[H]ard|OCP Radeon 8500 Review

[GTaudiophile]

Linky

 

[MoobyTheGoldenCalf]

We all know that Quake3 is one of the most utilized 3D benchmarks in the world, there is simply no arguing this. We came into possession of a program that you will see more of here soon. It is a .exe file that when placed in the Quake3 directory renames every single instance of the word "quake" to "quack" as well as builds a new quake.exe file name quack.exe. When you run quack.exe, you will find that Quake3 runs exactly as it should.

Here is the interesting part. The top three scores are "Quack" scores, and the bottom three are "Quake" scores taken both with the Radeon 8500 using the drivers on their website. It is my guess that the Radeon 8500 drivers have Quake specific instructions embedded in them to give better Quake3 scores. I did specifically ask an ATi engineer about this and he explained that ATi would not produce game specific drivers but they would surely optimize for game engines.

It seems to me that if this particular set of Radeon 8500 drivers was optimized for the Q3 game engine that we would see identical sets of data above. Now whether or not this is seen as good or bad, you still get the elevated performance when running Quake 3. I am not going to push this issue any further but to simply ask for logical explanations as to why this would occur.


WOW! Sounds like ATI's got some explaining to do.

 

[MoobyTheGoldenCalf]

Ooooooh, more good stuff:

Also, ATi is famous for releasing VidCards of a particular type with different clock speeds as we pointed to in our opening paragraph. Our card was clocked at 275/275 keeping our numbers rounded off. We have already had mail from one reader explaining that he purchased an OEM 8500 that was clocked at 250/250. Now at this time we have no proof of this and all other cards that we have checked and have had reported on were at the expected 275/275. This is an excerpt from an ATi mail that addressed the point of the reader we mentioned above.

Dear Asfand,

Thank you for contacting ATI Technical Support Canada.

----------------------------------
Snipped ticket number etc.
---------------------------------
Regarding clock rates on ATI products:

Clock rates on our products are never specified. We can and do change them without notice. Please direct your non-technical issues to our sales team.

Regards,

Doug McDonald
Technical Support Canada
ATI Technologies, Inc.

I have a problem with ATi's policy on this clocking issue and would like to see them hold themselves to producing OEM and retail cards of one specification and I am sure many of their potential customers would too. Please let us know if you run into this issue with an 8500 card.


Clock rates never specified huh? Sounds like everyone's not on the same page over at ATI.

Well, I for one am very happy to see a reviewer have the balls to address these issues. Thank You Kyle!!!!

 

[MoobyTheGoldenCalf]

Ok, last excerpt (I swear)

We are surprised that ATi launched this product at this time. It is obvious that all the features are not operating properly and testing was lackluster at best. We found compatibility problems with the card after only spending ten minutes with it after getting the drivers installed. We would think that if we were putting out a VidCard that we would be embarrassed by such events. You might have seen this card referred to by other hardware sites as having potential. We know that when we buy a piece of hardware, potential is absolutely not what we are looking for. We want to open the box and get our money's worth and ATi has not done that with the Radeon 8500 in its current state.

My thoughts exactly!

 

[GTaudiophile]

Dude, nVidia does the same thing to their drivers to improve Q3 performance!

 

[MoobyTheGoldenCalf]

<< Dude, nVidia does the same thing to their drivers to improve Q3 performance! >>



I think you missed the point. I'm sure nVidia does optimize for Q3, but that is not what ATI did. It seems that ATI has optimized their card to give a better benchmark in Quake 3, not optimize the card for the Quake 3 engine. That is wrong, dude. I don't care how you look at it.

 

[MrGrim]

Yep, if it was optimized for the game engine, giving the files a different name shouldn't make a difference! Somebody should do the same thing with the GF3 though.

 

[Tominator]

OUCH! ATI takes it on the chin....

 

[MoobyTheGoldenCalf]

One more thing that has really got me ticked off is the RAMDAC issue. From ATI's website:

Q6: What is the DAC speed?

A6:
The DAC frequency is 400MHz.


From this thread over at Rage3D, it looks like ALL boards, Retail and OEM are 350mhz, not 400mhz.

http://www.rage3d.com/board/showthread.php?s=388340b62c66e3269a2f06fdca56ac26&threadid=33585863

ATI's business practices are crap.

 

[MrGrim]

Oh dear ...

 

[Ryan]

Damn - ATI got slammed. Hmm, the Ti200 is looking REALLY sweet right now

 

[Czar]

Like before ATI makes great hardware but horrible horrible drivers

 

[MrGrim]

<< Like before ATI makes great hardware but horrible horrible drivers >>



Actually, it's not just bad drivers now, it's bad cheating drivers by the looks of it. :|

 

[MoobyTheGoldenCalf]

<< Like before ATI makes great hardware but horrible horrible drivers >>



Well, the hardware looks great on paper and on the ATI website, but as ATI sez, "Clock rates on our products are never specified. We can and do change them without notice. "

So what you get in your box may be a lil' different from what they claim

 

[Pariah]

"It seems that ATI has optimized their card to give a better benchmark in Quake 3, not optimize the card for the Quake 3 engine."

H]ard|OCP made no such accusation, those are your words. Here is what they said:

"Now whether or not this is seen as good or bad, you still get the elevated performance when running Quake 3."

Note, it says better performance, not better benchmarks. Creating optimized builds of games is certainly not rare. Matrox released Quake specific drivers a while back. 3dfx always had quake specific drivers as well. Also, why has no one else seen or heard of this file? I'm sure it exists, but a user has to knowingly get the file first (anyone else seen this? Anand, Tom?) and put it in the Quake directory for it to have any effect, making it very ineffective as a tool of deception if that ever was ATI's goal. I can assure you it is not installed automatically by the ATI setup CD.

The RAMDAC issue looks like plain false advertising. However, this issue only affects people running extreme resolutions and has no affect at all on performance. A 350Mhz DAC can drive 1600x1200 at 100hz and 2048x1536 at 85Hz. How many people have monitors capable of more than that? As with power supplies, it's quality over quanity for RAMDAC's.

 

[MoobyTheGoldenCalf]

Ummmm, let's put the whole quote in here to get the right perspective:

It seems to me that if this particular set of Radeon 8500 drivers was optimized for the Q3 game engine that we would see identical sets of data above. Now whether or not this is seen as good or bad, you still get the elevated performance when running Quake 3. I am not going to push this issue any further but to simply ask for logical explanations as to why this would occur.

And as I said, "It seems . . ." I was not stating a fact, that's my opinion.

As for the RAMDAC issue, you are correct it is false advertising, no question. But please don't try to justify it but claiming "no one will use those features anyway". That is just pathetic.

 

[The Sauce]

<< The Radeon 8500 drivers are certainly far from perfect and it would shock me that anyone could call them "Fully Tested". An example of this would be the fact that the Vsync switch in DirectX is backwards. When you toggle it on, it is really off and vice versa. We fought with Max Payne benchmarks for hours trying to get a tweaking program to disable Vsync and then we learned that to disable all we needed to do was "turn it on". >>



Ok, that one got a chuckle out of me.

So now not only did ATI write $hitty drivers but they wasted time they could have spent making them better by designing them to cheat on benchmarks. This is going from woeful to pitiful.

 

[Pariah]

Quoting the entire quote makes no difference. They still make no implication that the benchmark results are inflated or cheated. They conclude that the performance increase is not due to driver improvements but probably due to actual code optimization. Wow, a hardware optimized version of a game, now there's a rarity and an obvious attempt to fool the public. I probably have a dozen or more special edition games in my room that are optimized to run on the piece of hardware they came bundled with.

 

[nortexoid]

every hear of S3 Metal?

you know, the optimizations made specific for Savage series cards for Unreal Tournie and same-engine based games?

it's the same thing....drivers/hardware recognize specific code and carry out specific optimization instructions accordingly...nothing wrong with that...
parallels MMX, SSE, etc....

if they're changing the conditions by which the card can recognize specific code, it will fail to recognize it and wont' carry out the specific instructions for optimized performance...no biggie.

the only problem i have w/ ATI at this point is their non-disclosure and inconsistency of their clock speeds...

 

[NFS4]

Quack Quack! HardOCP really did lay into ATi with that review...I don't think I've ever seen them ripped that bad:Q

 

[oldfart]

It was pretty brutal. OUCH!

 

[BlvdKing]

This is *almost* a mirror image of the original Radeon launch. Typical ATI. They will never change!

 

[AGodspeed]

It is obvious that all the features are not operating properly and testing was lackluster at best.

We found compatibility problems with the card after only spending ten minutes with it after getting the drivers installed.

We are certainly not giving up on the Radeon 8500 this early in the game, but ATi releasing this card with these problems sure makes me wonder if they have the ability to get the job done at all.

It truely amazes me that this card seems no better than it was two months ago when it supposedly was using beta drivers and silicon that wasn't truely final. The 8500 was actually pretty competitive then with the GeF3 a couple months back. And now it can barely hold it's own with a GeF3. If anything, the 8500 has gone no where since two months ago.

I was looking at the 8500's specs in HardOCP's review, and was completely drooling. Let's hope that ATi has a huge driver ace up it's sleeve to activate the potential these specs will bring.

 

[MrGrim]

Guys what are you talking about? It's the same game with different file names! What ever it is they did will only impact a game with the name quake, games using the Quake3 engine will not be benefited. If it was an optimization for a game engine why did they associate it with the file name?! That's just not right.

 

[The Sauce]

Thank you MrGrim...that's the whole issue. They are still lying and saying that they optimized it for the Q3 engine when this is clearly not the case. Any other game using the Q3 engine will not get the performance indicated on the benchmarks. If the only game you play (or plan on ever playing) is Q3 then by all means, the 8500 is super-duper. What they did was CHEAT, plain and simple....and now we know. The Q3 scores were artificially inflated. I'm just disgusted with the whole thing and I'm glad someone found out about this.