ExtremeTech Article Up

CosmoKramer said:
OT: So what does "Reve" mean? Just curious... :)
Unless my French has completely left me, it means "dream" (I left out the "ê" in "rêve" b/c of ASCII incompetence and general laziness ;)). I believe ED (Evildeus :eek: ) is French, so his spelling of "Rev" as "rêve" is probably just the French one for the same pronounciation (he probably doesn't mean "rêve" as in "dream").
 
Pete said:
CosmoKramer said:
OT: So what does "Reve" mean? Just curious... :)
Unless my French has completely left me, it means "dream" (I left out the "ê" in "rêve" b/c of ASCII incompetence and general laziness ;)). I believe ED (Evildeus :eek: ) is French, so his spelling of "Rev" as "rêve" is probably just the French one for the same pronounciation (he probably doesn't mean "rêve" as in "dream").
Yes rêve baby
 
jjayb said:
Someone posted this at the Nvnews forums:

ok, got word from futuremark that this 3dmark03 behaviour is a bug and that they are currently working together with nvidia to solve the glitch.

Expect further explanations from futuremark in the coming days..

It was posted by vacs here: http://www.nvnews.net/vbulletin/showthread.php?s=&postid=123105#post123105

Any truth to this? Worm?

If this is the same nv driver issues the nvnews people are talking about, let me ensure you that nothing has so far suggested that this would be a bug in 3DMark03. Also, I don't see how this could be just a bug. A bug that accidentally clears the back buffer only when the default cameras show the background sounds a bit too good to be true :LOL:
 
FiringSquad now has their own blurb up:

http://firingsquad.gamers.com/news/newsarticle.asp?searchid=5117

The Bad: They display a lack of technical understanding of what the "issue" actually is.

The Good: Even given that, they take a very sensical approach to the whole thing:

As a result, the GeForce cards we tested in last week’s articles were essentially doing less work than their rivals from ATI, boosting their final frame rate. Since the 3DMark 03 overall score is based on the frame rate results from game tests one through four, this not only invalidates the frame rates we reported, but also the final scores. Therefore, we’ve decided to remove these pages from our GeForce FX 5900 preview article, and have removed the affected NVIDIA cards from the corresponding graphs in our RADEON 9800 PRO review. We also won't be posting 3DMark 03 scores in our upcoming Detonator FX driver report.

According to NVIDIA, a bug in the Detonator 44.03 driver is the cause of this issue; some have speculated that the inflated 3DMark 03 scores are the result of cheating. Whatever the case, we hope this issue can be resolved quickly. While it has its critics (NVIDIA included), the relevance of 3DMark 03 cannot be denied – over 2.5 million copies were downloaded in its first month, despite its 100MB+ size.

1) They report on the issue (it's obviously newsworthy), and gave the "overall" two sides of the story: ET: it's cheating, nVidia: it's a bug.
2) They pulled 3DMark scores from reviews that are effected by it
3) They won't publish further 3DMark scores for nVidia products until the issue is resolved.
4) Recongnize the importance and relevance of 3DMark itselef.

Aside from actually trying to invistigate the issue yourself (which FiringSqaud is likely not capable of doing), this is about the best approach I can see having towards the whole mess.

One additional thing that I would do, is in the articles where the 3DMark scores are ommitted, have an explanation of WHY they are omitted. We'll see if FiringSquad does this when the publish their driver article.
 
... Leeeeet's Ruumblllle!!!

Quote - "Someone posted this at the Nvnews forums:
Quote:eek:k, got word from futuremark that this 3dmark03 behaviour is a bug and that they are currently working together with nvidia to solve the glitch. Expect further explanations from futuremark in the coming days.."

Consider the Source. I have yet to see any determination to this effect anywhere else. Just checked FM's website. Nothing there. No comment at Nvidia either. No surprise there actually. They seem to be adopting a wait and see attitude, probably hoping this will blow over. I personally hope it blows up! I did find this little tidbit of news though... this is a gem.

"SANTA CLARA—MAY 14, 2003—NVIDIA Corporation, the worldwide leader in visual processing solutions, today announced that the Stanford Business School Alumni Association has named NVIDIA as the Entrepreneurial Company of the Year for 2003."

That takes the cake!


And this bit from an earlier post...
"Finally, i don't see your point. Every website has their customers. I like the way B3D does, but i also like [H] reviews. They are not directed to the same public. Moreover, i think more people like "kickass" reviews than you think"

I do not care about "Kickass" as much as I care about Integrity. If it is Honestly Kickass that's one thing, If it is kickass because of cheating, that's something entirely different.

I feel sorry for "Joe Public", who doesn't see these Forum discussions and hasn't a clue he's being cheated out of his money with bogus Benchmark Numbers. In my Book this is Theft by Deception. When you make purchasing decisions based on intentionally Falsified data, your being Ripped off. Since the Benchmark results are defacto Advertising for the Company, this starts looking like False Advertising to me.

Unless Both ATIs' and Nvidias' cards are doing the same amount of work each run, as intended, the results are worthless. The Benchmark is supposed to measure how Fast a Given work load for the scene is run. If Card 1 is doing X rendering workload and Card 2 is doing X-20% workload the results are Bogus and misleading. I don't like being misled.

Yes! I'm sure ATI has Optimized it's drivers for 3dMark03, and Nvidia had the same opportunity until they dropped out of the Beta Program. By that time, Nvidia pretty well had their drivers tweaked anyway. It seems to me that Optimizing the driver code to render the scene faster such as combining several Shader passes down to one pass or something similar is different from reducing overall scene complexity by not rendering the bits not seen by the viewer. The first one increases efficiency while running all the work, the other is only rendering Part of the work. Not the same at all.

But hey!, If Nvidia wants to play that way, then let's let ATI do the same thing and we'll run the benches again on an even playing field. Now what are the Bench results? I think the "Whoa!!! FX5900Ultra Rocks" posts will be as rare as a Virgin at a Frat Party.

my last 4 cards were Nvidia products. My next Purchase won't be. I'll speak with my $$$.

... and thanks for the linkage Joe. After reading it, I think FiringSquad handled the situation in a very Intelligent manner. Just the Facts and what they planned to do until the issue was resolved.
 
martrox said:
Wasn't it Rene Descarte who said, I think....therefore I think nVidia cheated? :LOL:

Heh...Heh...close, but this is the actual quote from nVidia, I believe:

nVidia: "I cheat, therefore I am."



(Sorry, couldn't resist).....;)
 
ladies and gentlemen we have a winner


But hey!, If Nvidia wants to play that way, then let's let ATI do the same thing and we'll run the benches again on an even playing field. Now what are the Bench results? Whoa!!! FX5900Ultra Rocks" posts will be as rare as a Virgin at a Frat Party
 
beyondhelp said:
... Leeeeet's Ruumblllle!!!

Quote - "Someone posted this at the Nvnews forums:
Quote:eek:k, got word from futuremark that this 3dmark03 behaviour is a bug and that they are currently working together with nvidia to solve the glitch. Expect further explanations from futuremark in the coming days.."

People working at Futuremark have denied this (think it was on either these forums or rage3ds, forgot the link :( ).
 
People working at Futuremark have denied this (think it was on either these forums or rage3ds, forgot the link ).

Heh, it was in this topic. Just a few replies down from my original post. ;)
 
Xspringe said:
beyondhelp said:
... Leeeeet's Ruumblllle!!!

Quote - "Someone posted this at the Nvnews forums:
Quote:eek:k, got word from futuremark that this 3dmark03 behaviour is a bug and that they are currently working together with nvidia to solve the glitch. Expect further explanations from futuremark in the coming days.."

People working at Futuremark have denied this (think it was on either these forums or rage3ds, forgot the link :( ).


As of that writing, I could find no reference to the issue on FM's website or Nvidia's(Of Course). I was deriding the post as unsubstantiated. my $.02.

What's the Problem?
 
In the end this just helps we consumers become better and smarter shoppers. I like to take and try a product first and draw my own conclusion approach. Hmmm...that is why I buy from retail instead of online, because if I don't like it I take it back enough said.

I still have the Banshee, V3 3000, V5 5500, G3 Ti 200, G4 Ti 4400. I did not like any Nvidia card until the G3 series, didn't like Redention cards, S3 2k card (was ok but was a broken chip from the start). Maxtox, just never tried them and that Intel card that Intel half ass worked on drivers for.

This is the first ATI I have used in a while that I have kept in my system. Tried the 8500 and didn't like (drivers at the time). After the 9700 pro came out I was going to jump on the ATI bandwagon for a while but held off until drivers improved. Now I have had a 9800 pro since April 3rd and have had no problems in any of my games. Matter of fact, any card I have had I have never had any real problems unless they were already broke.

I would say that 3dfx always had the best drivers when they were around. To expand on that, I think we should hold all GFX vendors to high standards as far as drivers out of the box, and broken hardware upon release (especially when they know it is broken).

Nvidia should offer all GeForceFX 5800 Ultra non Ultra owners refund for their products and they should receive a 5900 Ultra. We need a IHV Governing or Guidance Council to police vendors hardware and drivers.

B3D keep up the good work. I have been reading your site from the start when you were around, disappeared, and came back.

LOL, this is my small rant. Now back to hiding. hehe
 
Ok it took me a few hours to finish reading every post except ED's I skipped over his rubbish after the first few times I seen him post.

We need to clear the air on the ATI quak compared to Nvidia 3dmark and be done with it. I think you have to look at a few things like:

1. What did they have to gain ? Was their a big jump in performance ?
2. What negative effect did the driver have ?
3. Did the company address the issue or hide ?
4. Was the issue fixed and if so did the performance stay the same ?

OK ATI First

1. They had everything to lose at this time ATI was making a name for itself as a better IQ product with the near the same power as nvidia cards, the only benefit was a slight performance increase.
2. Some alpha texture quality wwas degraded, a overall bad thing because ATI is known for better IQ.
3. ATI did interview on the issue very quickly they claimed it was a bug and would be cleared up with the next patch.
4. The issue was in fact fixed and performance was on the same level as the previous drivers, while the IQ problem was fixed as well.

A few things to note ATI at the time was known as having bad driver support , yet many could not accept this as the bug it was. The new driver silenced many critics but not hardcore nvidia enthusiast who still refer to the whole issue negatively as the quack fiasco. Tom's hardware soon after released a play by play on how nvidia fed hardops the information, he did not discover maybe that's why this non-issue is still a issue.

About Nvidia

1. Everything to gain by releasing drivers that allow them to claim the lead in benchmark perfomance giving them the speed crown and sales. YES.
2. IQ in AA, textures are sacrificed for speed, clipping/culling issues allowing them to do less work yet seem to be doing the same work faster than competitors.
3. Other than blaiming their onw card for doing it and 3dmark for favoring ATI, nvidia has maintianed media silence on the issue.
4. Seeing how what the card was rendering was affected it is highly doubtful it can redo the test properly and retain the same scores, but as of yet nothing has been said or done.

A few notes, while i don't expect tom's hardware to post this news on their first page I do expect them to handle this like they handled the ATI issue. And as for sites like anaandtech, hardop and etc who went nuts over the ati issue they should have launched the same type investigation into this nvidia issue. I question the crediability of any site who hasn't done so so far.

The verdict is when looking at the 2 issues side by side they aren't even close to being the same issue. And as far as I am concerned nvidia is cheating, until they can step up and not only say why but show us why they weren't. All of the evidence points to it so, and that's where nvidia needs to start.
 
DaveB and FM: has anyone found a way to switch this 'feature' of the NV drivers on and off a la quak?
 
Back
Top