R8500 20% Faster Than GF4 Ti4600?

[Joe DeFuria]

Actually, Mouse could be telling the truth, you know.

Not the way he presented it. It could be true that ATI has some "experimental drivers" that have advanced HSR (like 3dfx's drivers at one time).

However, Mouse is also saying that some web reviewers are using these, and getting "artificially" high benchmark scores, and that "newer" (non-experimental) drivers drop performance significantly.

I haven't seen any reviewer have any drivers that significantly outpace any of the currently available ones. (Ones that don't have visual anomalies.) So what is Mouse's argument based on?

 

[Ichneumon]

A quick google search turns up a handful of Radeon 8500 driver comparisons:

http://www.subzerotech.com/articles/drivers/ati8500comparison/

http://www.epigamer.com/article.php?a=28

There's just a couple... As the driver versions get newer, some things get faster, some stay about the same, some loose a few fps, but the general trend is up.

Dunno where you get this sillyness about a s00p3r l33t 4x performance gain driver that is ONLY used by reviewers, but it is quite the amusing bag of horse-pucky.

I've run just about every driver that ATI has put out (official, beta, developer or otherwise) for the radeon series cards since the original Radeon was released. The general performance and quality trend has always been to go up, with some drivers loosing a bit here or there, but gaining it back and more in later drivers.

Stop being silly. There is no such thing as a l33t benchmark driver given to a select few sites that gives 4x performance games or some BS by cutting out 20% of the scene info.

8500 performance really equals that of an MX440... And some of you guys took this guy even slightly seriously?

 

[LittlePenny]

Actually, Mouse could be telling the truth, you know.

I mean, what if I were to say the Voodoo4 and Voodoo5 had/has HW T&L but it was broken so badly that 3dfx had to disable it entirely but later on decided to give it a try with the help of some drivers...??

But, you still have a reputation that we can trust. Not only does reputation matter but the style at which you present it. Nappe1 has some crazy rumor posts, but I still think of him as an integral member of the Beyond3d forums because his posts are fun to read and presented in a mature format.

<edit>spelling</edit>

 

[OpenGL guy]

Re: Too funny

But no one has these "special drivers that increase performance by 4x"

HAIL THE NEW GOD OF THIS WORLD WHO KNOW'S EVERYTHING GOING ON IN THIS WORLD!!!

I searched the whole codebase... not one occurance of #define SPECIAL_DRIVER_THAT_INCREASES_PERFORMANCE_BY_4X

I assume you are a senior level code developer then? Because you must realise that these drivers were developed quite a while ago (based on the current drivers at the time) and are seperate from the real drivers. They aren't updated (except the version info is updated some times so reviewers can claim a new version). They are controlled quite tightly which isn't too hard because not many people need to have access to them. (They did not bother to optimise these special drivers so the quality loss was minimal because it was not necessary). I won't even bother to comment on "SPECIAL_DRIVER_THAT_INCREASES_PERFORMANCE_BY_4X" since it proves you don't want to talk seriously if you make such a ridicolous comment.

I thought you were the one being ridiculous, so I replied sarcastically. BTW, I still think you are being ridiculous. The truth is that ATI has access to alien technology and we have incorporated this into our hardware and software design. nvidia is totally against using AT (alien tech.) in their products and that's why they will ultimately lose.

When did I ever claim to know everything in the world? I will claim that I know more about ATI than you do, however.

Also, do you really think ATI would spend the resources to maintain two separate drivers? I.e. the "special" driver and the normal one? If so, you need to take your medication.

Secondly, if you are referring to the "Quake/Quack" issue, that problem has been resolved

This was one of their first attempts at driver cheating and I think the success gave them a lot of confidence to continue.

Cheat or optimization? If you can make an application run faster without affecting visual quality, why wouldn't you do it? The problem with the early drivers was that there was a bug that did affect quality. That bug is fixed. End of story.

P.S. This is my last post to this thread.
P.P.S. coz = 'cause = because != course, so you don't say "Of coz" coz "Of because" makes no sense.

 

[Randell]

8500 performance really equals that of an MX440... And some of you guys took this guy even slightly seriously?

Nope he was asked for proof, I dont think anyone beleived in these special drivers. Of course the Savage2000 TnL debacle proved chips can ship with design flaws, so if he had proof, well then show it.

 

[Sharkfood]

I will desist since it's obvious no one here is going to believe even my most basic hypothesis (Ti200 is a better choice then the Rad LE8500) despite the undisproven facts (that the Rad 8500 is performing poorly at the moment).

That is the fundamental argument. Your spurious theory is *not* confirmed in the slightest and even your assertions are not supported- either by user reproduced testing, or by website reviews abroad.

You keep labeling things as "undisproven facts" yet provide zero of such deserving merit. Just baseless statements with the only support being 'people in the know should know!' drivel.

I can swap out a GF3 (not Ti200) and an 8500 and test both under a series of benchmarks and closely inspect IQ (yes, even scaled/zoomed 900% in a PS) and not see any indication of what you speak on current drivers for both. If anything, I can clearly see indications of IQ variance between the two, but not all in favor of ATI or NVIDIA- they both have their strongpoints and weaknesses and there is absolutely NO indication that "ATI cheats by removing 20% of the scene" nonsense that you are trying to uphold.

It should also be clearly noted that website review results are easily user reproducible and *very* conservative. If there were these spurious 4x drivers available, don't you think people that have purchased the 8500 would be screaming? This isn't the case as most users with properly configured systems are able to achieve the delivered performance *as advertised* and benchmarked at review sites- to meet OR exceed. And side-by-side IQ isn't this massive derrogatory suck-fest comparison like you are trying to assert. If anything, I'd simply argue the reverse is true with the Geforce cards at "default" settings (mipmapping, TC, LOD bias, etc.etc.) but "tuned out" by any end users that purchase the hardware through settings.

If any attention to matching IQ is utilized, the performance gaps in the various products is narrowed, and all that is left is some driver issues that are almost totally mutually exclusive between ATI or NVIDIA. This is what makes the choice of cards not a hard-set rule and users interested in videocard choice don't have some clear cut 'buy the Ti200 coz it's bestest for all!' types of doctrines your alleged "anti-NVIDIA" crusade suggests.

Just my $0.02,
-Shark

 

[Nil Einne]

Re: Too funny

When did I ever claim to know everything in the world? I will claim that I know more about ATI than you do, however.

I think his/her point was that you claimed that you knew no one had these drivers which obviously isn't possible since.

Also, do you really think ATI would spend the resources to maintain two separate drivers? I.e. the "special" driver and the normal one?

Er, but he/she never said that they were maintaining these drivers. It sounded like he/she was claiming they were developed very quickly and after that left alone which wouldn't consume much resources and would be worth it if it would kill them which it would

Cheat or optimization? If you can make an application run faster without affecting visual quality, why wouldn't you do it? The problem with the early drivers was that there was a bug that did affect quality. That bug is fixed.

I would agree, IF they informed you what it was doing. In any case, according to this dummy, they did affect quality altho only minorly (supposedly the new drivers don't affect quality at all).

 

[Nil Einne]

However, Mouse is also saying that some web reviewers are using these, and getting "artificially" high benchmark scores, and that "newer" (non-experimental) drivers drop performance significantly.

I haven't seen any reviewer have any drivers that significantly outpace any of the currently available ones. (Ones that don't have visual anomalies.) So what is Mouse's argument based on?

Well not entirely true since in this post the reviewer did have siginificantly better performance. But this was traced to a setting issue. And it is supposedly true that some new drivers drop performance but this is because of a bug.

 

[Nil Einne]

A quick google search turns up a handful of Radeon 8500 driver comparisons:

http://www.subzerotech.com/articles/drivers/ati8500comparison/

http://www.epigamer.com/article.php?a=28

There's just a couple... As the driver versions get newer, some things get faster, some stay about the same, some loose a few fps, but the general trend is up.

Dunno where you get this sillyness about a s00p3r l33t 4x performance gain driver that is ONLY used by reviewers, but it is quite the amusing bag of horse-pucky.

I've run just about every driver that ATI has put out (official, beta, developer or otherwise) for the radeon series cards since the original Radeon was released. The general performance and quality trend has always been to go up, with some drivers loosing a bit here or there, but gaining it back and more in later drivers.

Obviously you and the reviewers above don't have access to these "special" drivers then

 

[Nil Einne]

It should also be clearly noted that website review results are easily user reproducible and *very* conservative. If there were these spurious 4x drivers available, don't you think people that have purchased the 8500 would be screaming? This isn't the case as most users with properly configured systems are able to achieve the delivered performance *as advertised* and benchmarked at review sites- to meet OR exceed.

Well some of those ATIphilic sites seem to have oddly fast benchmarks (and they are the ones which are supposedly getting these 'special' drivers) but usually it's related to purposeful settings manipulation and are reproducible if you know what the dummy did. So I suppose it's partially plausible (in these terms, obviously it's illogical on the whole). In any case, no one reads these ATIphilic sites anyway (I think) so it isn't too important. And he/she did have a point that one new (now old?) driver version did reduce performance in some drivers due to a bug.

 

[Doomtrooper]

It should also be clearly noted that website review results are easily user reproducible and *very* conservative. If there were these spurious 4x drivers available, don't you think people that have purchased the 8500 would be screaming? This isn't the case as most users with properly configured systems are able to achieve the delivered performance *as advertised* and benchmarked at review sites- to meet OR exceed.

Well some of those ATIphilic sites seem to have oddly fast benchmarks (and they are the ones which are supposedly getting these 'special' drivers) but usually it's related to purposeful settings manipulation and are reproducible if you know what the dummy did. So I suppose it's partially plausible (in these terms, obviously it's illogical on the whole). In any case, no one reads these ATIphilic sites anyway (I think) so it isn't too important. And he/she did have a point that one new (now old?) driver version did reduce performance in some drivers due to a bug.

I have never seen ONE review that showed the 8500 with inflated benchmarks but I've seen reviews with much lower benchmarks then what I've experienced. Now I'm usually running leaked Dev drivers too, and alot of the sites still use the original CD drivers
All the reviews clearly state the driver used during the review and the idea of special drivers for certain reviewers is a JOKE.


Anandtech truck Demo..compare mine to his

Comanche 4- Detail settings maxxed using FRAPS 57 fps

Unreal Tournament OGL