PDA

View Full Version : I want to buy a 5900U but all this cr@p is putting me off


Pages : [1] 2

ZenmasteR
06-06-03, 06:14 AM
this cheating/optimization business is going to harm NVIDIA and not 3dmark2003 imo.

I want to buy a FX5900U but nvidia are shooting themselves in the foot with all this malarky, it makes me wonder if timedemo scores are compromised aswell..

panzaman
06-06-03, 06:52 AM
.....of course they are.........

..potentially every public benchmark/demo can be cheated, nVidia drivers department evaluates how much relevance has this benchmark/demo in hardware reviews and then starts working on "optimizations".

I am tellin' ya, they have just opened a Pandora's Box.

pgn.inertia
06-06-03, 06:53 AM
Have you _ever_ seen _any_ reviewer state that NV35 distorts the picture or anything. They did on 3DMark03, so if it would happen on other benchmarks or games, you would've noticed...

gstanford
06-06-03, 07:07 AM
Don't let the fanATIcs get to you.

Remember it is still very early days for NV3x drivers, and this is where most of the problems stem from.

The NV3x is the first really different core nVidia has had since the TNT-1 (all the geforces are basically TNT architecture at heart).

Just like it took nVidia a while to produce the first detonators for the TNT series, it is taking them a while to find their feet with NV3x.

I think NV3x is the biggest challenge nVidia's unified drivers have ever had to face.

Couple that with the manfucaturing issues, and NV3x's multi-precision design that it isn't allowed to exploit properly yet in direct X at least (we'll see what happens with actual game support), and you can see why the picture looks less rosy than it should.

extreme_dB
06-06-03, 07:11 AM
Originally posted by pgn.inertia
Have you _ever_ seen _any_ reviewer state that NV35 distorts the picture or anything. They did on 3DMark03, so if it would happen on other benchmarks or games, you would've noticed...

Since when has any cheat been readily noticeable? It takes investigation to discover these kinds of cheats.

The hardware-reviewing community had to be tipped off by ATI, just like they had to be tipped off by Nvidia with Quack. So your assertion is false.

panzaman
06-06-03, 07:13 AM
....:rolleyes: ...uhm..

if u want to call Clipping planes, replacement ov shader's code, FP16 precision instead of FP32 and deactivation of AF a "very early days for NV3x drivers" well you must have an abstract vision of the whole issue mate!

gstanford
06-06-03, 07:17 AM
Panzaman,

That stuff is specific to debunking 3DMark2003 which nVidia publiclly stated they were going to do.

It applies nowhere else (unless you have hard evidence to prove otherwise).

panzaman
06-06-03, 07:18 AM
Originally posted by extreme_dB
Since when has any cheat been readily noticeable? It takes investigation to discover these kinds of cheats.

The hardware-reviewing community had to be tipped off by ATI, just like they had to be tipped off by Nvidia with Quack. So your assertion is false.

..yeah, but hang on a minute, the need to cheat in those benchmark it must mean something to us mustn't it???

By definition if u cheat means that you can't reach a result in a fair way, it means, in this case, the NV hardware is not up to what they hyped and most of all is not up to the competition despite the fact that is some instances is priced 40% more than the top offer from a competitor.

pgn.inertia
06-06-03, 07:19 AM
Originally posted by extreme_dB
Since when has any cheat been readily noticeable? It takes investigation to discover these kinds of cheats.

The hardware-reviewing community had to be tipped off by ATI, just like they had to be tipped off by Nvidia with Quack. So your assertion is false.

Investigation is going on all the time by the reviewers out there, and they did notice a noticable degrading in IQ (look to some 3DMark03 pictures @ Digit-Life) and a noticable increase in 3Dmarks that could not be explained right-away...

extreme_dB
06-06-03, 07:29 AM
Originally posted by gstanford
Panzaman,

That stuff is specific to debunking 3DMark2003 which nVidia publiclly stated they were going to do.

It applies nowhere else (unless you have hard evidence to prove otherwise).

That's ridiculous.

There's no debunking going on through this kind of cheating because ANY benchmark/timedemo can be cheated in the same way.

Why would Nvidia repeatedly try to make a point that they already made public? They've only made liars of themselves by saying they wouldn't waste time optimizing for 3dmark.

gstanford
06-06-03, 07:37 AM
Wh is it ridiculous?

It is precisely what happened.

BTW the clipping planes are a perfectly legitimate optimization. Why? Because in the freely/publically available and downloadable versions of 3Mark2003, the user is "on a rail" and cannot alter his viewpoint in the scenes or magnify the scenes. Since there is no opportunity for the viewpoint to change, rendering anything outside of that viewpoint is a total waste since it will never be seen.

The only way futuremark can prevent this from happening is to release a freely/publically available and downloadable version of 3DMark2003 where the user can alter the viewpoint at anytime, in any scene.

extreme_dB
06-06-03, 07:38 AM
Originally posted by pgn.inertia
Investigation is going on all the time by the reviewers out there, and they did notice a noticable degrading in IQ (look to some 3DMark03 pictures @ Digit-Life) and a noticable increase in 3Dmarks that could not be explained right-away...

That doesn't rule out the fact that more cheats may be discovered in actual game benchmarks.

Digit-life actually believes that the clipping plane "optimization" is valid because it doesn't affect visible IQ! :rolleyes:

extreme_dB
06-06-03, 07:44 AM
Originally posted by gstanford
Wh is it ridiculous?

It is precisely what happened.

BTW the clipping planes are a perfectly legitimate optimization. Why? Because in the freely/publically available and downloadable versions of 3Mark2003, the user is "on a rail" and cannot alter his viewpoint in the scenes or magnify the scenes. Since there is no opportunity for the viewpoint to change, rendering anything outside of that viewpoint is a total waste since it will never be seen.

The only way futuremark can prevent this from happening is to release a freely/publically available and downloadable version of 3DMark2003 where the user can alter the viewpoint at anytime, in any scene.

If the benchmark has "invisible" work to do for the GPU in an attempt to better simulate gaming conditions, and the GPU discards it, that's a cheat.

What if the benchmark included AI routines in the background that are totally unseen by the viewer? Would disregarding them when they were specifically put there for a reason to boost performance be considered valid? I don't think so!

3dmark03 involves more than what is visible to the viewer. Futuremark themselves said the clipping planes were invalid! It doesn't matter what we argue, that's the fact.

deejaya
06-06-03, 08:05 AM
Originally posted by extreme_dB It doesn't matter what we argue, that's the fact. [/B]

:p Going by your logic, Futuremark also stated nVidia weren't cheating, so it doesn't matter what we argue, that's the fact. ;)

extreme_dB
06-06-03, 08:15 AM
Originally posted by deejaya
:p Going by your logic, Futuremark also stated nVidia weren't cheating, so it doesn't matter what we argue, that's the fact. ;)

Exactly. Futuremark said they weren't cheating, but they did say they were using "invalid optimizations". However, that's just a politically-correct term for cheating. :D

In any case, what Nvidia did is NOT allowed - end of story.

jbirney
06-06-03, 08:17 AM
Originally posted by deejaya
:p Going by your logic, Futuremark also stated nVidia weren't cheating, so it doesn't matter what we argue, that's the fact. ;)

Here in the states to claim legally that one company is cheating its a very serrious thing. We know that there were some legal ramifications and thus 3dmark removed the word Cheat from its PDF..thanks to nV lawers.

solofly
06-06-03, 11:27 AM
Originally posted by ZenmasteR
this cheating/optimization business is going to harm NVIDIA and not 3dmark2003 imo.

I want to buy a FX5900U but nvidia are shooting themselves in the foot with all this malarky, it makes me wonder if timedemo scores are compromised aswell..

That is why I purchased ATI for the time being. Not that I can't afford 5900 as well but I'll sit back and see what happens. ATI fanboys sure make it more difficult to be an nVidia fan these days. Keep it mind it's not over till it's over and we are far from it!
Cheers

Kihon
06-06-03, 02:28 PM
Originally posted by ZenmasteR
this cheating/optimization business is going to harm NVIDIA and not 3dmark2003 imo.

I want to buy a FX5900U but nvidia are shooting themselves in the foot with all this malarky, it makes me wonder if timedemo scores are compromised aswell..

End of the day ask your self two things

1. Will it play the games I want it to play, with the quality/speed I want to play them at?

2. Am I comfortable going with a company like NVidia?


The second is only important to some people, and if it doesnt matter to you - ignore it.

The first is the key thing. Check out reviews where games of the type you like to play are being tested with the cards, and where you can compare things like IQ.

Check out the forums to see how good support is - usually, you will find some ranting/raving going on, but check what kind of responses you get.

Then make your decision.

extreme_dB
06-06-03, 03:03 PM
Originally posted by solofly
That is why I purchased ATI for the time being. Not that I can't afford 5900 as well but I'll sit back and see what happens. ATI fanboys sure make it more difficult to be an nVidia fan these days. Keep it mind it's not over till it's over and we are far from it!
Cheers

The only reason it's "difficult" to be an Nvidia fan these days is because of the company's deceptive actions and PR lies. The cards themselves are not bad if you value their strong aspects much more than their weak aspects, despite any cheating. But there's no sense taking personal offence to criticisms of the cards.

If the 5900U is not as fast as the 9800Pro during actual gameplay, or is faster with IQ hacks, then that in itself doesn't make it a bad choice, because there are lots of other things to consider in a video card.

If you're a fan of a company and want to support it by buying its product even when the competition has an overall superior product, that's a valid choice. Maybe you're more comfortable with that company's version of things, and prefer the distinct qualities and unique style of its products, or maybe the reason has more to do with the company itself than the product.

For me, I think that if Matrox came out with a highend card that was at the performance level of a 9500pro but cost as much as a 9800pro, then I'd probably buy it, not only for its traditionally strong 2D quality and features, but also out of brand loyalty and to support a third player in the industry. :)

That doesn't mean I would bash people for criticizing my card, or blindly recommend my card to others, or deny its shortcomings, or blindly defend the company when a scandal arises. I would however explain and defend my reasons for my satisfaction with the card, and perhaps recommend it on that basis.

Nvidia is in an entirely different position though as the dominant company in the industry. Being at the center of attention, it's expected that they'll be continually scrutinized and criticized where appropriate. Except when they pay off the media. :p

It's the same with a superstar athlete or a movie star, or anybody that rises to prominence. :)

StealthHawk
06-06-03, 03:35 PM
Originally posted by pgn.inertia
Have you _ever_ seen _any_ reviewer state that NV35 distorts the picture or anything. They did on 3DMark03, so if it would happen on other benchmarks or games, you would've noticed...

No, you wouldn't have noticed. Remember the NV30 previews and how [H] and Anand criticized nvidia's QCA and 2x FSAA? All because the screen caps were incorrect. Did they even play the games? Apparently not, because they didn't realize that the screenshots didn't mimic what they saw in the actual game.

Same thing happened with the NV35 previews. Some sites said that NV35 when using 4x FSAA and AF blurred the screen. Again, it is apparently a screen capture issue.

With these multiple blunders from multiple sites, you have to wonder how much time these reviewers spend familiarizing themselves with the cards by playing with them. Additionally, I doubt most reviewers actually watch the benchmark runs, they use tools to automate it.

digitalwanderer
06-06-03, 03:53 PM
Originally posted by solofly
ATI fanboys sure make it more difficult to be an nVidia fan these days.
But you're really giving us credit we ain't earned. All we do is point out the reasons that nVidia is giving you to not be an nVidia fan, the rest is all up to you.

Don't give us the credit, give it to nVidia...they've more than earned it. :)

solofly
06-06-03, 03:55 PM
Originally posted by extreme_dB
The only reason it's "difficult" to be an Nvidia fan these days is because of the company's deceptive actions and PR lies. The cards themselves are not bad if you value their strong aspects much more than their weak aspects, despite any cheating. But there's no sense taking personal offence to criticisms of the cards.

Good point but at the same time nothing new. Obviously itís the people that make things happen and not video cards alone. I guess itís all about Ati users trying to convince nVidia users to jump the ship. Unfortunately the problem is, ATI users see no boundaries where to stop and keep beating on a dead cat. There are people that feel good about buying an nVidia card just simply because itís an nVidia card and that is usually based on a past experience. Now I say what is wrong with that? Nothing, but you know, Ati users canít seem to handle it and it shows...

ricercar
06-06-03, 04:14 PM
Buying an NVIDIA video card does mean the purchaser endorses the quality and functionality of the drivers.

With the current fiasco, it is nearly impossible maintain unshaken confidence in NVIDIA drivers when they are obviously and unarguably enhanced for benchmark performance instead of a better gaming experience. Is that really the way it's meant to be played? I'd rather have NVIDIA spend those person-hours on the gaming experience, thankyouverymuch.

These days I think often about a friend who warned me about one NVIDIA interview question, "It is more important to win or to do your best?" To get the job at NVIDIA, you had to answer "to win."

If you want buy a card from NVIDIA these days, you'll be tacitly deciding that what NVIDIA defines as "winning" also fits into what you want in a video card.

<--beleaguered NVIDIA fanboi

extreme_dB
06-06-03, 04:24 PM
The same old arguments go on and on probably because the diehard Nvidia fans simply won't acknowledge the validity of criticisms made by ATI fans, and counterattack instead. At the same time, the ATI fans don't acknowledge the criticisms by Nvidia fans, as well as the reasons for sticking with Nvidia. It's everyone's fault. :)

I myself am biased against Nvidia at the moment (meaning that if both companies had the same product, I'd definitely choose ATI) simply because of their shameful business practices of late.

But I believe I'm arguing the issues as fairly as possible.

I've been using Nvidia in my computer for over 4 years now, but my next card will be an ATI. If I end up experiencing problems with the ATI that are serious or too much of an inconvenience, I'll return it. It's as simple as that.

de_telegraaf
06-06-03, 04:25 PM
Originally posted by ZenmasteR
this cheating/optimization business is going to harm NVIDIA and not 3dmark2003 imo.

I want to buy a FX5900U but nvidia are shooting themselves in the foot with all this malarky, it makes me wonder if timedemo scores are compromised aswell..

Buy the card NVDIA Rulez!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!:rw: :rw: :rw: