R520 benchmarks - Hardware Analysis

[Regeneration]

OK...

Let's just wait until the NDA will expire.

 

[WaltC]

The thing that upsets me is that suddenly claims pop up from ATI staff that make all sorts of claims.

Since you yourself popped up with benchmark results they dispute, why should this surprise you at all? Heh...

I'm sure Chris said what he's quoted to have said, he usually blurts out all kinds of stuff, some true, some absolutely false, I have yet to receive and email from him. We talked last week on the phone so he has my cell and knows I'll pick it up. But how about these other claims, who is being quoted? The person serving the coffee at ATI HQ? The person that stapling the leaflets that go out to AIBs that detail the specs for the R520? Or is it Dave Orton's nephew having a go at things?

Regards,

Sander Sassen
http://www.hardwareanalysis.com

While it's gratifying to know that you have a cell phone (so do I, btw), it would appear from the many unanswered questions you ask about what goes on internally at ATi that you have very little knowledge at all about the company.

Thus, I am not at all surprised to find that ATi disputes your report...

 

[ssassen]

WaltC,

I'm 100% with you, I don't know what was tested exactly, I don't have the board, don't have pictures, etc. All that I asked for is if they could get me some benchmarks using our benchmark scripts on the R520 architecture. What they came back with were scores said to respresent that of an X1800XT and a X1800Pro, no more, no less. I know that's far (very far) from what we'd normally post, hence I apply a disclaimer, as I have no way to verify these, they could even be extrapolated from a different clockspeed R520 that was available at the AIB (X1600 or X1300 perhaps?), it is all guesswork and hence these benchmarks are preliminary. Yet I found them interesting enough to share as I was not bound by an NDA and motivated by an ongoing tiff in which ATI stated and I quote 'you can write what you like'.

Regards,

Sander Sassen
http://www.hardwareanalysis.com

 

[BruteForce170]

Of course! Of course it is humiliating and ridiculous to have one's published material exposed as a fraud!

I'm curious... how has any of this been exposed as a fraud? certainly the underhanded digs and the denials from the ati camp suffer from the same lack of substance. at this point it's nothing more than "he said", "she said", hearsay....

if nothing else sanders' statements have more substance, as the inq has stated third party sources as confirming the #'s are in the ballpark, tho may not be reflective of production hardware.

while i have always taken the inq reports with a "grain of salt", they still provide more substance/backing than what the "anti-sander" crowd has.....

it certainly would not surprise me one bit if we find those #'s in the general ballpark, even if slightly slower as it's on pre-release...

 

[Ryano]

That is the point, it seems to me. The people who are in possession of real production-grade 520 products know that the numbers Sanders got are not representative of the 520 products. How do they know? Because they have the hardware themselves and their results are much different.

Man, I sure wish I had benchmarks like you. I'd so totally tell that sander guy off. I mean it's like he's trying to give out info that he has. A hardware website shouldn't do that, they should repackage leet pr packages and make cool graphs. I mean, this info couldn't possibly be pre-production R520 (not even XT as he says as much) results from a AIB, it's totally unpossible!

 

[digitalwanderer]

I need to register, but what for? so you could abuse my account? I've heard about you, Mr. DW (aka Mighty Abuser).

Yet you joined anyways, go figure.

I wouldn't worry, DPJ seems to have taken a liking towards you...and I did promise him I'd let him keep a pet there if he found one he liked.

 

[WaltC]

WaltC,

I'm 100% with you, I don't know what was tested exactly, I don't have the board, don't have pictures, etc. All that I asked for is if they could get me some benchmarks using our benchmark scripts on the R520 architecture. What they came back with were scores said to respresent that of an X1800XT and a X1800Pro, no more, no less. I know that's far (very far) from what we'd normally post, hence I apply a disclaimer, as I have no way to verify these, they could even be extrapolated from a different clockspeed R520 that was available at the AIB (X1600 or X1300 perhaps?), it is all guesswork and hence these benchmarks are preliminary. Yet I found them interesting enough to share as I was not bound by an NDA and motivated by an ongoing tiff in which ATI stated and I quote 'you can write what you like'.

Regards,

Sander Sassen
http://www.hardwareanalysis.com

OK, that's what I thought might be the case all along...you really have no idea who had what hardware, the numbers reported to you being entirely beside the point. I have but one further possibility to add to your conjectures and suppositions as to the source of the numbers presented to you: they could be wholly fictitious. I'm not sure what's really "interesting," though, about 3rd-party reports of this nature which you are completely unable to verify. Maybe it's just me, but being unable to even corroborate for myself that R520 hardware was even in fact used would itself make such reports very uninteresting.

 

[digitalwanderer]

I'm curious... how has any of this been exposed as a fraud? certainly the underhanded digs and the denials from the ati camp suffer from the same lack of substance. at this point it's nothing more than "he said", "she said", hearsay....

if nothing else sanders' statements have more substance, as the inq has stated third party sources as confirming the #'s are in the ballpark, tho may not be reflective of production hardware.

while i have always taken the inq reports with a "grain of salt", they still provide more substance/backing than what the "anti-sander" crowd has.....

it certainly would not surprise me one bit if we find those #'s in the general ballpark, even if slightly slower as it's on pre-release...

ATi has said it's bullshit, ATi has said no partners have production grade boards to bench with; that is nothing?

Man, tough crowd here.

 

[Ryano]

ATi has said no partners have production grade boards to bench with; that is nothing?

Yep, he clearly says what was used wasn't a production board. So that statement is nothing.

 

[WaltC]

Man, I sure wish I had benchmarks like you. I'd so totally tell that sander guy off. I mean it's like he's trying to give out info that he has. A hardware website shouldn't do that, they should repackage leet pr packages and make cool graphs. I mean, this info couldn't possibly be pre-production R520 (not even XT as he says as much) results from a AIB, it's totally unpossible!

Thanks for informing me that I'm in possession of R520 benchmarks as I wasn't aware of it... I guess and suppose that if I had a web site you'd think I was honor bound to publish as fact any and all information emailed to me even if I could verify none of it. I must tell you frankly, though, that if I had such a web site I would never do that.

Heh... Anything at all "is possible" but what is possible is of no personal concern to me. What actually happens that I can verify is of great concern to me, however. But if I am unable to verify that something happened, then as far as I am concerned the event did not occur, whether or not it might have been "possible." Many things are possible, but only certain specific events actually occur. It is only those which interest me.

 

[ssassen]

OK, that's what I thought might be the case all along...you really have no idea who had what hardware, the numbers reported to you being entirely beside the point. I have but one further possibility to add to your conjectures and suppositions as to the source of the numbers presented to you: they could be wholly fictitious. I'm not sure what's really "interesting," though, about 3rd-party reports of this nature which you are completely unable to verify. Maybe it's just me, but being unable to even corroborate for myself that R520 hardware was even in fact used would itself make such reports very uninteresting.

This normally isn't something we'd bother with either as it indeed borders on being a rumor if you have to classify it. In my defense however this is a reputable source, that genuinely seemed able to help us with accurate scores. If you've read the 2nd post in this thread you know the full story behind this. In ATI's defense however these scores could be way off, but as already mentioned 3rd parties, Fudo at TheINQ for example, have said the scores to be close, but a bit low. In hindsight, would I do it again? Probably not. To satisfy your curiosity, it wasn't for the hits though or see the dust flying when the server had a busy afternoon, it was a mixed bag as I pointed out in the 2nd post. Maybe I should've just posted the scores and not write up that conclusion, or leave out the intro or both. Considering I received many emails of support from fellow journalists that are also caught in a fix with ATI I'm hoping this has helped ease their pain somewhat. Would the benchmarks looked any differently when I left all of that out though? And would the exposure been less, or maybe more?

Regards,

Sander Sassen
http://www.hardwareanalysis.com

 

[WaltC]

Yep, he clearly says what was used wasn't a production board. So that statement is nothing.

Since the only boards people will buy will be production boards, then testing a pre-production board has no value. What he told me just a couple of posts ago, however, is that he has no idea what board was used...or even if it was a 520 board of any description.

 

[Ryano]

What? He basically says as much. He thinks they're R520 architecture benchmarks and probably will be close to the retail stuff. I don't see what the big fuss is.

Reading is fundamental.

 

[Junkstyle]

Am I the only person that's impressed by these suspicious benches? I dont know if they are made up or actually representative of the 520 but I was impressed.

Put it in perspective. A 16pipe part competing neck to neck with a 24pipe part. ATI is already at 90nm. There next part will probably have 24pipes or more and that will be crazy fast. Also, the fact that they are going with 512MB ram...wow we are truly entering a new era of graphics cards.

 

[Ryano]

Since the only boards people will buy will be production boards, then testing a pre-production board has no value. What he told me just a couple of posts ago, however, is that he has no idea what board was used...or even if it was a 520 board of any description.

That's not even close to the point of the article. You know outside of the hardware industry people are _encouraged_ to dig for info. They don't wait for the company to authorize and stamp their articles.

 

[Headstone]

That's not even close to the point of the article. You know outside of the hardware industry people are _encouraged_ to dig for info. They don't wait for the company to authorize and stamp their articles.

Does that mean that you think that there are no NDAs outside the hardware (computer) industry? If so you are way off the mark

 

[Ryano]

Am I the only person that's impressed by these suspicious benches? I dont know if they are made up or actually representative of the 520 but I was impressed.

Put it in perspective. A 16pipe part competing neck to neck with a 24pipe part. ATI is already at 90nm. There next part will probably have 24pipes or more and that will be crazy fast. Also, the fact that they are going with 512MB ram...wow we are truly entering a new era of graphics cards.

I agree. If they'd been releasing fast like nvidia a 24pipe part would be out that'd be just awesome. Competition = cheaper upgrade for me

 

[Ryano]

Does that mean that you think that there are no NDAs outside the hardware (computer) industry? If so you are way off the mark

That's not the point. He didn't have a NDA.

 

[Headstone]

That's not the point. He didn't have a NDA.

Nor did he have a card. And so it is all speculation. Really it is a lot like if you being a car entusiast and had a friend who worked at Ford. You ask him to test drive the new hot model and tell you the results. He being in a capacity that only you know (could be in the secretarial pool) test said car. Turns out the car was only a prototype and yet gives you the data as being production ready...

 

[ssassen]

Headstone,

Read the 2nd post of this thread in full? It says pre-production, I never claim to have production model scores. If everybody would just read prior to commenting this thread would be (a) a lot shorter and (b) filled with more sensible comments and (c) actually go somewhere without the same questions being asked over and over again.

Regards,

Sander Sassen
http://www.hardwareanalysis.com