PDA

View Full Version : 3DVelocity Review- interesting results


Pages : [1] 2

CapsLock
06-13-03, 07:47 PM
anyone check out the review as was posted on the front page?

http://www.3dvelocity.com/reviews/5900u/5900u.htm

It seemed to be about equally biased both ways!

The gamebench marks were a toss up, even though he was only comparing against a 9700pro. There were a few shockers for me, as I expected the 5900U to easily beat a 9700P. Such as:

in No one lives forever 2 (no rez given)
5900U 9700P
0aa 0af 73.7 94.0
0aa 2af 70.1 87.6
0aa 4af 64.7 84.1
0aa 8af 61.1 83.1
0aa 8af (16Xaf on radeon) 56.2 67.3
4aa 8af (16Xaf on radeon) 43.3 59.2
8aa 8af (6Xaa 16af radeon) 33.7 48.5

This was recorded with FRAPS 1.9D from the opening animation sequence.

WTF!!??? Thats not even close. At any point. And using fraps should be making it quite objective. Could drivers be still that big of a problem here?

Then its also really weird in UT2K3 and F1 2002. Check out page 9: http://www.3dvelocity.com/reviews/5900u/5900u_9.htm The only time the 5900U won clearly in UT2K3 was without any AA or AF. When any was applied the 9700P won the majority! (and sometimes by a good margin)

Could someone please explain this to me. I thought the 5900U was supposed to be a 9800P beater never mind the older 9700P.
I really don't get it.

CapsLock

attempted edit for clarity, stupid formatting in program

digitalwanderer
06-13-03, 07:49 PM
Originally posted by CapsLock
Could someone please explain this to me. I thought the 5900U was supposed to be a 9800P beater never mind the older 9700P.
I really don't get it.
Mebbe the 5900 is a flop too and no amount of nVidia PR assault and site strong-arming is gonna be enough to hide that fact?

(Remember, I said it was a POSSIBLE explanation...far be it from me to accuse nVidia of anything! ;) )

GlowStick
06-13-03, 08:07 PM
if you look at other reviews you will see that the 5900ulra wins in ut2k3, i woudlnt be to worried about this one

digitalwanderer
06-13-03, 08:19 PM
Originally posted by GlowStick
if you look at other reviews you will see that the 5900ulra wins in ut2k3, i woudlnt be to worried about this one
Hmmmmm...are you saying that nVidia's newest card wins in the first game that joined their "Way you were meant to be played" campaign? Was that using the same everything settings or d'ya think nVidia cards might have a SLIGHT advantage on that one... :eek2: ;) :lol:

CapsLock
06-13-03, 08:52 PM
Originally posted by GlowStick
if you look at other reviews you will see that the 5900ulra wins in ut2k3, i woudlnt be to worried about this one

well so what r you saying, these results (all of them for UT2K3) are wrong?

Don't you think that maybe anyone with the balls to go against nvidia (and suffer the consequences of displeasing a corporation that has 1.3 billion in liquid assets) might be telling the facts?

I'm not being a fanATIc but it seems like most web reviews have been biased towards NV for that reason.

Not to mention he went to the bother of using FRAPS, which seems to indicate an attempt at honest results.

Wouldn't that reviewer, unlike any who erred the other way, be asking for serious trouble if his review results were wrong?

CapsLock

(I really don't know, it just seems with all the NV cheat scandals going on, that negative results seem more believable.)

GlowStick
06-13-03, 08:55 PM
just look at every other review site who beggs to differ.

digitalwanderer
06-13-03, 09:00 PM
Originally posted by GlowStick
just look at every other review site who beggs to differ.
Replace "beggs to differ" with "sold out or is trying to" and I might just agree.

Let's face it, we can't trust ANY reviews out there about nVidia anymore. :(

CapsLock
06-13-03, 09:03 PM
fine whatever, I can't really prove anything myself. But what about NOLF2 results and F1 2002 results? I mean c'mon! This is only a 9700pro, what the hell is going on here?

CapsLock

G6-200
06-13-03, 09:08 PM
That AA comparison has to be the flat out worse I have ever seen, what exactly are we supposed to be looking at?
The GFFX gun comparison does illustrate how crappy the 6x and 8x modes are though.

G6

Hellbinder
06-13-03, 09:17 PM
just look at every other review site who beggs to differ

Well actually more than one review shows the 9800pro beating the 5900U in UT2003 with AA+AF depending on the level. Some levels the 9800pro wins by a decent margin. Remember the Drivers used on the 9700pro give it about a 15% boost in AA+AF scores.

CapsLock
06-13-03, 09:27 PM
Originally posted by Hellbinder
Well actually more than one review shows the 9800pro beating the 5900U in UT2003 with AA+AF depending on the level. Some levels the 9800pro wins by a decent margin. Remember the Drivers used on the 9700pro give it about a 15% boost in AA+AF scores.

I was just cruzin around waiting for a reply when I found a new UKGamer review of the MSI 5900U. For the UT results click:

http://www.ukgamer.com/article.php4?id=3&page=8

The 5900 seemd to do ok at 1024 but again even the 9700 beat it bad at higher rez. There's no way this is drivers as Dig pointed out, this game has the NV logo on it and would be THE first one they would code for.

Frankly these scores SUCK. There's no way I'm going to pay 400 freakin bucks for the latest card when it doesn't crush the competition. It might be time to save some money and get a 9700.

Caps

On the other hand looking at the splinter cell scores (same review) http://www.ukgamer.com/article.php4?id=3&page=9
maybe I need to look at more reviews. CRAP

edit for spell/grammer/link

digitalwanderer
06-13-03, 09:31 PM
Originally posted by CapsLock
It might be time to save some money and get a 9700.
Not meaning to sound too much like an ATi fanboy, but I really do think that would be your best bet. You can pick up a 9700 Pro now for around $250-300us and it'll pretty much play any game out there right now as well as the 9800 Pro will, it's a much better "bang-for-buck" solution than either the 9800 Pro or the 5900 Ultra.

But I am biased since I have one and love it to death and all and am still jaw-dropping amazed at what it can do. ;)

Hellbinder
06-13-03, 09:42 PM
Caps you should know that the idiot reviewer supposedly doing AA tests in splinter cell... :rolleyes: :rolleyes: :rolleyes:

Splinter cell Does not allow AA to be on with any Nvidia card regardless of Driver setting. It disables it internally. Ati cards Still use AA but it is Buggy. This is becuase of rendering issues With Splinter cell itself. There are a lot of posts about this at B3D from the Splinter Cell Lead Programer.

So what you are really seeing is the FX playing the game with only AF being applied against AA+AF on the ATi card. Somehow this Rocket Scientist Reviewer failed to notice all the Jaggies on the Nvidia board.. little odd eh??? :rolleyes: :rolleyes: :rolleyes:

digitalwanderer
06-13-03, 09:46 PM
Originally posted by Hellbinder
Somehow this Rocket Scientist Reviewer failed to notice all the Jaggies on the Nvidia board.. little odd eh??? :rolleyes: :rolleyes: :rolleyes:
My personal theory is that nVidia is now including a special set of glasses that they're requiring the reviewers to wear while reviewing their cards... :eek2:

;) :lol:

Morrow
06-13-03, 10:24 PM
Originally posted by digitalwanderer
My personal theory is that nVidia is now including a special set of glasses that they're requiring the reviewers to wear while reviewing their cards... :eek2:

;) :lol:

digi, would it hurt you much if you just stopped posting useless biased posts here? You topped yourself here in this thread. It's so pathetic...

and congrats to you, you again managed to make a perfectly legitimate thread useless by adding your comments. Keep on with your good work and ATI will continue to send you more Radeons the next few months...

rakestone
06-13-03, 11:17 PM
http://www.3dvelocity.com/reviews/5900u/5900u_15.htm

Hellbinder
06-14-03, 12:08 AM
Ok now i am going to be SICK!!

So many people have now accused me of screwing up the benchmarking based on our UT results that I figured I should set the record straight. I quite clearly demonstrated in the review how setting the Mipmap detail level to either "Best image quality" or to "Best performance" makes very little difference but obviously UT is a different kettle of fish. The following scores (which were run on a 5800 Ultra for the record) on Detonators 44.03 were recorded running [H]'s script at 1024x768 with 4xAA and 4xAF. The difference is, as I hope you can see, substantial. Running UT with the Mipmap detail level set to "Best performance" at 12x10 would probably have put the 5900 Ultra ahead of the Radeon, as it was on all the other sites that everyone keeps banging on about, however the image quality would have vastly inferior.

You guys asked for a like-for-like comparison and I hoped that having large boxes on top of each page showing the driver settings used would help it seems it hasn't stopped the 99% of superb feedback I've had from being peppered with verbal abuse from people accusing us of being in ATi's pocket. It's a shame but it's true.

So there we are, our UT scores are lower because I set the drivers to a mipmap detail level that offered a fairer level of image quality.

The tests were alreay done to start with with ATi's AF set to 16x and Nvidias to 8X. But Supposedly thats still not fair enough??? Give me a freaking break!!! I cant believe Some Complete Fanboys claiming the benchmark was Biased towards ATi wanted them to rerun the test with Best PERFORMANCE selected on the Nvidia card???? Claiming that somehow the IQ is more Equal?? That is Complete CRAP!!

G6-200
06-14-03, 12:49 AM
Originally posted by Hellbinder


So what you are really seeing is the FX playing the game with only AF being applied against AA+AF on the ATi card. Somehow this Rocket Scientist Reviewer failed to notice all the Jaggies on the Nvidia board.. little odd eh??? :rolleyes: :rolleyes: :rolleyes:

It is quite mind boggeling that review after review gets this so wrong. Maybe the reviewer has someone run the tests for them and they just dont pay attention to what's being displayed.
It's nice at least to see that UKgamer review point out how bad nVidia's AA looks compared to ATI's, and the last paragraph of the conclusion is somewhat satisfying.

G6

CapsLock
06-14-03, 12:50 AM
ok, so there were other people out there as shocked as I was about the less than stellar performance of the 5900.

and I was right, judging by the reviewers prompt response that he was trying to be as honest as possible, whereas other reviews it seems, were not.

hellbinder-
sorry I forgot about the splinter cell problem, very dumb of me.

so, I still like an explanation for these results. the 5900 has 135 million transitors clocked significantly higher than a 9700 (450 vs 325) which only has 110 million transistors. plus it has hugely faster ram as well (425 vs. 310).

HOW could this happen?
WHO'S been smoking what??!!!
WHAT the freakin bleep is going on at NV?

Caps

lukar
06-14-03, 12:53 AM
NV35 is bad card. There are only two benchmarks where NV35 wins over r350, guess what? quake 3 and serious sam 2. Really pathetic...

Ninja Prime
06-14-03, 01:10 AM
NV35 is bad card.

I wouldn't say exactly that. I'd say that NV30 was a bad card, and since NV35 is derived directly from NV30, it suffers from most of its problems, however, nVidia has done a pretty good job fixing some of the crap from NV30. Overall, NV35 is pretty good, but it still suffers from the major problems of NV30, and thus still loses to R3XX.

Hellbinder
06-14-03, 01:23 AM
NV35 is pretty good, but it still suffers from the major problems of NV30, and thus still loses to R3XX.

Its really more like Ties it. Each one leads depending on the game and settings.

I think it would be hard to argue however that the 9700pro (R300) Grandpa that it is.. is not the best card for the money on the market today. Its pretty close to the speed of the big boys. Has all the features you will need for a while and kicks some serious frame Rate Booty with 4x FSAA and 8xAF at 1024x768 in pretty much every game.

Its the best Video card I have ever Owned. The only one that gave me a similar feeling of satisfaction was my Voodoo 2 SLI Rig from back in the day.

digitalwanderer
06-14-03, 02:24 AM
Originally posted by Morrow
digi, would it hurt you much if you just stopped posting useless biased posts here? You topped yourself here in this thread. It's so pathetic...
I'll stop making fun of them if nVidia cuts with all the silliness and starts shooting straight a bit. :lol:

Keep on with your good work and ATI will continue to send you more Radeons the next few months...
Really? You think so? Ok, but only since you insisted.... :lol:

:jumping: :jumping: :jumping:

Dazz
06-14-03, 04:36 AM
I don't know some of them pixle shader results looked fixed to me as at first even the Radeon 9700Pro butched the FX5800Ultra, now we have them leading with no less then 25%. Also if they wanted a true comparaion they should of used the Radeon 9800Pro.

StealthHawk
06-14-03, 05:20 AM
Originally posted by CapsLock
anyone check out the review as was posted on the front page?

http://www.3dvelocity.com/reviews/5900u/5900u.htm

It seemed to be about equally biased both ways!

The gamebench marks were a toss up, even though he was only comparing against a 9700pro. There were a few shockers for me, as I expected the 5900U to easily beat a 9700P. Such as:

in No one lives forever 2 (no rez given)
5900U 9700P
0aa 0af 73.7 94.0
0aa 2af 70.1 87.6
0aa 4af 64.7 84.1
0aa 8af 61.1 83.1
0aa 8af (16Xaf on radeon) 56.2 67.3
4aa 8af (16Xaf on radeon) 43.3 59.2
8aa 8af (6Xaa 16af radeon) 33.7 48.5

This was recorded with FRAPS 1.9D from the opening animation sequence.

WTF!!??? Thats not even close. At any point. And using fraps should be making it quite objective. Could drivers be still that big of a problem here?

Then its also really weird in UT2K3 and F1 2002. Check out page 9: http://www.3dvelocity.com/reviews/5900u/5900u_9.htm The only time the 5900U won clearly in UT2K3 was without any AA or AF. When any was applied the 9700P won the majority! (and sometimes by a good margin)

Could someone please explain this to me. I thought the 5900U was supposed to be a 9800P beater never mind the older 9700P.
I really don't get it.

CapsLock

attempted edit for clarity, stupid formatting in program

Simple. The drivers are optimized for games used in benchmarks, and not games like NOLF2, which aren't used widely.

From what I've been hearing, it seems like ATI's drivers/cards perform better than nvidia cards in games that aren't used as benchmarks, while nvidia dominated in games used as benchmarks. I would like to see some Serious Sam numbers recorded with FRAPS, as opposed to the timedemo though. Same for UT2003.