HD 2900XTX Benches

[Munky]

Originally posted by: Cookie Monster
Thats because of
a) different hardware sites use different setups hardware/software configurations
b) different time demos/method of benchmark

Or DT is lying which i suspect they arent because large number of hits is not worth the credibility that they lose in a long term sense.

That's true, but you also have to question DT's results if they are that much different from what other sites get. If in Oblivion DT used a 5 second long run, that would not be nearly enough sample data to reflect accurate performance, so even if DT did get 98fps average for the gtx at 19 x 12, that level of performance would not be sustainable over a longer duration. Moreover when testing with the FRAPS method, good review sites do several runs for each card to remove the outliers, and DT does not mention doing so. The point is, DT might honestly be reporting exactly the numbers they saw, but those numbers could be way off from actual performance due to inadequate testing methods.

 

[Cookie Monster]

Originally posted by: munky

Originally posted by: Cookie Monster
Thats because of
a) different hardware sites use different setups hardware/software configurations
b) different time demos/method of benchmark

Or DT is lying which i suspect they arent because large number of hits is not worth the credibility that they lose in a long term sense.

That's true, but you also have to question DT's results if they are that much different from what other sites get. If in Oblivion DT used a 5 second long run, that would not be nearly enough sample data to reflect accurate performance, so even if DT did get 98fps average for the gtx at 19 x 12, that level of performance would not be sustainable over a longer duration. Moreover when testing with the FRAPS method, good review sites do several runs for each card to remove the outliers, and DT does not mention doing so. The point is, DT might honestly be reporting exactly the numbers they saw, but those numbers could be way off from actual performance due to inadequate testing methods.

That my friend is one of the reasons why i am holding off my judgement based on the current R600 benches.

But i think people are forgetting that AMD themselves said the XT model will not beat the GTX model. Another thing is price, so the XT might compete against the GTS in terms of street price.

We will have to see.

 

[Munky]

Haven't seen this posted yet, so here's the latest from DT re: r600
http://www.dailytech.com/DailyTech+Digest+Making+Sense+of+R600/article7079.htm

 

[Matt2]

Originally posted by: munky
Haven't seen this posted yet, so here's the latest from DT re: r600
http://www.dailytech.com/DailyTech+Digest+Making+Sense+of+R600/article7079.htm

Pretty good explanation. Smart move to make those comments and put out some of the fires burning across tech forums.

Kristopher Kubicki makes a good point about there not being much of a difference between 1GB GDDR4 and 512mb GDDR3 at this time. I know the benchmarks weren't extensive, but so far it looks like the extra 30GB/s of memory bandwidth did absolutely squat for the XTX.

 

[HurleyBird]

They defended their position, but unfortunately didn't apologize for the things that mattered.

There's no way these tests can be taken seriously. And really, you're an idiot (of the clueless fanboy variety) if you do take them seriously.

There is no doubt that the testing was biased. You don't put a water cooled (just check the specs) and OCed video card up against a preproduction model that if anything is running below what stock speed will be when it's released. The only reason to do that is if you *want* one card to win over the other. The proof of bias automatically invalidates *all* of the testing.

 

[cheesecurd]

This is what happens when you buy Bit Boys Oy.

 

[Matt2]

Originally posted by: HurleyBird
They defended their position, but unfortunately didn't apologize for the things that mattered.

There's no way these tests can be taken seriously. And really, you're an idiot (of the clueless fanboy variety) if you do take them seriously.

There is no doubt that the testing was biased. You don't put a water cooled (just check the specs) and OCed video card up against a preproduction model that if anything is running below what stock speed will be when it's released. The only reason to do that is if you *want* one card to win over the other. The proof of bias automatically invalidates *all* of the testing.

Or if the cards werent their's and they had limited time to do the tests like they said 100 times.

 

[HurleyBird]

Originally posted by: Matt2

Originally posted by: HurleyBird
They defended their position, but unfortunately didn't apologize for the things that mattered.

There's no way these tests can be taken seriously. And really, you're an idiot (of the clueless fanboy variety) if you do take them seriously.

There is no doubt that the testing was biased. You don't put a water cooled (just check the specs) and OCed video card up against a preproduction model that if anything is running below what stock speed will be when it's released. The only reason to do that is if you *want* one card to win over the other. The proof of bias automatically invalidates *all* of the testing.

Or if the cards werent their's and they had limited time to do the tests like they said 100 times.

First of all, I would like to see a link to that.

Secondly, if true, they had enough time to install games and FRAPS, but didn't have enough time to install Riva Tuner and lower the 8800 back to stock (ten minutes at most)? Give me a break!

 

[Matt2]

Originally posted by: HurleyBird

Originally posted by: Matt2

Originally posted by: HurleyBird
They defended their position, but unfortunately didn't apologize for the things that mattered.

There's no way these tests can be taken seriously. And really, you're an idiot (of the clueless fanboy variety) if you do take them seriously.

There is no doubt that the testing was biased. You don't put a water cooled (just check the specs) and OCed video card up against a preproduction model that if anything is running below what stock speed will be when it's released. The only reason to do that is if you *want* one card to win over the other. The proof of bias automatically invalidates *all* of the testing.

Or if the cards werent their's and they had limited time to do the tests like they said 100 times.

They had enough time to install games and FRAPS, but didn't have enough time to install riva tuner and lower the 8800 back to stock? Give me a break!

8800GTX Roundup

No where in that article do I see the reference clock GTX getting stomped by 30+ fps like DT shows the XTX by the OC versions.

10-12fps difference at max.

Take your FUD elsewhere. I'm sorry R600 didnt live up to your expectations.

 

[HurleyBird]

Originally posted by: Matt2
No where in that article do I see the reference clock GTX getting stomped by 30+ fps like DT shows the XTX by the OC versions.

10-12fps difference at max.

You don't get it do you? The fact that they use a water cooled 8800GTX puts their entire methodology in question which invalidates the entire testing. The difference between stock and OCed parts is *irrelevant*. This is a very simple concept to grasp. Therefore, I don't beleive you're purposefully spreading FUD (since you don't seem to have the demeanor), but you *are* an idiot. (note: call me names, and I'll call you names)

 

[Matt2]

Originally posted by: HurleyBird

Originally posted by: Matt2
No where in that article do I see the reference clock GTX getting stomped by 30+ fps like DT shows the XTX by the OC versions.

10-12fps difference at max.

You don't get it do you? The fact that they use a water cooled 8800GTX puts their entire methodology in question which invalidates the entire testing. The difference between stock and OCed parts is *irrelevant*. This is a very simple concept to grasp. Therefore, I don't beleive you're purposefully spreading FUD (since you don't seem to have the demeanor), but you *are* an idiot. (note: call me names, and I'll call you names)

1. I didn't call you "names", I said take your FUD somewhere else so I'm done with you.

2. Since ATI doesn't let board partners OC their cards and Nvidia does, I can't wait till you try to "invalidate" Anand's or any other reviewers review because you are crazy if you think that R600 wont be benched against the NUMEROUS OC versions of G80 availabel.

 

[HurleyBird]

Originally posted by: Matt2


1. I didn't call you "names", I said take your FUD somewhere else so I'm done with you.

Which is the equivalent of calling me a lying fanboy. Just because you cloak it doesn't mean you aren't name calling. I don't feel bad about insulting you though, because you obviously deserve it. See below.

Originally posted by: Matt2
2. Since ATI doesn't let board partners OC their cards and Nvidia does, I can't wait till you try to "invalidate" Anand's or any other reviewers review because you are crazy if you think that R600 wont be benched against the NUMEROUS OC versions of G80 availabel.

Type "radeon factory overclocked" in Google. You'll be surprised. See above.

I've also never (until now) seen a water cooled card with insane clocks from one vendor compared to a stock card from the other.

 

[Cookie Monster]

Originally posted by: HurleyBird

Originally posted by: Matt2
No where in that article do I see the reference clock GTX getting stomped by 30+ fps like DT shows the XTX by the OC versions.

10-12fps difference at max.

You don't get it do you? The fact that they use a water cooled 8800GTX puts their entire methodology in question which invalidates the entire testing. The difference between stock and OCed parts is *irrelevant*. This is a very simple concept to grasp. Therefore, I don't beleive you're purposefully spreading FUD (since you don't seem to have the demeanor), but you *are* an idiot. (note: call me names, and I'll call you names)

How is the stock and OCed parts irrelevant? For one, i would agree that they shouldve used a stock GTX, which however still would have beaten the XT and the 1gb model anyway. (according to their test)

Secondly I dont see your logic about cards being water cooled. Water cooled OR not, if the core/mem clocks of the card is same, that it doesnt really matter does it. E.g water cooled G80 at 575/1800 vs stock cooled at 575/1800. You tell the difference in performance from the two models. So since its water cooled, its magically getting more performance from thin air?

They probably benched it since the 8800 ultra is clocked at 650/2350. And its being released on May15th with the tech reviews on 1st. Kind of ironic how the R600 reviews are going to be online during May 2nd.

 

[HurleyBird]

Originally posted by: Cookie Monster
Secondly I dont see your logic about cards being water cooled. Water cooled OR not, if the core/mem clocks of the card is same, that it doesnt really matter does it. E.g water cooled G80 at 575/1800 vs stock cooled at 575/1800.

The fact is, the only card I've been able to find that clocks as high as the one DT used is the ultra expensive (especially if you add the cost of a water cooler) BFG 8800 GTX OC Water Cooled Ed. It would be very hard to attain those clocks with standard air cooling.

Originally posted by: Cookie Monster
How is the stock and OCed parts irrelevant? For one, i would agree that they should've used a stock GTX, which however still would have beaten the XT and the 1gb model anyway. (according to their test)

Exactly, *according to their test*. The entire point is that using such an extremely high clocked GTX vs. an XTX that probably isn't even stock clock (they were able to OC the XT almost 100MHz above the XTX) shows either stupidity or bias. Either way, we can't possibly take *their tests* seriously anymore because the entire testing methodology is put into question by said stupidity or bias.

 

[HurleyBird]

In other news, from the author of the "making sense of R600" article:

Someone "reached out and touched us" today and gave us 2 Radeon XT cards for use in Crossfire and a different driver (350MB ?!?!). Expect benchmarks this week.

350MB? Despite DT's laughable review practices, my interest is peaked.


The same guy also had something else to say, something disturbing...

Where did you read it was a watercooled card? It wasn't.

Now hold on a second! I spent something like an hour and a half searching for 8800 GTX cards that are factory overclocked to 650MHz. There was only *one*, the BFG 8800 GTX OC Water Cooled. Something is *very* fishy here.

Either:

-The guy hasn't seen the GTX they used, and quickly replied to save face (or he's lying).

-They overclocked an air cooled card past it's original factory overclock.

-They reported the frequency wrong in the original article.

-They are pitting the XTX against some unreleased, ultra high clocked, air cooled monster of a GTX, and conveniently forgot to mention it in their original article (as that would, of course, cause much backlash on the forums)

Whatever the reason, none of the above scenario's cast DT's benchmarking practices in good light.

 

[Keysplayr]

I suppose we'll find out after May 2.

 

[Cookie Monster]

Originally posted by: keysplayr2003
I suppose we'll find out after May 2.

Agreed.

And just chill hurley bird. Its like you found out you got cancer today or something

 

[jpeyton]

The "overclocked" 8800 GTX they used is a small bump in memory frequency away from being a stock 8800 Ultra (same core clock, 80MHz slower memory clock).

All I know is that someone (either DailyTech or a lot of critics around the web) are going to be eating crow in a few weeks.

The skepticism being hurled at DailyTech's benchmarks is not petty; it's fairly serious.

 

[erwos]

Originally posted by: jpeyton
The skepticism being hurled at DailyTech's benchmarks is not petty; it's fairly serious.

I agree - but in the sense that the criticisms pass dangerously close to libel territory.

Seems to me it's just a bunch of folks going sour grapes that their preferred GPU provider under-delivered, and did so late. What's the response? Shoot the messenger. Sad.

 

[Zstream]

Originally posted by: SexyK

Originally posted by: apoppin
thanks for the 'catch up'

it seems to me that although DT is sloppy, their conclusions seem in line with what we are hearing

and overall it shouldn't change by much

r600 IS disappointing

Exactly. The DT scores were confusing due to the different individuals benching different cards in different situations and my guess is that probably happened because of a lack of time to play with the cards before publishing the article.

However, I have a serious gripe with Zstream running around posting that the scores are completely made up and flaming anyone who disagrees with him. I promise you that he WILL NOT admit he was wrong if these numbers prove accurate, and he definitely won't apologize to the folks at DT who are working hard so we can have the benefit of an early look at what R600 is going to offer.

The multiple delays, the number of respins, and the marketing doublespeak coming out of AMD all add up to a botched launch. These numbers explain a lot of AMD's previously-unexplainable behavior and, as apoppin said, are in line with all the latest information we have about this chip.

Anyone who still thinks the XTX is going to arrive in stores by mid-May with a huge performance lead over the GTX is just out of touch with reality at this point. We all wish AMD would come out with something groundbreaking, but they aren't this time around and we've just got to deal with that at this point.

Interesting quotes from Tom's HW forums, found from B3D:

Quote:
While I am required to follow the NDA, the stuff up on Daily Tech today is almost worthless. Yes Anandtech was present in Tunisia (signing Non-disclosure agreements like the Inquirer), why they are posting this stuff is beyond me because their numbers are off. They must be only using the XP drivers and OS because the numbers in CF vs the GTX are very much different. So until I can officially comment on the architecture and the performance.. hold all of this as useless until the rest of the world writes about it.


I really would love to comment on this stuff...

I understand that DT and Anand are seperate but that is so childish. Derek was there and his cards got to his place of business before he returned home from Tunisia. That long board they have ... Not what Derek should have gotten in his delivery. That is all I will say before I go too far.

 

[jpeyton]

Two more posts on Rage3D/HardForum:

Originally Posted by loafer87gt
Kombatant, an employee of AMD/ATI, has also said that the benchmarks are bunk and that the card Daily Tech tested was not the XTX, but rather an OEM XT. He also said earlier on that the card will be well worth the wait, and hinted at some sort of surprise for ATI faithful. Not too sure what that is supposed to mean. ??

Originally Posted by Kombatant
They christened the OEM version we've known and loved for quite a few months as an "XTX". That should tell you a lot about their credibility actually.

Even Kris mentioned the following in his blog:

Originally Posted by KristopherKubicki
It would be ludicris to think ATI would ever release the XTX with the clocks that its at right now.

Originally Posted by KristopherKubicki
The XTX and the XT are identical except for the footprint and memory type. It's likely the core clock will change as well, but the core is physically the same.

I think it is highly likely what DailyTech benchmarked will not reflect the performance of the retail XTX. In their eagerness to be the first with XTX benchmarks, they might have mistook an OEM XT (or even an OEM FireGL) as an XTX.

We might see updated, more valid data from DailyTech sooner than later, given that after they posted the controversial XTX benchmarks, we hear the following:

Originally Posted by KristopherKubicki
Someone "reached out and touched us" today and gave us 2 Radeon XT cards for use in Crossfire and a different driver (350MB ?!?!). Expect benchmarks this week.

Sounds like someone at AMD (or someone close to AMD) has quickly provided DailyTech with additional hardware and new drivers almost immediately after the benchmarks were published. I know ATi's driver teams can work quickly, but in less than 48 hours? It's more likely that DailyTech was using an older driver revision.

 

[HurleyBird]

Wow. After the NDAs lift DT is either going to go up in flames, or get a lot more traffic.

 

[Regs]

Well...this is interesting...

 

[Aikouka]

Originally posted by: HurleyBird
The same guy also had something else to say, something disturbing...

Where did you read it was a watercooled card? It wasn't.

Now hold on a second! I spent something like an hour and a half searching for 8800 GTX cards that are factory overclocked to 650MHz. There was only *one*, the BFG 8800 GTX OC Water Cooled. Something is *very* fishy here.

Read the original article with a bit of insight, they never mentioned the name/manufacturer of the card and they seemed reluctant to. There's most likely a reason for this... quite possibly an unreleased card; however, there would be no NDA as DT doesn't sign NDAs. It's quite possible that a vendor gave them a card to use as DT isn't Anandtech and most likely doesn't just have 8800GTXs lying around. However, the card they got was an OC'd model. Frankly, if that's what happened, it's better that we got at least something to compare it to rather than nothing. You might say, "we... well, we could've compared it to other GTX specs!" Wrong. You can't just pick and choose benchmarks from places to compare a card to .

 

[SexyK]

Only time will tell I guess. After all the FUD, lies and general BS that AMD and their "followers" have been spreading about R600 and this launch, though, I tend to believe that the DT benchmarks are valid and represent a close approximation of final performance.