Why did AT and THG test R350 on 1 platform?

[KGB1]

AT chose the PIV 3.06 + 850E (why not a Granite Bay?)
THG chose a Nforce 2 with a Athlon XP 2700+ (why not a Barton 3000+?)

I feel this is not how you make a benchmark on a certain graphics card. I'd understand if a person were testing the sub system when a HDD/CDROM/NIC on a system with PCI slots... UDMA 100. Pretty standard on that part.
But when it comes to graphics its important to test it out on 2 platforms. Not to judge how good the platform is.. but how a certain graphics card performs on a particular system

I don't get it.. I mean its so hard to throw in the card into a second testing machine. I'm taking these benchmarks with a grain of salt. :|

I just hope they get their acts straight when a FULL review comes out.

 

[paralazarguer]

AT chose the PIV 3.06 + 850E (why not a Granite Bay?)

Because RD ram is still faster than dual DDR. That's a fact.

Full benchmark? What are you talking about?

 

[Smilin]

Because they only had a couple days between when they received the cards and when the NDA expired that allowed them to write the articles.

 

[Spicedaddy]

I'm taking these benchmarks with a grain of salt.

LOL

 

[NFS4]

Originally posted by: Smilin
Because they only had a couple days between when they received the cards and when the NDA expired that allowed them to write the articles.

Ain't that the truth!! They haven't had these cards for very long and they have tons of reviews to put out. Anand was working on NV34, NV31, Radeon 9800 Pro, a new motherboard chipset, a new hard drive, and a few other things at one time.

If you think it's so easy to throw the card in another system and get right to work, then so be it. I honestly don't think you're being fair at all with this.

 

[PrinceXizor]

And obviously, you know nothing about benchmarking or systematically examining a single component. When you are comparing features, performance, benchmarking, call it what you will, in ANY field, be it hardware, engineering, etc. the idea is to REDUCE variability, NOT increase it. Performing benchmarks on two systems would increase variablilty due to the differences in systems. So the key component is producing a stable, repeatable environment in which to run a variety of tests within.

Additonaly as was mentioned, you don't just "throw the card in another system" and out pops your reliable data lickety-split. You would effectively be doubling the time required for a review. This is a case of diminishing returns. The likelihood of your system affecting the performance of an add-in card is extremely slim (else ATI/Nvidia et.al. would be seriously limiting the available market for their card). Couple this with the doubling of time required for the review and the fact, as you pointed out yourself, that OTHER hardware sites will be using different systems for their own bench-marking that you can look at if you so chose, really closes the book on the subject.

On a side note, if your willing to pay for Bartons and Granite Bays, I'll gladly provide you whatever benchmarks you want

 

[paralazarguer]

The setup that anand used in his review was the fastest available. The fast processor is the 3.06 P4 and PC1066 is the fastest ram with the i850 being the best platform going price aside. My guess would be that he was trying to use all the fastest components to remove the system bottleneck as much as possible. He COULD have used a nforce 2 board with an athlon XP3000+ but since it's slower, that wouldn't make a lot of sense.

 

[KGB1]

Originally posted by: PrinceXizor
And obviously, you know nothing about benchmarking or systematically examining a single component. When you are comparing features, performance, benchmarking, call it what you will, in ANY field, be it hardware, engineering, etc. the idea is to REDUCE variability, NOT increase it. Performing benchmarks on two systems would increase variablilty due to the differences in systems. So the key component is producing a stable, repeatable environment in which to run a variety of tests within.

Additonaly as was mentioned, you don't just "throw the card in another system" and out pops your reliable data lickety-split. You would effectively be doubling the time required for a review. This is a case of diminishing returns. The likelihood of your system affecting the performance of an add-in card is extremely slim (else ATI/Nvidia et.al. would be seriously limiting the available market for their card). Couple this with the doubling of time required for the review and the fact, as you pointed out yourself, that OTHER hardware sites will be using different systems for their own bench-marking that you can look at if you so chose, really closes the book on the subject.

On a side note, if your willing to pay for Bartons and Granite Bays, I'll gladly provide you whatever benchmarks you want

Umm yeah... that's what I want to happen How does a R350 work on a P4 and a Athlon XP? Is the way it SHOULD be done. I'm not saying test it on Cyrix 900Mhz.. pc.. or a dual CPU system.

How would it not be stable? I didn't even mention overclocking, even AT overclocked the R350.

 

[chizow]

Actually, I think the 3000+ was right there with the P4 3.06 in gaming applications.

Chiz

 

[NFS4]

Originally posted by: KGB

Originally posted by: PrinceXizor
And obviously, you know nothing about benchmarking or systematically examining a single component. When you are comparing features, performance, benchmarking, call it what you will, in ANY field, be it hardware, engineering, etc. the idea is to REDUCE variability, NOT increase it. Performing benchmarks on two systems would increase variablilty due to the differences in systems. So the key component is producing a stable, repeatable environment in which to run a variety of tests within.

Additonaly as was mentioned, you don't just "throw the card in another system" and out pops your reliable data lickety-split. You would effectively be doubling the time required for a review. This is a case of diminishing returns. The likelihood of your system affecting the performance of an add-in card is extremely slim (else ATI/Nvidia et.al. would be seriously limiting the available market for their card). Couple this with the doubling of time required for the review and the fact, as you pointed out yourself, that OTHER hardware sites will be using different systems for their own bench-marking that you can look at if you so chose, really closes the book on the subject.

On a side note, if your willing to pay for Bartons and Granite Bays, I'll gladly provide you whatever benchmarks you want

Umm yeah... that's what I want to happen How does a R350 work on a P4 and a Athlon XP? Is the way it SHOULD be done. I'm not saying test it on Cyrix 900Mhz.. pc.. or a dual CPU system.

How would it not be stable? I didn't even mention overclocking, even AT overclocked the R350.

Why bother??? The difference between a top of the line Barton and a 3.06 P4 wouldn't warrant the need for two seperate systems.

What you're asking Anand to do is basically DOUBLE his workload for nothing. Instead of standard and overclocked benches on one system, you'd need two systems. Plus all the non AA/AF settings vs the AA/AF settings.

Do you know how many HOURS it took for Anand for just the Radeon 9800 Pro tests on ONE system??? Multiply that by two to satisfy your wishes!

The man is a working machine, but even he has limits. I saw him Tuesday and he was battling sickness, hadn't slept much in days, and he STILL drug himself to class where he snuck in some time to work on the write-up for a hard drive review (yes, in class). This man is THAT dedicated to this site.

CUT THE MAN SOME SLACK!!!

 

[PrinceXizor]

Originally posted by: KGB

Originally posted by: PrinceXizor
And obviously, you know nothing about benchmarking or systematically examining a single component. When you are comparing features, performance, benchmarking, call it what you will, in ANY field, be it hardware, engineering, etc. the idea is to REDUCE variability, NOT increase it. Performing benchmarks on two systems would increase variablilty due to the differences in systems. So the key component is producing a stable, repeatable environment in which to run a variety of tests within.

Additonaly as was mentioned, you don't just "throw the card in another system" and out pops your reliable data lickety-split. You would effectively be doubling the time required for a review. This is a case of diminishing returns. The likelihood of your system affecting the performance of an add-in card is extremely slim (else ATI/Nvidia et.al. would be seriously limiting the available market for their card). Couple this with the doubling of time required for the review and the fact, as you pointed out yourself, that OTHER hardware sites will be using different systems for their own bench-marking that you can look at if you so chose, really closes the book on the subject.

On a side note, if your willing to pay for Bartons and Granite Bays, I'll gladly provide you whatever benchmarks you want

Umm yeah... that's what I want to happen How does a R350 work on a P4 and a Athlon XP? Is the way it SHOULD be done. I'm not saying test it on Cyrix 900Mhz.. pc.. or a dual CPU system.

How would it not be stable? I didn't even mention overclocking, even AT overclocked the R350.

Stable = no variation b/n testing of other components i.e. Radeon 9700 Pro, GeForceFX Ultra. It has nothing to do with overclocking at all. If you test environment is different for each product you are testing, your benchmarks don't mean a thing. Also, from a computer architecture standpoint, I'm trying to figure out why it would matter. The point of a benchmark is to compare one video card to another. Whatever affects a system might have on a card i.e. how that system would affect the performance of the video card would be applied in equal measure to any OTHER card put in that exact same system. So it doesn't matter. What we are interested in is the relative performance of the card vs. other video cards, not its specifice performance on any one particular system....

 

[WyteWatt]

NFS4 yep and he even did some minimum fp benchmarks just like everyone wanted!

 

[KGB1]

Originally posted by: NFS4

Why bother??? The difference between a top of the line Barton and a 3.06 P4 wouldn't warrant the need for two seperate systems.

What you're asking Anand to do is basically DOUBLE his workload for nothing. Instead of standard and overclocked benches on one system, you'd need two systems. Plus all the non AA/AF settings vs the AA/AF settings.

Do you know how many HOURS it took for Anand for just the Radeon 9800 Pro tests on ONE system??? Multiply that by two to satisfy your wishes!

The man is a working machine, but even he has limits. I saw him Tuesday and he was battling sickness, hadn't slept much in days, and he STILL drug himself to class where he snuck in some time to work on the write-up for a hard drive review (yes, in class). This man is THAT dedicated to this site.

CUT THE MAN SOME SLACK!!!

Hey dude.. take a chill pill or something. I'm not bashing the guy on anything, I'm not going to get all fussy and think up an email to send to him. I appreciate the work he does, but all I want is dual platfrom so I can see how it performs. Is that so much to ask, and by your answers I see that I am asking too much from him, and I shouldn't keep pushing his buttons. I don't want to scouver around many sites to see the performance of 1 products on several platforms. If a review was done with the platforms and hardware accessible to them, why not put it up?

 

[PrinceXizor]

Because as I've been trying to tell you, its pointless and has no bearing. For example,

Your Athlon XP 3000+ affects your card becaus of "Insert reason here"

That "anonymous reason" will affect you whether you put a Radeon 9800 Pro in there, a 9700 Pro, a GeForceFX 5800 Ultra or whatever. This "anonymous reason" will affect each and every card that you put in that system. So who cares.

The point of a video card review is to demonstrate how that video card compares to all other video cards. Its performance on any one particular system is meaningless because everyone doesn't have the exact same system as everyone else. And any advantage/disadvantage of any particular system will be applied in exactly the same fashion no matter what card you put in there, so it doesn't matter.

 

[Spicedaddy]

I don't want to scouver around many sites to see the performance of 1 products on several platforms. If a review was done with the platforms and hardware accessible to them, why not put it up?

Because like it was said earlier, it effectively doubles the time spent on a review for something very few people care about. (because they understand the point of a video card review is to use a fast system so you don't have bottlenecks anywhere else, be it AMD or Intel, it doesn't matter)

If you're too lazy to read 2 different reviews to get the info you want, well, what can I say...

 

[NFS4]

Originally posted by: KGB

Originally posted by: NFS4

Why bother??? The difference between a top of the line Barton and a 3.06 P4 wouldn't warrant the need for two seperate systems.

What you're asking Anand to do is basically DOUBLE his workload for nothing. Instead of standard and overclocked benches on one system, you'd need two systems. Plus all the non AA/AF settings vs the AA/AF settings.

Do you know how many HOURS it took for Anand for just the Radeon 9800 Pro tests on ONE system??? Multiply that by two to satisfy your wishes!

The man is a working machine, but even he has limits. I saw him Tuesday and he was battling sickness, hadn't slept much in days, and he STILL drug himself to class where he snuck in some time to work on the write-up for a hard drive review (yes, in class). This man is THAT dedicated to this site.

CUT THE MAN SOME SLACK!!!

Hey dude.. take a chill pill or something. I'm not bashing the guy on anything, I'm not going to get all fussy and think up an email to send to him. I appreciate the work he does, but all I want is dual platfrom so I can see how it performs. Is that so much to ask, and by your answers I see that I am asking too much from him, and I shouldn't keep pushing his buttons. I don't want to scouver around many sites to see the performance of 1 products on several platforms. If a review was done with the platforms and hardware accessible to them, why not put it up?

How often do you see a site do a review of a video card on two different systems?? It just doesn't happen very often b/c of the workload involved. I find it best not to rely on just one site anyway. I take a look at reviews from all the major sites when evaluating the performance of a product. In fact, I can't think of one of the sites that posted 9800 Previews that tested on two systems.

No reason to single out Tom and Anand for it. It's just A LOT of work with diminishing returns.

 

[KGB1]

EXACTLY SSE2, Hyper-Threading, 3D-Now does not matter when you compare 3D cards. They're all the same. I think I'm going to test it on my Pentium 200

 

[bigshooter]

The benchmark were also there to show how it compared to other cards, namely the 9700 pro and the geforce fx. It wasn't to show how the card did on various systems, but how it compared to the two other high end cards. You can take the benchmarks for the geforcefx or radeon 9700 pro on an AMD system, i'm sure anand or someone else included them a long time ago, and then figure how much faster the 9800 would be on that system, by looking at how much faster it is on the p4 system. (hope nobody said this, i didnt read all the threads =P)


I hope the hard drive writeup he has is a final (or really close to final) version of the WD raptor. I'm really interested in that drive.

 

[PrinceXizor]

Originally posted by: KGB
EXACTLY SSE2, Hyper-Threading, 3D-Now does not matter when you compare 3D cards. They're all the same. I think I'm going to test it on my Pentium 200

I'm not saying they don't affect the performance of a video card, what I'm saying is whatever positive affect they have is applied equally to a Radeon 9700 or a Radeon 9800 or a GeForceFX or whatever card you put in there. It affects all the cards the same so its irrelevant to determining which card performs better than another card. I can illustrate it mathematically too.

The first letter indicates the baseline performance of the video card. The second letter indicates the increase due to whatever you want it to be SSE2, etc...

9800Pro = AX
GEForceFX = BX

Ratio of Perfomance = (AX/BX) = A/B

Ratio of Performance on another system without said improvement

9800Pro = A
GeForceFX = B

Ratio of Performance = A/B its EXACTLY THE SAME

 

[Spicedaddy]

Originally posted by: KGB
EXACTLY SSE2, Hyper-Threading, 3D-Now does not matter when you compare 3D cards. They're all the same. I think I'm going to test it on my Pentium 200

this thread is now officially certified ***retarded***

 

[paralazarguer]

The thread...and the original poster.