My observations on the infamous R520 benchmarks...

[Chocolate Pi]

As a resident lurker who seldom says anything, I feel the need to speak up; people are actual debating these false benchmark statistics! Bias is clearly showing through, on BOTH sides of the ridiculous fence.

Benchmarks are NOT news items, as some treat them. Even the dinkiest tech sites do benchamrks scientifically. A specific procedure is used with a constant platform, documented standards and controls, all to provide a result that can be reproduced.

But our friends with the R520 benchmarks did none of this. Why, they got the magic numbers in an email from Taiwan! How lucky of them! We should all be glad HA has such deticated friends in Taiwan that they are willing to take the time to carefully find all the components needed to re-create the testbed and laboratory conditions to create an identical copy to the original platform. I'm sure it wasn't that hard though, as we all know computer parts of all types and brands grow on Silicon trees in Taiwan.

So, after HA sent "Taiwan" their top-secret benchmark scripts (that obtain totally different scores from other sites like Andandtech) the ever-deticated pals in Taiwan run these benchmarks on the scientifically re-created testbed at various low resolutions. You'd think faster clock speeds, top-of-the-line memory, and various brand-spanking-new design decisions would make a positive difference, but I'm glad "Taiwan" is here to set the record straight.

 

[compgeek89]

Kinda confusing... you arent carrying a consistent line of thought but I think I get your point.

 

[Chocolate Pi]

My overall message is that scientific benchamrks are very different from numbers got in an email from Taiwan.

 

[compgeek89]

Well Im not looking at anymore "R520 benchmarks" til the card is out. They dont matter til then.

That is... if it ever comes out.

 

[stnicralisk]

dedicated has another d in it im pretty sure

 

[Polish3d]

I'm from the future and the GTX wins

 

[ddogg]

Originally posted by: Frackal
I'm from the future and the GTX wins

lol :thumbsup:

 

[Keysplayr]

Chocolate Pi, about your opinion. It only works in dreamland.

"Benchmarks are NOT news items, as some treat them. Even the dinkiest tech sites do benchamrks scientifically. A specific procedure is used with a constant platform, documented standards and controls, all to provide a result that can be reproduced."

And every freakin review site has variations in there platforms, documented standards, and little if any controls. Designed specifically to confuse us all and envokes constant arguments over which review site is trustworthy with reliable results.
There exists no standards board to dictate how benchmarking should be performed and not guidelines for reviewers to adhere to. Every review site has at least one (minimum) difference to all others.

"But our friends with the R520 benchmarks did none of this. Why, they got the magic numbers in an email from Taiwan! How lucky of them! We should all be glad HA has such deticated friends in Taiwan that they are willing to take the time to carefully find all the components needed to re-create the testbed and laboratory conditions to create an identical copy to the original platform. I'm sure it wasn't that hard though, as we all know computer parts of all types and brands grow on Silicon trees in Taiwan. "

This second paragraph is pure assumption at best. HOW THE HECK do you know anything about how these benchmarks were done? This just sounds petty and very strange.

"So, after HA sent "Taiwan" their top-secret benchmark scripts (that obtain totally different scores from other sites like Andandtech) the ever-deticated pals in Taiwan run these benchmarks on the scientifically re-created testbed at various low resolutions. You'd think faster clock speeds, top-of-the-line memory, and various brand-spanking-new design decisions would make a positive difference, but I'm glad "Taiwan" is here to set the record straight. "

Incredible. Here, go have yourself a good laugh. Sounds like you need it.

Keys

 

[Genx87]

Granted these could be fake or they could be real. Who knows but I wouldnt throw them out just because they dont show ATI in the best light. In shader limited roles in theory ATI loses this round unless they magically come out with a 24 pipe design.

 

[crazydingo]

Originally posted by: keysplayr2003
Chocolate Pi, about your opinion. It only works in dreamland.

Why ? Because it is different from your opinion ? :roll:

Originally posted by: keysplayr2003
And every freakin review site has variations in there platforms, documented standards, and little if any controls. Designed specifically to confuse us all and envokes constant arguments over which review site is trustworthy with reliable results.
There exists no standards board to dictate how benchmarking should be performed and not guidelines for reviewers to adhere to. Every review site has at least one (minimum) difference to all others.

The variation in performance numbers is due to the variation of platform specifications. Replicate any platform bed and you should be able to replicate the numbers of that specific review. Isnt that the purpose of mentioning the "setup" or "testing procedure" part of the reviews ? :roll:

Originally posted by: keysplayr2003
This second paragraph is pure assumption at best. HOW THE HECK do you know anything about how these benchmarks were done? This just sounds petty and very strange.

It is exactly what Sander claimed in his "review" that he got the numbers from a partner and he is not responsible by any length of imagination.

Originally posted by: Genx87
Granted these could be fake or they could be real. Who knows but I wouldnt throw them out just because they dont show ATI in the best light. In shader limited roles in theory ATI loses this round unless they magically come out with a 24 pipe design.

Funny how people make their mind up on pre-release products. :disgust:

Chocolate Pi, you didnt need to create a whole new thread as there are already many threads on topic already.

 

[vision33r]

Well, if they were fake, HA stands to lose reputation a lot more than would ATI lose theirs with poor R520 bench.

If you look at the numbers and look back at ATI's history and recent hooplas one can only say that the scores are in line with expectations which excites many fanboys on both sides.

 

[Genx87]

Funny how people make their mind up on pre-release products.

How did I make up my mind? I am reserving judgement for final reviews. I am making a point that these benchmarks may not be that far off due to shader performance limitations on the R520 given the current information we know.

 

[Drayvn]

Since Sander didnt do the benchmarking that means he ASSUMED the guys in Taiwan did it with an R520.

So pretty much everyone except the Taiwanese is assuming these are correct.

 

[Wreckage]

The benchmarks are fake, because the R520 is fake.

 

[Keysplayr]

Originally posted by: crazydingo

Originally posted by: keysplayr2003
Chocolate Pi, about your opinion. It only works in dreamland.

Why ? Because it is different from your opinion ? :roll:

I would appreciate it if you can show me more than 2 review sites that did identical benchmarks in every conceivable aspect (hardware, software (with all service packs, drivers, settings EXACTLY identical), method , etc etc. right down to the same model numbers of every single part. Now, you tell me how this is an opinion and not fact that this is never done. I think you'll find that you have your work cut out for you.

Originally posted by: keysplayr2003
And every freakin review site has variations in there platforms, documented standards, and little if any controls. Designed specifically to confuse us all and envokes constant arguments over which review site is trustworthy with reliable results.
There exists no standards board to dictate how benchmarking should be performed and not guidelines for reviewers to adhere to. Every review site has at least one (minimum) difference to all others.

The variation in performance numbers is due to the variation of platform specifications. Replicate any platform bed and you should be able to replicate the numbers of that specific review. Isnt that the purpose of mentioning the "setup" or "testing procedure" part of the reviews ? :roll:

Why yes it is, and I would guess this is probably as close as we will ever come to an agreement on something. No two sites have the exact same setups. You make my point for me. Thank you. No amount of eye rolling can change that.

Originally posted by: keysplayr2003
This second paragraph is pure assumption at best. HOW THE HECK do you know anything about how these benchmarks were done? This just sounds petty and very strange.

It is exactly what Sander claimed in his "review" that he got the numbers from a partner and he is not responsible by any length of imagination.

Still does not let us know how the benchmarks were done now does it? I think your just arguing for the sake of arguing at this point. Why, I have no clue.

Originally posted by: Genx87
Granted these could be fake or they could be real. Who knows but I wouldnt throw them out just because they dont show ATI in the best light. In shader limited roles in theory ATI loses this round unless they magically come out with a 24 pipe design.

Funny how people make their mind up on pre-release products. :disgust:

Well, we certainly know what column to place you under now don't we.... You made up your mind that the results are false. Isn't it still a pre-release product?
Or does it only count when it's not you. Hipocrit.

Chocolate Pi, you didnt need to create a whole new thread as there are already many threads on topic already.

 

[crazydingo]

Originally posted by: keysplayr2003
I would appreciate it if you can show me more than 2 review sites that did identical benchmarks in every conceivable aspect (hardware, software (with all service packs, drivers, settings EXACTLY identical), method , etc etc. right down to the same model numbers of every single part. Now, you tell me how this is an opinion and not fact that this is never done. I think you'll find that you have your work cut out for you.[/b]
What is the point of providing the testbed configs by ALL review sites ? It is so that you can by all means get the same set of hardware/software and replicate those results.

Originally posted by: keysplayr2003
And every freakin review site has variations in there platforms, documented standards, and little if any controls. Designed specifically to confuse us all and envokes constant arguments over which review site is trustworthy with reliable results.
There exists no standards board to dictate how benchmarking should be performed and not guidelines for reviewers to adhere to. Every review site has at least one (minimum) difference to all others.

Originally posted by: keysplayr2003
Why yes it is, and I would guess this is probably as close as we will ever come to an agreement on something. No two sites have the exact same setups. You make my point for me. Thank you. No amount of eye rolling can change that.

You said there were no testing standards, I proved my point that they existed. Please read properly.

Originally posted by: keysplayr2003
Well, we certainly know what column to place you under now don't we.... You made up your mind that the results are false. Isn't it still a pre-release product?
Or does it only count when it's not you. Hipocrit.

And we know which column you can be categorised under. I dont believe his numbers because Sanders say that explicitly but you still dont acknowledge that. And thanks for the insults, we know when you cant prove your arguments you resort to them.

 

[ArchAngel777]

Originally posted by: crazydingo

Originally posted by: keysplayr2003
I would appreciate it if you can show me more than 2 review sites that did identical benchmarks in every conceivable aspect (hardware, software (with all service packs, drivers, settings EXACTLY identical), method , etc etc. right down to the same model numbers of every single part. Now, you tell me how this is an opinion and not fact that this is never done. I think you'll find that you have your work cut out for you.[/b]
What is the point of providing the testbed configs by ALL review sites ? It is so that you can by all means get the same set of hardware/software and replicate those results.

Originally posted by: keysplayr2003
And every freakin review site has variations in there platforms, documented standards, and little if any controls. Designed specifically to confuse us all and envokes constant arguments over which review site is trustworthy with reliable results.
There exists no standards board to dictate how benchmarking should be performed and not guidelines for reviewers to adhere to. Every review site has at least one (minimum) difference to all others.

Originally posted by: keysplayr2003
Why yes it is, and I would guess this is probably as close as we will ever come to an agreement on something. No two sites have the exact same setups. You make my point for me. Thank you. No amount of eye rolling can change that.

You said there were no testing standards, I proved my point that they existed. Please read properly.

Originally posted by: keysplayr2003
Well, we certainly know what column to place you under now don't we.... You made up your mind that the results are false. Isn't it still a pre-release product?
Or does it only count when it's not you. Hipocrit.

And we know which column you can be categorised under. I dont believe his numbers because Sanders say that explicitly but you still dont acknowledge that. And thanks for the insults, we know when you cant prove your arguments you resort to them.

Sanders had to cover his bases... I think we know why. ATI might delay the release again due to these benchmarks, just to prove that they are incorrect. So they will delay the process and launch vaporware cards on Oct 5th. Of course, if any do exist, you can bet they are overclocked versions of the cards they were origionally planning. Yes, this is speculation, so save a line in your reply.

I read an article where they are trying to get the card up to 700Mhz... LOL, they are doing the intel approach.

 

[Powermoloch]

Another fanboyism e-penis pissing war. Getting tiring (yawns)

 

[Keysplayr]

Originally posted by: crazydingo

Originally posted by: keysplayr2003
I would appreciate it if you can show me more than 2 review sites that did identical benchmarks in every conceivable aspect (hardware, software (with all service packs, drivers, settings EXACTLY identical), method , etc etc. right down to the same model numbers of every single part. Now, you tell me how this is an opinion and not fact that this is never done. I think you'll find that you have your work cut out for you.[/b]
What is the point of providing the testbed configs by ALL review sites ? It is so that you can by all means get the same set of hardware/software and replicate those results.

Originally posted by: keysplayr2003
And every freakin review site has variations in there platforms, documented standards, and little if any controls. Designed specifically to confuse us all and envokes constant arguments over which review site is trustworthy with reliable results.
There exists no standards board to dictate how benchmarking should be performed and not guidelines for reviewers to adhere to. Every review site has at least one (minimum) difference to all others.

Originally posted by: keysplayr2003
Why yes it is, and I would guess this is probably as close as we will ever come to an agreement on something. No two sites have the exact same setups. You make my point for me. Thank you. No amount of eye rolling can change that.

You said there were no testing standards, I proved my point that they existed. Please read properly.

Originally posted by: keysplayr2003
Well, we certainly know what column to place you under now don't we.... You made up your mind that the results are false. Isn't it still a pre-release product?
Or does it only count when it's not you. Hipocrit.

And we know which column you can be categorised under. I dont believe his numbers because Sanders say that explicitly but you still dont acknowledge that. And thanks for the insults, we know when you cant prove your arguments you resort to them.

What insults? LOL. Are you referring to the hiprocrit remark? You proved nothing except that benchmark standard exist for a SINGLE reviewer. Duh. Of course. Your offtrack here. I am talking about one reviewer to another and you know it. Before we go any further with your circus, can you acknowledge that I have made it clear what I am talking about and that you understand that it is about one reviewer to another? Not just the standards of a single reviewer? Hello?

 

[Keysplayr]

Originally posted by: Powermoloch
Another fanboyism e-penis pissing war. Getting tiring (yawns)

Thanks for the narraration. How bout contributing something?

 

[Powermoloch]

Originally posted by: keysplayr2003

Originally posted by: Powermoloch
Another fanboyism e-penis pissing war. Getting tiring (yawns)

Thanks for the narraration. How bout contributing something?

I'll just wait when other websites posted their take on the cards. WE just can't rely on one source.

again, like I said. It is another fanboyism e-penis pissing war

 

[Keysplayr]

Originally posted by: Powermoloch

Originally posted by: keysplayr2003

Originally posted by: Powermoloch
Another fanboyism e-penis pissing war. Getting tiring (yawns)

Thanks for the narraration. How bout contributing something?

I'll just wait when other websites posted their take on the cards. WE just can't rely on one source.

again, like I said. It is another fanboyism e-penis pissing war

Agreed. And it looks like the OP went back to lurking.

 

[Powermoloch]

Originally posted by: keysplayr2003

Originally posted by: Powermoloch

Originally posted by: keysplayr2003

Originally posted by: Powermoloch
Another fanboyism e-penis pissing war. Getting tiring (yawns)

Thanks for the narraration. How bout contributing something?

I'll just wait when other websites posted their take on the cards. WE just can't rely on one source.

again, like I said. It is another fanboyism e-penis pissing war

Agreed. And it looks like the OP went back to lurking.

:Q !

 

[Duvie]

Originally posted by: Genx87

Funny how people make their mind up on pre-release products.

How did I make up my mind? I am reserving judgement for final reviews. I am making a point that these benchmarks may not be that far off due to shader performance limitations on the R520 given the current information we know.

Do not try to interject any commonsense from things said outside of this article....It only stands to confuse them....

 

[RampantAndroid]

Originally posted by: Powermoloch
Another fanboyism e-penis pissing war. Getting tiring (yawns)

Hmmm, if you guys like, I can go into photoshop, make up a "benchmark" and then email it to you guys saying that I'm from Taiwan and these are benchmarks comparing a ATI 7200 64mb card to nVidia's G135 chip. Up to you.