R8500 20% Faster Than GF4 Ti4600?

[Matt]

I saw this review over at OCWorkbench and this result they've posted up for Serious Sam SE:

Wow, 20% faster than the GF4 Ti4600. In a CPU limited situation, the GF4 wouldn't drop that much at just 1280x1024, and the guy is on a P4 1.5Ghz. Am I the only one who thinks something is weird with that result?

Guess I should throw away my GF4 and get a R8500 instead, because according to GameSpy and now OCWorkbench, the R8500 is faster than my GF4 or "just a little bit faster" than my GF4.

Anyways, am I the only one who think it's a good thing to add anisotropic filtering and FSAA results in benchmarks now? Considering the amount of power and the featuresets of both the R8500 and GF4, I think it should become the standard rather than the exception. This review didn't even include them, and after reading the review, I can't see how the gamer who are interested in either of the two cards will get any real information from their review. Heck, if you're not interested in FSAA or anisotropic filtering, get a GF2 GTS/Radeon DDR/Kyro2. No point investing in much else past that.

Now I really understand the frustration of 3dfx people when some sites would just benchmark Quake 3 and 3DMark 2000 and that's it.

 

[ben6]

It's even worse than that. He used the default settings at the resolutions specified. With those conditions the Serious Sam benchmark is unreliable at best. You should have seen what happened the first time I ran any of the SS:TSE demos with default settings .Consider these benchmarks from my nForce review:

http://www.pcrave.com/reviews/257.htm

The difference in performance between a 230/190 DDR Radeon 8500DV and a AIW Radeon at 166/166 DDR in the same default benchmark settings/resolution is less than 20% at 1280x1024 32bit settings? Less than 15% at 1024x768? Ugh....

 

[ben6]

Whelp , forgot I ran the first encounter not the second encounter in that review . But there is a reason why the Reverend made up his own benchmark script for Serious Sam..

 

[LittlePenny]

Anyways, am I the only one who think it's a good thing to add anisotropic filtering and FSAA results in benchmarks now?

It certainly would be nice if Aniso implementations were the same so comparisons could be made. Another problem with FSAA is that image quality is subjective, but FPS numbers are not. So, while 3dfx's FSAA might have been superior, I doubt we will see that style used until they figure out a way to minimize the hit. I think the real reason we haven't seen better FSAA yet is because of this view we have about FPS.

 

[Dave Baumann]

But there is a reason why the Reverend made up his own benchmark script for Serious Sam..

If you alter the setting in the advanced setting that effectly overrides the 'defaults' that are loaded.

 

[ben6]

Thanks Dave , I found that out after running the default benchmarks on a : Geforce2 GTS, 3, Ti200, Ti500, Ti4600, Radeon 32mb, AIW Radeon, Radeon 7500, 8500, 8500DV, Kyro I and II heh (borrowed some of them for the nForce review).I ran into all kinds of weirdness, including a Ti200 beating a Ti500 at one setting and the 7500 beating my 8500DV with default settings at one setting.

 

[Matt]

LittlePenny:

I agree that comparing anisotropic filtering isn't fair, as it's trilinear vs. bilinear, and FSAA with supersampling vs. multisampling, but it does give the readers an indication of what you do get with the card when they're enabled. That's one of the reasons why I don't do comparisons in my reviews, because a review should be a review of the card, not a comparison against the competition. Comparisons should be just that... comparisons.

 

[MouseAnony]

I reckon that person was probably using the special review drivers, reserved for ATI favoring reviewers. They not only have the original 25% info removal 'bug' (refer my other post on the Rad 8500 if you're not sure what I'm talking about), but it's an enhanced 'bug' removing between 35-40% of the scene info thereby giving a massive performance boost.

 

[Dave Baumann]

Please start posting substatiated information, rather than fabrication.

 

[Sabastian]

I reckon that person was probably using the special review drivers, reserved for ATI favoring reviewers. They not only have the original 25% info removal 'bug' (refer my other post on the Rad 8500 if you're not sure what I'm talking about), but it's an enhanced 'bug' removing between 35-40% of the scene info thereby giving a massive performance boost.

Hmm that would be quite a large number of special reviewers. You would think that one of them would leak it out somewhere. I have NEVER seen so many positive glowing reviews of ATi graphics products. It seems there is an extrodinary number of extremely positive reviews poping up everywhere. ATi must have gotten their driver issues ironed out entirerly. Here is a relatively small list I have found recently.

http://www.hardocp.com/reviews/vidcards/ati/8500le/index.html

"As for the card built by ATI, it's rock solid. Other then the odd visual glitch problem, I had this card overclocked and performing extremely well. If you want to mod a card out, this is a good card to do that with. Arguing best price versus performance, I have to say the 8500LE GPU is definitely on the top of the chart. Coming in under $200 and having the potential to overclock higher than a retail 8500, puts this card in a class of its own! I can easily recommend this chip to anyone who wants to save a little cash and still get top of the line performance."

http://216.123.84.163/custom/layout/3b/radeon_8500.wmv

ATI Radeon 8500 64MB Video Card in Video Review #135:
"This Video Card has crisp 2D and dual display support, excellent 3D gaming performance and it's affordable. Bang for the buck, this card is hard to beat! The Radeon 8500 64MB Video Card definitely falls in the KICKASS category" ~3dGameMan.com

http://www.bench-house.com/video/ati/radeon_8500/radeon_8500_001.html

"Because all of this ATI Radeon 8500 receives Editor's Choice award as the best product in its class."

http://www.bytesector.com/data/bs-article.asp?faq=4&fldAuto=60

ATi has done a marvelous job with its new Radeon 8500. The fast clock speeds crush its opponents into dust. In addition, the 8500 is one solid piece of hardware with great integration, and latest technologies incorporated. And while ATi released the Radeon 8500 for the purpose of keeping up with nVidia’s Geforce 3 series; their competitor is on their toes fearing enormous success to follow their opponent, the maker of the Radeon 8500! Overall, the ATi Radeon 8500 is a great contender in the graphics board industry, and it impacts us to see how technology is progressing ever-so-fast, and that’s why we give the Radeon 8500 our first 100%! Congratulations go to ATi and all of its fans

http://www.designtechnica.com/reviews.php?op=showcontent&id=19&page=1

http://www.tweakers.com.au/articles/graphics/auriga_r8500/page1.asp

http://www.3dchipset.com/articles/ati.aiwr8500dv/

http://www.hardocp.com/reviews/vidcards/ati/8500le/index.html

http://www.escortcosworth.com/petrucciweb/rubriques/hardware/geforce3vsradeon/gf3vsradeon.htm

http://www.tweakersasylum.com/Reviews/FIC/AT008/00000001.htm

http://www.3dxtreme.org/8500p1.shtml

http://www.h2h.ca/8500review.shtml

You would think that with all these "special review drivers" floating around someone would be able to snap one up somewhere.

Sabastian

 

[Sabastian]

I reckon that person was probably using the special review drivers, reserved for ATI favoring reviewers. They not only have the original 25% info removal 'bug' (refer my other post on the Rad 8500 if you're not sure what I'm talking about), but it's an enhanced 'bug' removing between 35-40% of the scene info thereby giving a massive performance boost.

Your first post was sufficient to make your point.

Edited: John Reynolds

 

[OpenGL guy]

LOL

I reckon that person was probably using the special review drivers, reserved for ATI favoring reviewers. They not only have the original 25% info removal 'bug' (refer my other post on the Rad 8500 if you're not sure what I'm talking about), but it's an enhanced 'bug' removing between 35-40% of the scene info thereby giving a massive performance boost.

As an empolyee of ATI, I guess I'll have to pull some strings and get ahold of these "extra special drivers".

Or, I guess I could just look at the source code and search for #ifdef ATI_CH43T0RZ_DR1V3R



P.S. Sorry for the sarcasm... I couldn't help myself.

 

[Sabastian]

Re: LOL

I reckon that person was probably using the special review drivers, reserved for ATI favoring reviewers. They not only have the original 25% info removal 'bug' (refer my other post on the Rad 8500 if you're not sure what I'm talking about), but it's an enhanced 'bug' removing between 35-40% of the scene info thereby giving a massive performance boost.

As an empolyee of ATI, I guess I'll have to pull some strings and get ahold of these "extra special drivers".

Or, I guess I could just look at the source code and search for #ifdef ATI_CH43T0RZ_DR1V3R



P.S. Sorry for the sarcasm... I couldn't help myself.

Yeah if I worked at ATi I think I would be pretty upset at all the FUD on the internet with regards to my work. Anyhow I don't think the appology was nessesary. I thought that it was quite creative.

Sabastian

 

[OpenGL guy]

Re: LOL

[quote="Sabastian]Yeah if I worked at ATi I think I would be pretty upset at all the FUD on the internet with regards to my work. Anyhow I don't think the appology was nessesary. I thought that it was quite creative.
[/quote]

None of my work has seen the light of day, so why should I be upset? (I just started here last month.)

 

[Doomtrooper]

Re: LOL

[quote="Sabastian]Yeah if I worked at ATi I think I would be pretty upset at all the FUD on the internet with regards to my work. Anyhow I don't think the appology was nessesary. I thought that it was quite creative.

None of my work has seen the light of day, so why should I be upset? (I just started here last month.)[/quote]

Huh. so what are you working on

Secrets can be told here..really

 

[Sabastian]

Re: LOL

[quote="Sabastian]Yeah if I worked at ATi I think I would be pretty upset at all the FUD on the internet with regards to my work. Anyhow I don't think the appology was nessesary. I thought that it was quite creative.

None of my work has seen the light of day, so why should I be upset? (I just started here last month.)

Huh. so what are you working on

Secrets can be told here..really [/quote]

Hehe, yeah at worst it would be some sort of harmless rumor of many on the net. Besides I love a good rumor, come on out with it, just a modest crumb on the R300. I won't tell anyone, honest.

Sabastian

 

[Matt]

Hmm, I wasn't thinking of cheating drivers, I was thinking more along the line of making a mistake and letting it go in favor of getting the card for free. I've made mistakes in reviews, although I do correct them right away.

I agree, the Radeon 8500 been getting a lot of glowing positive reviews, and deservedly so, considering how much it has matured and the steal it now pose at ~$150. I just don't think it should be faster than a Ti4600, considering the specs, bandwidth, and generation it's in (disregarding CPU limited situations).

 

[Randell]

agreed

 

[MouseAnony]

These are all ancient reviews therefore using the originally 'buggy' drivers which naturally cut off 25% of the less important info (read my first post). The special drivers only go to a select few. If you see the Rad 8500 beating a Ti4600 in a recent review it's probably the special drivers in play.

The special drivers cut of about 35% thereby increasing performance by about 2x again. So a total of 4x performance increase over the current drivers.

You should look at recent reviews if you don't believe. Anandtech etc. You will see performance is now terrible. Show me two recents review by a reliable site (Anandtech etc) and two recent reviews from less reliable sites (but not crapola like Ace's hardware, www.iloveati.org, Tom's Hardware).

Why would you actually want these drivers? True they increase performance by about 4x but the quality is so crap it isn't worth it. If you don't believe me try the old 'buggy' drivers and compare this with the new 'fixed' drivers.

 

[Doomtrooper]

These are all ancient reviews therefore using the originally 'buggy' drivers which naturally cut off 25% of the less important info (read my first post). The special drivers only go to a select few. If you see the Rad 8500 beating a Ti4600 in a recent review it's probably the special drivers in play.

The special drivers cut of about 35% thereby increasing performance by about 2x again. So a total of 4x performance increase over the current drivers.

You should look at recent reviews if you don't believe. Anandtech etc. You will see performance is now terrible. Show me one recent review by a reliable site (Anandtech etc not crapola like Ace's hardware, www.iloveati.org).

Why would you actually want these drivers? True they increase performance by about 4x but the quality is so crap it isn't worth it. If you don't believe me try the old 'buggy' drivers and compare this with the new 'fixed' drivers.

OMG it is Nam_ng