PDA

View Full Version : nV News 5900(SE) vs. 5600U Upcoming Review... UPDATE:


Pages : [1] 2

ragejg
10-17-03, 08:23 AM
FRESHLY EDITED 11/14:

The preliminary plans were great and all, but this review took a different path due to time constraints and the fact that I suffered some equipment failure (dead 8RDA+) ...

FYI folks, here's the brief skinny on the review:

Cards:

Creative Labs Europe 5900 Non-Ultra (400/700, primary focus of review)
vs.
XFX 5600 Ultra (400/800)

Testbeds:

Soltek nforce2 mobo
XP1700 @200/400fsb, 2.2ghz
2 sticks samsung PC2700 256mb @ dual channel, 200/400mhz

Abit KR7A-133R kt266a mobo
XP1700 @ 166333fsb, 1.89ghz
1 stick Samsung PC2100 256mb @ 166/333mhz

Tests Run:

(we had no 16x12 capable monitors, btw...)

SpecViewPerf 7.1
ChameleonMark
Dawn Ultra
3Dmark2k1
3Dmark2k3
Gunmetal gameplay (1st level)
Halo timedemo
halo gameplay (Battle Creek)
TMNT demo gameplay (1st level)
MOHAA gameplay (Normandy, God Mode)
SSSE (Elephant Atrium Demo)

Overclocking:

- Covered on the 5900... there will be a section at the end with OC results for many of the tests run...

AA/AF Comparisons:

Most tests were run with no aa/af, then just 8xaf, then 4xaa/4xaf... a few were also run @ 4xaa/4xaf ... and also a few were run @ 6xaa/8xaf and 8xaa/8xaf.

Image quality:

There will be a couple screenshots of 4xaa/8xaf vs. 8xaa/8xaf from SSSE, and a couple 4x/6x/8xAA with 8XAF shots from TMNT...

I will be providing links to StealthHawks IQ threads, for they are chock full of informative info, and completely usable in this review.


...

I don't need to mention the filler of the review... it'll be there (Intro, pics, conclusion)...

...

This is my first review, folks. I did my best. Hope you like it when it's posted!! :).

ragejg
10-17-03, 12:22 PM
Oh, I've done that good, huh? So perfect that no-one can comment? :p:p

jeez, maxpower got BUTTLOADS of community response on his...

c'mon, folks! I need some feedback!! :D:D

darkmiasma
10-17-03, 12:40 PM
ragejg -

no offense, but the only point of interest would be point #3 (I have a lower spec vidcard. I love it and am keeping it, but how does it fare vs. newer competition)...

other than that, we can see the other points from any other review out there.

since you arent really focusing on IQ (which is somewhat subjective anyway) i really dont see the point in even bothering with this test.

i applaud you for wanting to do a review (i would love to do one, but i frankly dont have the patience) but i really can't see what will make this one any different than any other one on the web.

additionally, i think its a poor idea to test the DX8 cards while they are overclocked, simply because some people don't overclock their hardware, and overclocking results vary from person to person anyway.

- mike

ragejg
10-17-03, 01:27 PM
Originally posted by darkmiasma
ragejg -

no offense, but the only point of interest would be point #3 (I have a lower spec vidcard. I love it and am keeping it, but how does it fare vs. newer competition)...

other than that, we can see the other points from any other review out there.

since you arent really focusing on IQ (which is somewhat subjective anyway) i really dont see the point in even bothering with this test.

i applaud you for wanting to do a review (i would love to do one, but i frankly dont have the patience) but i really can't see what will make this one any different than any other one on the web.

additionally, i think its a poor idea to test the DX8 cards while they are overclocked, simply because some people don't overclock their hardware, and overclocking results vary from person to person anyway.

- mike

That's good that you at least have one point of interest in this...

But maybe you're forgetting that there's a plenty of people wanting to:

A. Know how well the 5600U and 5900 perform with the newer drivers (and possibly DX9.1).

B. Know how these cards (more-so the 5600U) perform as a step up from their previous hardware.

...

Most people that frequent hardware sites overclock their cards. And though OC'ing varies person to person, you can safely say that "most GF4ti4200 owners can get 280-310 core speeds" ... as an example...

And also: some older cards when OC'ed perform far better than the generalized stats given upon introduction and the first wave of reviews, which alwayys leave an impression on card buyers... What if you want to play Halo and only have $50 and you discovered that a $50 R8500 fit your needs perfectly because of it's PS 1.4 support and higher FPS in Halo than a GF3 or maybe a GF4 for instance?

Look, people have been begging for scaling info. I'm giving it to them. People wonder how *now*'s midrange cards fare vs. what they currently have. I'm giving them some facts on that.

I suppose this would've recieved MORE attention if I'd have said right off the bat "testing and benchmarking 52.16 vs. 52.13" even though I don't have the driver yet? ...

Thanks for the input though.

Clay
10-17-03, 01:33 PM
Originally posted by ragejg
Oh, I've done that good, huh? So perfect that no-one can comment? :p:p

jeez, maxpower got BUTTLOADS of community response on his...

c'mon, folks! I need some feedback!! :D:D
Whoa, must be a slow day or something. I bet you'll get more feedback this evening. Maybe you presented so much info that people may not know exactly how to tackle a response. It looks like you have it well thought out. I can see this taking you a TON of time though, might not hurt to scale it back a bit for your first review...they take a lot more time than you might think. ;)

Looks great overall! One thing I might suggest is to add another game and maybe drop the GF3ti500 and R8500...I see where you're coming from including those but it will really lengthen your timeframe. I'd stick with the 3.8 Cats and definitely wait for the 52.16's if you can. Unless you just really want to do a full blown driver comparison I'd just stick to one version for ATI and NV each.

Good luck!

darkmiasma
10-17-03, 02:15 PM
That's good that you at least have one point of interest in this...

I would have read the article regardless, but I am interested in your conclusions regarding these cards.

But maybe you're forgetting that there's a plenty of people wanting to: A. Know how well the 5600U and 5900 perform with the newer drivers (and possibly DX9.1).

I think people would want to know about what changes went into these drivers to make them perform however they perform (including the IQ)

B. Know how these cards (more-so the 5600U) perform as a step up from their previous hardware.

I agree. But I would like to see a 9600 in there too

Most people that frequent hardware sites overclock their cards. And though OC'ing varies person to person, you can safely say that "most GF4ti4200 owners can get 280-310 core speeds" ... as an example...

But you should show the cards @ stock and overclocked then ... not just overclocked is my point.; especially if you are trying to show how worthy of a step-up the card will be.

And also: some older cards when OC'ed perform far better than the generalized stats given upon introduction and the first wave of reviews, which alwayys leave an impression on card buyers... What if you want to play Halo and only have $50 and you discovered that a $50 R8500 fit your needs perfectly because of it's PS 1.4 support and higher FPS in Halo than a GF3 or maybe a GF4 for instance?

I agree, which is why I think you should show stock vs. overclock to show the performance you might gain.

Look, people have been begging for scaling info. I'm giving it to them. People wonder how *now*'s midrange cards fare vs. what they currently have. I'm giving them some facts on that.

Thats cool with me. I have friends who would be interested in this, but they are also considering a 9600.

I suppose this would've recieved MORE attention if I'd have said right off the bat "testing and benchmarking 52.16 vs. 52.13" even though I don't have the driver yet? ...

Not really, but it probably wouldn't hurt to mention it.

Thanks for the input though.

No problem. I hope it didn't come off offensive or harsh ... now that you explained it more, it seems better. I wish you could throw the 9600 in though.

- Mike

ragejg
10-17-03, 02:22 PM
Sadly enough, I have no access to a 9600P for this review, which is my 1st... So I figured I'd fill it out with as much other info as possible.

... Oh wait, I do have an RV370 here, but there's an NDA... plus there's some new Cats sittin here with me... lemme call ATI up and see if they'll let me use this...







...









j/k! :p

darkmiasma
10-17-03, 04:04 PM
Originally posted by ragejg
Sadly enough, I have no access to a 9600P for this review, which is my 1st... So I figured I'd fill it out with as much other info as possible.

I think that is unfortunate, especially since you seem to be trying to do a mid-range upgrading article, and that is an excellent mid-range card.

Good luck with the review.

- Mike

Rytr
10-18-03, 05:35 AM
Is it worth the extra money to go with a 5900 over the 5600U? This was of interest for those trying to decide between Ati's 9700 and the 9500P and likewise the 9800 and the 9600P. What are the benefits of one card over the other and of course how will the newest drivers enhance those.

Use of additional drivers and additional card comparisons may take away from what the actual tests reveal (muddy the water, so to speak).

Some disagree but I see significant differences between the Ti series and the FX series and anything close to apples to apples comparison is really difficult to achieve other than speed tests.

Good luck with the review!

ragejg
10-18-03, 10:21 AM
FYI:

My helper (goofjb)and I are mentally picking this apart with a fine-tooth comb...

We MIGHT do away with come of the cards in the test, but probably not the ti4600...

Maybe a wider comparison will have to wait. We really want to include R200 and NV20 in a review :D:DONE MORE TIME:D:D... they're still powerful enough in some ways to deserve some recognition and attention... if a card can do XIII @ 10x7, all IQ at max, with 4XAA and 4XAF and remain very smooth, it still deserves some attention. ;)

Rytr
10-18-03, 10:28 AM
I totally agree with that ragejg! Just avoid using the CAT 3.8's with the 8500. Instead the released drivers on the Dell site or Omega's modified 3.8's. Either of these will correct most of the conflicts between the 8500 and the 3.8 drivers.

I think the 8500 deserves another review and the CATs thru 3.7 have really improved overall performance.

ragejg
10-18-03, 10:38 AM
Ok, I must have missed something... what's the issue with R200 and cat 3.8's??


ALSO: I need your help, folks. This Creative 5900 has ram that's supposedly only rated to 700mhz. I can't seem to find the part number or any information on the web to confirm...

"SAMSUNG 316
K4D263238-6C2A"

Without any ramsinks, it so far appears that these run out of gas at around 780-790... UPDATE: autodetect just allowed me 474/850... I'll run some tests when I get back from the mall and see if it can stand it...

If anyone can help with some info on this BGA chip, it would be appreciated...

Rytr
10-18-03, 04:03 PM
It rated at 2.86ns, 350MHz.

ragejg
10-18-03, 08:36 PM
meh...
most 5900 non-U's I've seen have faster memory than straight 700...

That's ok, the show will go on... :)

Lfctony
10-19-03, 06:30 AM
Isn't there a code to distinguish between versions of 5900s? Like Albatron's ones, the 5900CP runs at 400/700, the 5900EPV runs at 405/810 etc.

ragejg
10-22-03, 02:36 PM
http://www.nordichardware.se/recensioner/grafikkort/2003/FX-Mani/Creative/front_thumb.jpg

Big HSF/no ramsinks... (hsf would get in the way of ramsink installation)

That's some heavy copper...

It's quiet though.

Stability cuts out @ anywhere above 450/805 at this point... not too bad...

ragejg
10-22-03, 02:43 PM
is this board compliant with the reference design?

ragejg
10-24-03, 07:43 AM
FYI peeps:

Review has been pared down (dang it!!) to:

Creative 5900 vs. XFX 5600U vs. Chaintech ti4600.

My nforce2 mobo was DOA, so I've had to realistically re-organize...


... so, Maxpower, what Humus demos do you think I should run?

Clay
10-24-03, 08:59 AM
Originally posted by ragejg
FYI peeps:

Review has been pared down (dang it!!) to:

Creative 5900 vs. XFX 5600U vs. Chaintech ti4600.

My nforce2 mobo was DOA, so I've had to realistically re-organize...


... so, Maxpower, what Humus demos do you think I should run?
Well Humus stated that many of the OpenGL demos use PS1.1 so that explains the decent performance by the NV36 whereas the D3D demos he has use PS2.0 moreso which results in lower scores by the NV36. I just picked some of the ones that looked more stressful than others. It might be neat for you to use the sames ones I did, you could always throw in some others as he has quite a few.

Morrow
11-02-03, 09:05 AM
Originally posted by ragejg
1. No AA/no AF, all settings @ highest performance (10x7 and 12x10)
2. Highest Quality settings, 8XAF only (10x7 and 12x10)
3. Highest Quality settings, 4XAA only (10x7 and 12x10)
4. Highest Quality settings, 4XAA/8XAF (10x7 and 12x10)
5. nv cards, same as #4, but add texture sharpening…


point 5 makes no difference because texture sharpening does nothing at 8xAF

ragejg
11-02-03, 10:22 AM
Well, texture sharpening's not in these new drivers anyway, so, yeah, #5 is out...

:p

Morrow
11-02-03, 12:04 PM
Originally posted by ragejg
Well, texture sharpening's not in these new drivers anyway, so, yeah, #5 is out...

:p same thing for drivers prior to 5x.xx.

Texture sharpening does just increase the selected AF level by one step.
Since 8xAF is already the maximum texture sharpening cannot increase anyhow further and therefore does nothing

ragejg
11-14-03, 06:16 PM
1st post on thread edited...

... It's almost done!!!

http://www.nvnews.net/vbulletin/attachment.php?s=&postid=232458

volt
11-14-03, 06:29 PM
Why so many non-game benchmarks?

ChrisRay
11-14-03, 06:41 PM
Well since I have a Geforce FX 5900SE From leadtek (400/700) Quite curious to see this review.