PDA

View Full Version : Tired of FX5900 Bashing


Pages : [1] 2

cwalker
06-14-03, 11:46 PM
I am really geting tired of all the FX5900 bashing. The reviews I have read are pretty much bashing the performance of the FX5900.

Let us remember that this card is just hitting the streets.

Plus, the drivers have not been optimized yet.

I would also like to see more real world "tests". All of the synthetic benchmark programs mean nothing. Benchmarking is only worthwile to see if your configuration has any serious problems (ie: your 3DMark2001 is only 6000, and should be up near 13,000), then you know to look for problems.

How about more "real world" testing on games showing FPS etc.
Until the newest drivers (Det 50.xx) hit the streets, maybe everyone should just calm down a bit and wait.

Quint
06-15-03, 12:18 AM
Where are these reviews?
I've only seen good ones...sure, theyre not the be all end all review, that says, BUY THIS CARD-IT IS THE BEST!...but theyre good.
I'm really looking forward to this card, and will hopefully get it soon. (128)

~Q~

Skynet
06-15-03, 01:34 AM
Nvidia only has themselves to blame. They pumped up the FX5800 to be the greatest piece of silicon ever made. Then came the salvation chip known as the 5900, twice as fast as the 5800 etc. blah blah.

And the 5900 is still barely making its way into the channels. Immature drivers? No, just flawed design. When Nvidia shows us a world-class bug free product we will stop reigning down on them.

sbp
06-15-03, 01:47 AM
There does seem to be an attitude amongst certain people the 5900 is nothing more than chopped liver.

nVidi0t
06-15-03, 02:22 AM
Well, it's perfectly understandable why everyone is bashing nVidia/FX5900. It's what the NV30 was meant to be and because the NV30 was a little dissapointing it left a sour taste in most users mouths, which will probably last through the entire NV3* range.

I doubt anyone will give nVidia a chance or a break until the NV40 is released and they fix the FX architecture which uses a 3 year old AA technique.

About drivers, come on. nVidia already used clip planes, app detection tweaks and water shader "optimization". I dont think they can milk benchmark scores any more.

What I'm worrying about, is everyone is now talking about how "real world performance" is what really matters, and sure as ell nVidia and ATI will start "optimizing" those aswell.

Sazar
06-15-03, 02:40 AM
Originally posted by sbp
There does seem to be an attitude amongst certain people the 5900 is nothing more than chopped liver.

the card is good... but it is not as good as the first previews were supposed to show it to be after the driver hacks are removed... :)

it is still an excellent buy and I don't recall seeing too many posts saying the nv35 is a bad buy...

nvidia drivers == evil... perhaps... nv35 == bad ? I dunno bout that one..

Behemoth
06-15-03, 03:27 AM
Originally posted by reever2
Oh come on, dont even attempt this excuse anymore, eveyrbody said that about the 5800 ultra and look at it now, the only "optimizing" i have seen in the Nv drivers have been for synthetic benchmarks
performance has been improving from driver to driver on nv30, 3DMark issue and nvidia attitude wont make the fact go away.

sbp
06-15-03, 04:14 AM
Originally posted by Sazar
the card is good... but it is not as good as the first previews were supposed to show it to be after the driver hacks are removed... :)

it is still an excellent buy and I don't recall seeing too many posts saying the nv35 is a bad buy...

nvidia drivers == evil... perhaps... nv35 == bad ? I dunno bout that one.. Quake 3 Arena, UT2003, Jedi Knight II, Serious Sam:SE, Commanche 4 benchies are not good due to driver hacks?

Eymar
06-15-03, 04:43 AM
I think the 5900 is a good card, but I also think it didn't meet everyone's (especially nVidia fans) expectations. Expectations that were probably too high because of the long wait (the PR department had to do something so people wouldn't buy ATI's stuff). However, expectations for nVidia card always high and usually higher than what is seen in the finished product. In the end though, I was always satisfied or more than satisfied with the finished product in spite of my high expectations. I think that will be the case of 5900. The only problem I see now is I played the waiting game so long that I'm thinking it's probably best to wait a little long for NV40 (hoping nVidia doesn't miss on the date too much).

StealthHawk
06-15-03, 05:19 AM
Originally posted by Behemoth
performance has been improving from driver to driver on nv30, 3DMark issue and nvidia attitude wont make the fact go away.

Yeah...timedemo benchmark scores have been improving. Has real world gaming performance also seen the same increase? Has real world gaming performance increased period?

I am now very weary about any huge increases that magically appear in new drivers as shown in timedemos that are used in popular benchmarks.

ChrisW
06-15-03, 05:19 AM
Where are these reviews bashing the 5900? All I ever see is reviewers praising it. They don't even mention the 'cheating' part. They are giving nVidia what they would never give ATI...a pass.

Ady
06-15-03, 05:36 AM
Originally posted by cwalker
Plus, the drivers have not been optimized yet.


:confused: ok.

I haven't seen one review bash the 5900. I've seen nothing but praise. Please point these reviews out.

btw I thought the problem was the drivers have been too "optimized"!?!?

Ady
06-15-03, 05:39 AM
Originally posted by sbp
Quake 3 Arena, UT2003, Jedi Knight II, Serious Sam:SE, Commanche 4 benchies are not good due to driver hacks?

EXACTLY!!! :afro:


*runs and hides*

CapsLock
06-15-03, 07:47 AM
Originally posted by ChrisW
Where are these reviews bashing the 5900? All I ever see is reviewers praising it. They don't even mention the 'cheating' part. They are giving nVidia what they would never give ATI...a pass.

I don't know about bashing exactly, but I did have a thread going on this topic before Dig and Hell got it locked. There were two reviews just recently done which didn't show the 35 in a very good light. The UKGamer review showed the 5900 being completely beaten in UT2K3.

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

There was also the 3DVelocity review which showed the same thing plus pretty bad performance in NOLF2 and F1 2002.

http://www.3dvelocity.com/reviews/5900u/5900u_9.htm

On the other hand, 3D's review showed the 5900 winning in UT without any aa or af and Morrowind. Of course they were only using the 9700P, which sort of makes the losing scores even worse.

PCmag had the 5900 lose in 5 of 7 benchmarks against the 9800.
The original gameguage in the ExtremeTech review was about the same.

Frankly I'm not upset at the reviewers, I'm PO'd at NV.
What is going on here? With more trans, more clockspeeds, the magical 256 bit bus and almost a year after the r300 you'd think they'd have been able to do better than this!!!

I was more upset when I originally read the two recent reviews, now I'm more dejected. The NV 40 better be good, I don't think any of the nv3x series are going to do it.

CapsLock

Ady
06-15-03, 07:54 AM
Originally posted by CapsLock
UKGamer review showed the 5900 being completely beaten in UT2K3.

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



The UKgamer review is of the 128mb 5900. Not the ultra.

CapsLock
06-15-03, 08:13 AM
ok, but how much of a perf difference is there between the two?

i don't think its very much if any.

Caps

surfhurleydude
06-15-03, 09:01 AM
ok, but how much of a perf difference is there between the two?

i don't think its very much if any.

Well, there's about 50 mhz RAM difference and a 50 mhz core difference between regular and ultra.

Dazz
06-15-03, 09:54 AM
FX5900 Ultra 450MHz / 850MHz + 256MB
FX5900 Regular 400MHz / 800MHz + 128MB

Not really much diffrence, although the biggest diffrence will be in fillrate intensive games where that 50MHz in the core will shine.

I myself think the NV35 range an't bad, would was expecting it to of been faster with the vast amounts of bandwith it now has. Me thinks the crossbar controller is still 128bit optimized & not 256bit.

Behemoth
06-15-03, 11:49 AM
Originally posted by StealthHawk
Yeah...timedemo benchmark scores have been improving. Has real world gaming performance also seen the same increase? Has real world gaming performance increased period?

I am now very weary about any huge increases that magically appear in new drivers as shown in timedemos that are used in popular benchmarks.
i have suspected nvidia might cheat on timedemos as well, so i ran some tests on ut2k3 using fraps for a few sessions, the results were pretty consistent with the timedemo results. AA/AF performance has improved dramatically either on timedemo or in-game(fraps).
but would they increase in-game performance as every timedemo indicates? i dont know, i dont want to take the time to "fraps" everything to find out only nothing, so i have to trust nvidia for this for now. Given so much hatred nvidia got nowadays, i have faith that they cant hide cheats anymore, at least not as easily as before, as people are now so keen on uncovering their cheats, and even IHV competitor is participating in doing so. :)

muzz
06-15-03, 03:14 PM
Originally posted by Behemoth
and even IHV competitor is participating in doing so. :)

What go's around comes around.......:D

I think they were just returning the favor.:p

digitalwanderer
06-15-03, 03:24 PM
Nope, not at all...but thanks for asking! :cool:

Sazar
06-15-03, 03:28 PM
Originally posted by digitalwanderer
Nope, not at all...but thanks for asking! :cool:

dood... you sound like big gay Al from south park..

I'm Soooper... thanks for asking

:D

:rofl

sry... just watched the repeat last night of the Harbucks episode :D and cripple fight...

2 classics..

digitalwanderer
06-15-03, 03:30 PM
Originally posted by Sazar
dood... you sound like big gay Al from south park..

I'm Soooper... thanks for asking

:D

:rofl

sry... just watched the repeat last night of the Harbucks episode :D and cripple fight...

2 classics..
Funny, I actually had his "I'm thuper!" song from the movie in me head when I was posting that! :lol: (Watched it again last night while waiting for Bubbles to finish up rebuilding herself, it ALWAYS makes me smile. :) )

ChrisW
06-15-03, 03:51 PM
Originally posted by CapsLock
I don't know about bashing exactly, but I did have a thread going on this topic before Dig and Hell got it locked. There were two reviews just recently done which didn't show the 35 in a very good light. The UKGamer review showed the 5900 being completely beaten in UT2K3.

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

There was also the 3DVelocity review which showed the same thing plus pretty bad performance in NOLF2 and F1 2002.

http://www.3dvelocity.com/reviews/5900u/5900u_9.htm

On the other hand, 3D's review showed the 5900 winning in UT without any aa or af and Morrowind. Of course they were only using the 9700P, which sort of makes the losing scores even worse.

PCmag had the 5900 lose in 5 of 7 benchmarks against the 9800.
The original gameguage in the ExtremeTech review was about the same.

Frankly I'm not upset at the reviewers, I'm PO'd at NV.
What is going on here? With more trans, more clockspeeds, the magical 256 bit bus and almost a year after the r300 you'd think they'd have been able to do better than this!!!

I was more upset when I originally read the two recent reviews, now I'm more dejected. The NV 40 better be good, I don't think any of the nv3x series are going to do it.

CapsLock
So, you feel the reviewer should only show benchmarks that nVidia wins in or they are 'bashing' nVidia or 'showing these cards in an unfavorable light'?

CapsLock
06-15-03, 06:15 PM
Originally posted by ChrisW
So, you feel the reviewer should only show benchmarks that nVidia wins in or they are 'bashing' nVidia or 'showing these cards in an unfavorable light'?

Where the heck are you getting that?

I think reviewers should do their job to the best of their abilities and as honestly as possible (niave, I know).

What I'm PO'd about is the total LACK of new super chip performance here. The one review has a 5900 regular (UKGamer)
which is clocked 400/400 runinng against a 9700P clocked 325/310.

In SS2, the 5900 wins at 4X 8X (but the NV 4Xaa does not equal Ati's as the review just pointed out).

BUT in UT2K3 (4X8X):

at 1024 the 5900 wins/loses at about evenly
at 1280 the 5900 loses badly
at 1600 the 5900 loses worse

Against an almost one year old chip thats clocked much lower!!!

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

In the 3DVelocity review its a 5900 ultra (450/425) going against a 9700P (325/310).

In Morrowind the 5900 wins but the 9700 scores seem a bit funny, they seem to be stuck low all the way through.

Then we get to NOLF2 scores, 1600 rez:

5900 ultra 9700 pro
73.7 2Xaf 94.0 2Xaf
70.1 4Xaf 87.6 4Xaf
64.7 8Xaf 84.1 8Xaf
61.1 16Xaf 83.1 16Xaf
56.2 2X 8X 67.3 2X 16Xaf
43.3 4X 8X 59.2 4X 16Xaf
33.7 8X 8X 48.5 6X 16Xaf

Please note this time I am PO'd at the reviewer. Why use 16Xaf for the 9700? But more importantly, here we have a popular game where the MUCH lower clocked 9700 smokes the 5900!!

So WTF? What is wrong with NV? They can't say whoops forgot the 256 bit this time. You'd think if anyone could afford the best engineers here, it would be NV.

In F1 2002, the 9700 wins again. And the dolt reviewer uses Ati 16Xaf against NV 8Xaf AGAIN!??

In UT2K3, the only way the 5900U wins cleanly is without any aa or af. As soon as they are turned on, the 9700P wins. Don't tell me its drivers, UT is a first to code for game and in the NV program, "the way its meant to be played". Its even worse if you consider NV's 8X aa is what you need to equal Ati's 4X aa.

http://www.3dvelocity.com/reviews/5900u/5900u_9.htm

So WHAT was NV smoking? Thats what I'm asking. Its absurd.

CapsLock