HD 2900XTX Benches

[chizow]

Originally posted by: munky
What retaking the crown with the 7950gx2? Prior to that the 7900gtx never had the crown in the first place, even against the x1900xtx. The fact that it took two g71 chips to beat a single r580 wasn't too impressive, and if Ati does the same with the r600, then it basically confirms that a single r600 can't match a single g80.

Also, about the whole OC thing, 650mhz is not exactly easy for the 8800gtx. Looking at some oc results here, here, here, here, here, and here, only a minority of the tested cards reached 650 and higher.

Timeline might be a bit off but pretty sure it went 7800 > 1800 > 1900 > 7900 > 7950x2. I wasn't referring to the 7900 retaking the crown since ATI already had it, but ATI didn't take the performance lead from the 7800 until they released the 1900s, after the 7800. 7900 was just a die-shrink with a few enhancements.

And I didn't say hitting 650 on a GTX was easy, although the G80 has been out for 7 months since most of those were written. Its a relatively mature chip now and could easily have improved yields. I'm saying if you're going to assume a 100MHz higher clock on the XTX based on early XT results you can't discount the tested 650MHz GTX. I do agree they should've tested the OC'd XT to the OC'd GTX though.

 

[HurleyBird]

Originally posted by: chizow

Originally posted by: HurleyBird

Oh, I realize that. It's a bit of a different situation though. The XT overclocks by 100MHz on a weaker cooler, so it's pretty rational to claim 50MHz on a stronger cooler as reasonable. Claiming that the XTX will launch at 850MHz is silly though (even though it may be possible depending on differences between tested silicon and production silicon, if any).

The GTX can't hit higher clock speeds on a stock cooler because of safe working limits but we can expect the XTX to? Apples to apples.....

Ugh, this is why I find it so hard to not make ad hominem attacks. Just reading your posts makes me shudder involuntarily.

Obviously the 8800GTX can clock higher on a stock cooler because manufacturers have released factory OCed parts at speeds of up to 620MHz on said cooler. I've never claimed otherwise. Saying that the 8800GTX can be factory OCed to 650MHz on said cooler is probably approaching absurdity though, else Nvidia wouldn't need such a massive cooler on the Ultra.

In regards to the XTX, claiming that it will be released at 850MHz (since the XT overclocked to that level) is also silly. However, a more modest number, say 800MHz (which is both my estimate, and DT's lower end estimate), is a perfectly acceptable estimate. Note that I'm not claiming that the XTX will be released at 800MHz, I'm estimating that number based on a modest 50% of the XT overclock + a bigger cooler.

I very much hope that you can finally see the logic in this reasoning (I'm sure everyone else does) and I don't need to pound it into your head further.

 

[chizow]

Originally posted by: HurleyBird

Originally posted by: chizow

Originally posted by: HurleyBird

Oh, I realize that. It's a bit of a different situation though. The XT overclocks by 100MHz on a weaker cooler, so it's pretty rational to claim 50MHz on a stronger cooler as reasonable. Claiming that the XTX will launch at 850MHz is silly though (even though it may be possible depending on differences between tested silicon and production silicon, if any).

The GTX can't hit higher clock speeds on a stock cooler because of safe working limits but we can expect the XTX to? Apples to apples.....

Ugh, this is why I find it so hard to not make ad hominem attacks. Just reading your posts makes me shudder involuntarily.

Obviously the 8800GTX can clock higher on a stock cooler because manufacturers have released factory OCed parts at speeds of up to 620MHz on said cooler. I've never claimed otherwise. Saying that the 8800GTX can be factory OCed to 650MHz on said cooler is probably approaching absurdity though, else Nvidia wouldn't need such a massive cooler on the Ultra.

So what exactly are you assuming here? That the cooler used on the Ultra is something special that NV invented in order to get that extra 24MHz? Or that there's no way the 650MHz GTX can exist because that extra 24MHz would be crazy talk?

Either way you'd be making the wrong assumption. That cooler isn't anything new or special, EVGA has used it on their ACS3 parts since launch as have a few others (Sparkle and Gainward I believe).

In regards to the XTX, claiming that it will be released at 850MHz (since the XT overclocked to that level) is also silly. However, a more modest number, say 800MHz (which is both my estimate, and DT's lower end estimate), is a perfectly acceptable estimate. Note that I'm not claiming that the XTX will be released at 800MHz, I'm estimating that number based on a modest 50% of the XT overclock + a bigger cooler.

Right, totally rational to draw your own conclusions but totally discount others.

I very much hope that you can finally see the logic in this reasoning (I'm sure everyone else does) and I don't need to pound it into your head further.

Its really easy for me to avoid ad hominem attacks btw, especially now that I know you're "shuddering involuntarily" while I'm sitting here lol'ing.

 

[ss284]

Originally posted by: HurleyBird

Originally posted by: chizow

Originally posted by: HurleyBird

Oh, I realize that. It's a bit of a different situation though. The XT overclocks by 100MHz on a weaker cooler, so it's pretty rational to claim 50MHz on a stronger cooler as reasonable. Claiming that the XTX will launch at 850MHz is silly though (even though it may be possible depending on differences between tested silicon and production silicon, if any).

The GTX can't hit higher clock speeds on a stock cooler because of safe working limits but we can expect the XTX to? Apples to apples.....

Ugh, this is why I find it so hard to not make ad hominem attacks. Just reading your posts makes me shudder involuntarily.

Obviously the 8800GTX can clock higher on a stock cooler because manufacturers have released factory OCed parts at speeds of up to 620MHz on said cooler. I've never claimed otherwise. Saying that the 8800GTX can be factory OCed to 650MHz on said cooler is probably approaching absurdity though, else Nvidia wouldn't need such a massive cooler on the Ultra.

In regards to the XTX, claiming that it will be released at 850MHz (since the XT overclocked to that level) is also silly. However, a more modest number, say 800MHz (which is both my estimate, and DT's lower end estimate), is a perfectly acceptable estimate. Note that I'm not claiming that the XTX will be released at 800MHz, I'm estimating that number based on a modest 50% of the XT overclock + a bigger cooler.

I very much hope that you can finally see the logic in this reasoning (I'm sure everyone else does) and I don't need to pound it into your head further.

Too bad the 8800GTX at stock will still beat the overclocked XTX. 3Dfx 4 lyfe~

 

[swtethan]

http://www.overclock.net/nvidia/136207-official-8800gtx-gts-oc-thread.html

660/1220 (2440mhz) on STOCK cooling for a GTX

 

[Nightmare225]

I find the ATI fanboyism on this thread disturbing.
Come on, guys. Admit it, your precious R600 was a failure, even if it does perform better than these benchmarks, it was still half a year late.
This happens a lot, and neither company is immune. NVIDIA had it with the NV30, and it looks like ATI will have it with the R600.

 

[aka1nas]

Nightmare, I wouldn't say that it's at all the same. I owned an FX 5900, it was a crappy design that scaled poorly and had a very short usable lifetime before it's design deficiencies made it unusable in Dx9 games.

As far as we know at this point, there's nothing fundamentally wrong with the R600 other than it would have been nice to have had a chance to wrap our sweaty hands on one 6 months ago. The XT looks to be an excellent performer in current games at it's price point and we don't yet have any reason to believe that it will be a poor performer in Dx10 titles.

So far, the R600s issue was AMD's ability to actually execute on it and deliver it in a timely manner. I wouldn't consider something to be an NV30-esque disaster unless it has a correspondingly severe design flaw to go along with a failure to execute.

 

[Matt2]

Originally posted by: Nightmare225
I find the ATI fanboyism on this thread disturbing.
Come on, guys. Admit it, your precious R600 was a failure, even if it does perform better than these benchmarks, it was still half a year late.
This happens a lot, and neither company is immune. NVIDIA had it with the NV30, and it looks like ATI will have it with the R600.

Agreed.

I don't understand what the big deal is. Why is everyone trying so hard to debunk these benches? I mean, if R600 does turn out to be a flop, would it really kill you guys to buy an Nvidia card?

DailyTech is not exactly the Inq and they are getting these numbers from their own guys, it's not like the Inq getting info from some random Chinese site.

No matter what the methodology of the benches is, or what card they were run against, OC or not, the numbers are what they are. Not impressive.

 

[HurleyBird]

Originally posted by: chizow
So what exactly are you assuming here? That the cooler used on the Ultra is something special that NV invented in order to get that extra 24MHz? Or that there's no way the 650MHz GTX can exist because that extra 24MHz would be crazy talk?

Either way you'd be making the wrong assumption. That cooler isn't anything new or special, EVGA has used it on their ACS3 parts since launch as have a few others (Sparkle and Gainward I believe).

Again you need to read. I never claimed that Nvidia invented the ACS3 cooler. Other manufacturers have used the ACS3 cooler over the stock cooler for their OC parts because it is more efficient and thus makes it easier to clock parts higher. However, it should be obvious to anyone that if Nvidia had been comfortable releasing the Ultra with a stock GTX cooler they would have. It's simple economics. Obviously, Nvidia wasn't and had to resort to using the ACS3. Therefore, it's bloody obvious you aren't going to see a GTX OC edition card with the stock cooler that's running at Ultra levels. If you can't see the reasoning here, you are, quite simply, an idiot.

Originally posted by: chizow
Right, totally rational to draw your own conclusions but totally discount others.

Everyone's free to make estimates. If an estimate is out to lunch, I have the right to debunk it, which I did. However, just because I debunk another's estimate doesn't mean that my estimates are also out to lunch. If you disagree with my estimate, do so, and tell me why (based on your observations) you think it's wrong. What you're doing now is calling my estimate wrong because I disagree with someone else's, and that's a *very* stupid thing to say.

 

[HurleyBird]

Speed round!

Since I'm very annoyed right now, the gloves come off!

Originally posted by: swtethan
http://www.overclock.net/nvidia/136207-official-8800gtx-gts-oc-thread.html
660/1220 (2440mhz) on STOCK cooling for a GTX

We've just finished explaining how OC ability does not equal release ability. It's disturbing exactly how resilient you are to shedding your ignorance.

Originally posted by: Nightmare225
I find the ATI fanboyism on this thread disturbing.

I find all fanboyism disturbing. You are an Nvidia fanboy who would be complaining if a preview with this many inconsistencies popped up for an Nvidia card, no doubt. I guess I find you disturbing too.

Originally posted by: Matt2
No matter what the methodology of the benches is, or what card they were run against, OC or not, the numbers are what they are. Not impressive.

This quote really speaks for itself.

 

[Matt2]

Originally posted by: HurleyBird

Originally posted by: chizow
So what exactly are you assuming here? That the cooler used on the Ultra is something special that NV invented in order to get that extra 24MHz? Or that there's no way the 650MHz GTX can exist because that extra 24MHz would be crazy talk?

Either way you'd be making the wrong assumption. That cooler isn't anything new or special, EVGA has used it on their ACS3 parts since launch as have a few others (Sparkle and Gainward I believe).

Again you need to read. I never claimed that Nvidia invented the ACS3 cooler. Other manufacturers have used the ACS3 cooler over the stock cooler for their OC parts because it is more efficient and thus makes it easier to clock parts higher. However, it should be obvious to anyone that if Nvidia had been comfortable releasing the Ultra with a stock GTX cooler they would have. It's simple economics. Obviously, Nvidia wasn't and had to resort to using the ACS3. Therefore, it's bloody obvious you aren't going to see a GTX OC edition card with the stock cooler that's running at Ultra levels. If you can't see the reasoning here, you are, quite simply, an idiot.

Originally posted by: chizow
Right, totally rational to draw your own conclusions but totally discount others.

Everyone's free to make estimates. If an estimate is out to lunch, I have the right to debunk it, which I did. However, just because I debunk another's estimate doesn't mean that my estimates are also out to lunch. If you disagree with my estimate, do so, and tell me why (based on your observations) you think it's wrong. What you're doing now is calling my estimate wrong because I disagree with someone else's, and that's a *very* stupid thing to say.

Stop calling people idiots. You reek of fanboyism and you obviously have ten little flame throwers for fingers.

How about you get to 200 posts before you become the most annoying person on this board.

 

[Nightmare225]

Originally posted by: HurleyBird

Originally posted by: Nightmare225
I find the ATI fanboyism on this thread disturbing.

I find all fanboyism disturbing. You are an Nvidia fanboy who would be complaining if a preview with this many inconsistencies popped up for an Nvidia card, no doubt. I guess I find you disturbing too.

My last card was a 9700 Pro which I held onto for three years. I was amazed at its performance and loved it, it was an upgrade from an abysmal FX 5600. I resolved to upgrade for the DX10 generation and was blown away by the G80's performance. I even decided to wait a bit for R600, but it just got delayed way too much and I decided to screw it and get a G80. Never been this happy with a card since R300. I was hopeful that ATI would put the hurting on NVIDIA, and am still ready to sell my card on eBay. But, the fact that these preliminary benches show off abysmal performance for a card that comes out 6 months later... No, only the most diehard fanboys would excuse ATI for such a terrible performance... :|

Now, give me a solid reason why you aren't a fanboy.

 

[swtethan]

Originally posted by: Nightmare225

Originally posted by: HurleyBird

Originally posted by: Nightmare225
I find the ATI fanboyism on this thread disturbing.

I find all fanboyism disturbing. You are an Nvidia fanboy who would be complaining if a preview with this many inconsistencies popped up for an Nvidia card, no doubt. I guess I find you disturbing too.

My last card was a 9700 Pro which I held onto for three years. I was amazed at its performance and loved it, it was an upgrade from an abysmal FX 5600. I resolved to upgrade for the DX10 generation and was blown away by the G80's performance. I even decided to wait a bit for R600, but it just got delayed way too much and I decided to screw it and get a G80. Never been this happy with a card since R300. I was hopeful that ATI would put the hurting on NVIDIA, and am still ready to sell my card on eBay. But, the fact that these preliminary benches show off abysmal performance for a card that comes out 6 months later... No, only the most diehard fanboys would excuse ATI for such a terrible performance... :|

Now, give me a solid reason why you aren't a fanboy.

my last card was a x800xl no fanboy here

 

[BlizzardOne]

or how about we wait for some proper benchmarks, instead of going off of these, questionable at best, results and proclaiming it game over?

with such little time for DT to 'test' the card(s), in haste they could have missed something very simple, which skewed the results in a major way.

we won't know until more thorough data is out there.

 

[Cookie Monster]

They could unleash something along the lines of HD 2900X2 for that matter.

One thing that looks like a big limiting factor is that the R600 has 16TMUs. Even though its clocked at 740mhz, games still make heavy use of the TMUs. Compared to R580, which had 16 TMUs @ 650mhz this could be one reason why the R600 doesnt perform much better than the X1950XTX in some given scenario.

However, its a given SOMETHING is bottlenecking the R600, but the funny thing is that when it comes to GPGPU things, i do believe the R600 kill anything in those apps. Maybe this was the big decision AMD made, and focused more on the GPGPU side of things instead of gaming performance where video hardware should always be prioritized first.

 

[Nightmare225]

Originally posted by: BlizzardOne
or how about we wait for some proper benchmarks, instead of going off of these, questionable at best, results and proclaiming it game over?

with such little time for DT to 'test' the card(s), in haste they could have missed something very simple, which skewed the results in a major way.

we won't know until more thorough data is out there.

I think some of us are through "waiting" :thumbsdown: :|

 

[Janooo]

Originally posted by: Cookie Monster
...

However, its a given SOMETHING is bottlenecking the R600, but the funny thing is that when it comes to GPGPU things, i do believe the R600 kill anything in those apps. Maybe this was the big decision AMD made, and focused more on the GPGPU side of things instead of gaming performance where video hardware should always be prioritized first.

You could be right. ATI took a risk with R580 design and it paid off. It appears that they took it a step further with vec5 instead of scalar. Now, are they trying to be future proof and better prepared for DX10? This is a big question and only time will tell.

 

[HurleyBird]

Originally posted by: Matt2

Originally posted by: HurleyBird

Originally posted by: chizow
So what exactly are you assuming here? That the cooler used on the Ultra is something special that NV invented in order to get that extra 24MHz? Or that there's no way the 650MHz GTX can exist because that extra 24MHz would be crazy talk?

Either way you'd be making the wrong assumption. That cooler isn't anything new or special, EVGA has used it on their ACS3 parts since launch as have a few others (Sparkle and Gainward I believe).

Again you need to read. I never claimed that Nvidia invented the ACS3 cooler. Other manufacturers have used the ACS3 cooler over the stock cooler for their OC parts because it is more efficient and thus makes it easier to clock parts higher. However, it should be obvious to anyone that if Nvidia had been comfortable releasing the Ultra with a stock GTX cooler they would have. It's simple economics. Obviously, Nvidia wasn't and had to resort to using the ACS3. Therefore, it's bloody obvious you aren't going to see a GTX OC edition card with the stock cooler that's running at Ultra levels. If you can't see the reasoning here, you are, quite simply, an idiot.

Originally posted by: chizow
Right, totally rational to draw your own conclusions but totally discount others.

Everyone's free to make estimates. If an estimate is out to lunch, I have the right to debunk it, which I did. However, just because I debunk another's estimate doesn't mean that my estimates are also out to lunch. If you disagree with my estimate, do so, and tell me why (based on your observations) you think it's wrong. What you're doing now is calling my estimate wrong because I disagree with someone else's, and that's a *very* stupid thing to say.

Stop calling people idiots. You reek of fanboyism and you obviously have ten little flame throwers for fingers.

How about you get to 200 posts before you become the most annoying person on this board.

And you do nothing but talk, while never contributing any evidence to your claims. You're telling me my estimates are wrong or out to lunch, but never exactly *why*. I call them like I see them. I have no qualms with calling idiots people idiots when they act the part. I usually even give them three tries to stop being idiots before I say anything derogatory.

 

[fern420]

any news on an actual crossfire benchmark?

 

[apoppin]

Originally posted by: HurleyBird
Too many emoticons, apoppin. Like some preteen on MSN...

is that the very best you can do?
:thumbsdown:

and *who* is under NDA that is calling the DT numbers BS?


================

and dreddfunk ... i DO believe we have some really bitter AMD former fans posting here now

 

[yacoub]

The one thing that keeps me waiting is knowing that 320MB will not be enough memory to handle high res gaming with AA/AF on, and the 640MB GTS isn't quite to my price range ($350 before any rebates is the max I'd pay and I'd like to see it go even lower).
I also am curious what NVidia will announce for its refresh in a few months because often times the high-mid range refresh card is the one to get. With the G70 series the 7800GT was meh and the 7800GTX was expensive. Then the 7900 refresh series came out an the 7900GT had the performance of the 7800GTX for the price of a 7800GT. That's the card I got (7900GT KO) and I love it. It makes me willing to see what NVidia will do for its G80 refresh, especially since I want a card with at least 512MB of memory on it.

The other factor is that ATi's cards allow for dynamic fan throttling with temperature (via cool programs like ATi Tool which I love) and that is a feature I dearly miss from my X800 XL Ultimate (Sapphire card w/ Zalman cooler on it stock). It was so quiet even when gaming, where as my 7900GT of course has a fan control that amounts to an on/off switch with like 2-3 levels of fan speed at the most and basically it just turns the fan "on" to 100% whenever a 3D app is loaded. I'd much prefer the ATi style of fan control and hope that either NVidia gets their act together and enables that for their cards or ATi puts out a card that's appealing enough to get me to go with ATi instead of NVidia.

 

[apoppin]

i thought the 7800GS had more fan control than that

are you sure?

using the CP or Rivatuner?

and my x1950p is on high - all the time ... my fault for *maxing* my games
-i just haven't tried to OC it yet

 

[HurleyBird]

Originally posted by: apoppin

Originally posted by: HurleyBird
Too many emoticons, apoppin. Like some preteen on MSN...

is that the very best you can do?
:thumbsdown:

and *who* is under NDA that is calling the DT numbers BS?


and dreddfunk ... i DO believe we have some really bitter AMD former fans posting here now

Heh, I guess there's nothing actually wrong about going emoticon crazy... just is a bit annoying seeing them in every single sentence.

As for NDA guys, there's a poster on Toms' and an ATI guy at, I beleive it was Rage 3D.

Regarding bitter former AMD fans, I suppose there's probably a few, but there's also a lot of people who can't trust the DT numbers as authoritative due to many reasons.

Really, the only responsible thing to do right now is to wait until the NDA lifts.

 

[HurleyBird]

Originally posted by: Nightmare225

Originally posted by: HurleyBird

Originally posted by: Nightmare225
I find the ATI fanboyism on this thread disturbing.

I find all fanboyism disturbing. You are an Nvidia fanboy who would be complaining if a preview with this many inconsistencies popped up for an Nvidia card, no doubt. I guess I find you disturbing too.

My last card was a 9700 Pro which I held onto for three years. I was amazed at its performance and loved it, it was an upgrade from an abysmal FX 5600. I resolved to upgrade for the DX10 generation and was blown away by the G80's performance. I even decided to wait a bit for R600, but it just got delayed way too much and I decided to screw it and get a G80. Never been this happy with a card since R300. I was hopeful that ATI would put the hurting on NVIDIA, and am still ready to sell my card on eBay. But, the fact that these preliminary benches show off abysmal performance for a card that comes out 6 months later... No, only the most diehard fanboys would excuse ATI for such a terrible performance... :|

Now, give me a solid reason why you aren't a fanboy.

Ok, so aren't a fanboy. You have my deepest apologies. However, you do need to grasp that when a preview comes up that shows some bias, poor testing methods, and other people under NDA call it bunk, it might just be ok to put that preview into question. So far I haven't seen any 'ATI fanboys' (correct me if I'm wrong) claiming anything other than "wait until NDAs break, the DT preview might well be wrong" which is a perfectly sane point of view given the circumstances.

As for me proving I'm not a fanboy, well, I'll openly admit that I like Nvidia and AMD a little bit more than I like Intel and ATI, but not to the extent that I would call myself a fanboy. I'm just as critical of each company, but I only really feel the urge to chime in whenever I see mass ignorance.

 

[Matt2]

Originally posted by: HurleyBird
And you do nothing but talk, while never contributing any evidence to your claims. You're telling me my estimates are wrong or out to lunch, but never exactly *why*. I call them like I see them. I have no qualms with calling idiots people idiots when they act the part. I usually even give them three tries to stop being idiots before I say anything derogatory

Fair enough. You're an idiot.