what you believe) will not create XP like results. But as you can see, the Nvidia drivers are doing what it's suppose to do in XP. Currently, there are no forceware drivers that allow for XP type performance gains in Vista (again, using similar gpu/cpu, etc testbed). This also lends to the comments of how strong G80 is in XP vs vista. However, I reserve my opinion until we see more G80 reviews in Vista.

162.15 for Vista is apparently pretty strong, and improves sli support in vista.

However, it is a shame that nvidia's vista support isn't as strong as their xp support, but I think that's due to their initial instabilities in vista. They made a statement in march that they're first focusing on stability(thus the influx of drivers in the past few months), then they're moving on to performance. I know 162.15 added ~5 fps in COJ according to users at NVNews. In f.e.a.r. the 162.15's actually brought single card performance up to XP levels according to users on the same board(actually, it out perfomed the 158 series xp drivers in f.e.a.r.)

[zerazax]

Either way there were some very questionable review tactics in the whole thing

[Eastcoasthandle]

162.15 for Vista is apparently pretty strong, and improves sli support in vista.

However, it is a shame that nvidia's vista support isn't as strong as their xp support, but I think that's due to their initial instabilities in vista. They made a statement in march that they're first focusing on stability(thus the influx of drivers in the past few months), then they're moving on to performance. I know 162.15 added ~5 fps in COJ according to users at NVNews. In f.e.a.r. the 162.15's actually brought single card performance up to XP levels according to users on the same board(actually, it out perfomed the 158 series xp drivers in f.e.a.r.)

User opinion can vary from one PC to another. Those FPS increased could have come at the cost of a certain clock speed (cpu/gpu using lns, dry ice, etc), Bios tweaking (PCIe frequency for one example) and ram tweak (lower timings) etc to name a few. So I take that with a grain of salt. However, in this thread, when there is consistency across the board (sort of speak when a user gets similar results found in the article) in this example it's hard to ignore. And as you can clearly see, the G80 (along with it's drivers) do appear to have a perfect fit in performance gains in XP. The same cannot be said in Vista and, I do believe it's not just a driver issue either. But time will tell

[nemrod]

REMEMBER THIS IS STOCK CLOCKS ON BOTH CPU+GPUs

Bench 1: 1st Single Ultra A =
Bench 2: 2nd Single Ultra B =
Bench 3: Both on SLI =
Bench 4: XP QUAD OCD to 3600mhz + GPUs OCD =

Bench 5: Vista QUAD OCD to 3600mhz + stock SLI = ?

[DilTech]

Take a good look at nitteo's results. Peer very closely at the SM2 scores. Notice something? They don't change between single card and sli... Why?

Because with that driver SLi isn't working in sm2. The proof is right there infront of you. Unless, of course, you want to tell me that 3dmark06 bottlenecks the sm2 score at ~4980 on a QX6700.

That's why I asked him to run it again with the 162.15's.

[Eastcoasthandle]

Take a good look at nitteo's results. Peer very closely at the SM2 scores. Notice something? They don't change between single card and sli... Why?

Because with that driver SLi isn't working in sm2. The proof is right there infront of you. Unless, of course, you want to tell me that 3dmark06 bottlenecks the sm2 score at ~4980 on a QX6700.

That's why I asked him to run it again with the 162.15's.

And how can you prove that its a driver problem or a video card problem? It can be either or. Because of that you are either looking at a:
-driver problem that still doesn't prove it's capable of XP performance
-video card limitation in Vista
Regardless, as it stands now the article has been proven correct with what's currently available. No need to call foul as it can be said that the G80 has been out well over 7 months now. If it was just released (like 2 months ago) I would understand. But as it stands now that's no excuse. Also note, that his SM 3.0 results improved with his SLI however, it appears to be at the cost of SM 2.0 (hw limitation??). In any case you cannot just look at SM 2.0 only when using Vista. You cannot say that SLI is not enabled when he's getting a nice boost in SM 3.0. In fact it looks to me that his SLI is working in Vista. What's happening is that in Vista his SM 3.0 increases under SLI in Vista but his SM 2.0 decreases.

[DilTech]

And how can you prove that its a driver problem or a video card problem? It can be either or. Because of that you are either looking at a:
-driver problem that still doesn't prove it's capable of XP performance
-video card limitation in Vista
Regardless, as it stands now the article has been proven correct with what's currently available. No need to call foul as it can be said that the G80 has been out well over 7 months now. If it was just released (like 2 months ago) I would understand. But as it stands now that's no excuse.

Thus, why I again say, lets have him run it again with the 162.15's. Aren't you the least bit curious as to what the outcome will be with the latest drivers? I know I'm generally curious to always see hardware run on the latest available(and always dismissing reviews that don't state the driver versions run).

[Eastcoasthandle]

Thus, why I again say, lets have him run it again with the 162.15's. Aren't you the least bit curious as to what the outcome will be with the latest drivers? I know I'm generally curious to always see hardware run on the latest available(and always dismissing reviews that don't state the driver versions run).

Please read
This is why I express the importants of WHQL approved drivers when benchmarking like this.

driver has been taken offline at the request of NVIDIA. There are some isssues with these drivers.