Macci shows off the 5900FX Ultra !! [Archive]

retrospooty

07-06-2003, 11:36 AM

Originally posted by Raptor
i believe i might change my R9800Pro that went to RMA for a 5900... naaaaa...

Don't do it man !

If you want the best benchmarking scores, the 5900 is a bit faster, but it is done by reducing image quality. Especially when using AA. Nvidia cards just don't look as good. :D

Raptor

07-06-2003, 11:41 AM

i just bought a R9500 128mb for "backup" and i simply can't leave ATI... they're so great to play Flight Simulator 2004! :D

macci

07-06-2003, 12:13 PM

This is one Extremely interesting card for benching I must say :D
Not quite sure how much higher it will go cos I was forced to stop benching because of condensation. I've ran the mem at 1GHz aircooled and it was artifactin a bit so I guess its maxing out at around 990 mem without extra cooling.
I think I killed my P4C800 and thats the reason for using the Albatron. Using a faster mobo/higher CPU speed will obviously give some extra marks but today I was mainly trying to find the max clock for the card. But even with this fairly low system speed I was seeing peak FPS of 1164FPS in GT1 :D

As far as Mark2001 goes it flies thru Nature at over 190FPS but the rest of the tests suck badly. Guess they didn't test MArk2001 with those drivers :P

OPPAINTER

07-06-2003, 12:23 PM

Originally posted by macci

As far as Mark2001 goes it flies thru Nature at over 190FPS but the rest of the tests suck badly. Guess they didn't test MArk2001 with those drivers :P

Macci,

Could the mobo be any of the problem with 2001?

OPP

eva2000

07-06-2003, 01:24 PM

here's to Albatron board! :D

Kunaak

07-06-2003, 02:01 PM

wow, been a long time since I seen NVidia anywhere near the top.

Congrats there.

:)

Jeff

07-06-2003, 02:08 PM

Another top-notch job by Macci!!! :toast:

QuadDamage

07-06-2003, 03:35 PM

nice score macci. btw, 44.03 are still fastest in '01.

Larsa

07-06-2003, 04:31 PM

Congrats macci :toast:

"Larsa"

charlie

07-06-2003, 05:00 PM

I'd still like to see the 3dmk01 details to analyze where the 5900 falls short.
C

Chong345

07-06-2003, 09:43 PM

Great score as usual Macci! :toast:
I'm fixing to start benching again. I hope all goes well.

Joey C.

GreenBeret

07-06-2003, 09:55 PM

I hope you guys know about all the 3DMark cheats Nvidia has put in their drivers ?

What is there to be excited about this score when the driver contains tons of cheats ?...and you scream bloody murder when other people cheat to get to the top of the ORB...

Chong345

07-06-2003, 10:01 PM

I don't think there are many people that haven't heard about the "cheats". Supposedly some of them are fixed and futuremark doesn't consider them cheats anymore. They are optimizations. I'm not saying what I think they are but regardless, I will benchmark against this score cause its fun to see a new record.

If we knew how to design a really nice driver set that abused 3dmark, would that be considered a cheat or tweaking?

Joey

retrospooty

07-06-2003, 10:28 PM

If a driver uses some creative method to get more performance its an optimization, If they lower the image quality, its a cheat. Plain and simple Nvidia drivers are cheats.

Just because the VP's of Nvidia and Futuremark got together and made a tidy little deal and suddenly futuremark retracted all statements does not change the fact that Nvidia lowers their image quality to get faster results...

Nvidia got where they are by being the fastest, and best image quality. Now they are deceptive about everything. :confused:

Creative

07-06-2003, 10:38 PM

Nice score Macci :)

But I think it is a sad sad time we live in if a 'known' bencher gets praised like the sun shines out of ass when he posts an 'optimized' score......yet when an unknown member like Anabolic posts a high score every cries foul and cheat etc etc.

Pretty hypocritical actions from the XS.org 3DMark team if you ask me.....and to be honest, the crap that has gone on in the last few days is really honestly making me lose interest in 3Dmarking all together...

Thanks guys :(

I am not saying I dont respect you guys, but I am saying that the behavious here in the last few days has not been something I would have expected from 'professionals' like yourselves....

Kunaak

07-06-2003, 10:52 PM

creative the issue is not as simple as that.

theres a difference between a faster driver, and unexplainable score. if it was a simple driver, then everyone would have figured it out by now.
theres still the issue that he really does appear to have just stubbornly ran away, after telling everyone to piss off, which in the end only makes it look alittle more fishy.

bretts a good guy, but I am just not sure I believe this cold boot thing he's claiming gets him a 2,000 point instant increase...

this whole issue is a double edge sword at the moment.

:(

Creative

07-06-2003, 10:56 PM

Kunaak, appreciate the reply. I was scared people would see it as me bagging Macci etc but nah *not at all* :)

I personally am skepticle about the 25K score too. Its just not humanly physically possible to get 360fps with a 3.9ghz cpu....I may not be up the top of the ORB (I dont have pc shops supplying me with hardware;) ) but I do know alot about 3dmark being Australias top 3dmarker (score wise).

Im not saying I know everything but I do know what numbers are needed for certain scorees, and I know what numbers can be achieved in certain tests :)

Back ontopic though, Maccis score is unbelievable, optimisations or not. :):)

Kunaak

07-06-2003, 11:05 PM

Hey Macci,

I just realized something, in early reviews the 5900 had a temp sensor on the videocard.
does your have one?
does it show the super cooling sub zero temps from your hardware?

thats one thing I am very interested in.

:)

SpicyHuevos

07-06-2003, 11:06 PM

Nice going macci :)
Oh boy looks like we all get to go shopping again:(
Thats ok personally im ready to switch back anyway.
Ive had these damn canadian goods in my box far to long!
If we knew how to design a really nice driver set that abused 3dmark, would that be considered a cheat or tweaking?
I use to make driver sets for nvidia all the time.
But i also made them available for anybody to download.
instead of just me using them.
Its just more fun playing on a semi level field if you no what i mean.

macci

07-06-2003, 11:43 PM

does it show the super cooling sub zero temps from your hardware?
Yup there is a temp sensor and it obviously can't read sub zero temps (R404 on GPU makes the control panel display a temp of 216C etc). The coldest I've seen was +23C and that was when I cooled it down with +15C water.

OPPAINTER

07-06-2003, 11:48 PM

Macci,

Was your bench done without v-mods?

OPP

ctgilles

07-07-2003, 02:29 AM

I am guessing NOT on the DDR, maybe on the GPU, but I'm doubting.
There are a lot better specimens out there, but I'd need to supercool it and it's not my card :rolleyes:
Macci, what OC software'd you use? Rivatuner/Coolbits goes back to stock settings at the start of the benchmark :(

ALT-F13

07-07-2003, 02:50 AM

oooh... again... Guys, ok, lets install all the anti-detectors, remove any software tweaking and we'll be the most honest 3DMark benching team ever.
ROFL...

JeffPH

07-07-2003, 03:16 AM

congrats MACCI !!!!

Bra!nFreeZe^

07-07-2003, 04:03 AM

I really cant understand why every1 starts §§§§§ing when some1 like macci or every1 else really put themselves together and end up on top of the orb ,now i dont want to start flaming or sumfin' like that but i still think its cool that macci got a score that nice!

btw im new in here ;)

ALT-F13

07-07-2003, 04:51 AM

Welcome to Xtreme, BrainFreeze!

Bra!nFreeZe^

07-07-2003, 08:26 AM

Thanks ALT weeee i got noticed :banana: ;)

Juggernout

07-07-2003, 08:28 AM

That's awesome !

You guys really now how to make a benchmark.

One question macci: Are u sponsored ?

Greetz

charlie

07-07-2003, 10:15 AM

Still like to see some 3dmark 01 action...sucky or not.
C

Shroomalistic

07-07-2003, 11:29 AM

i have a 9700pro and i am starting to get sick of it - i really love nvidia and i was wondering if i should go with the 5900 non ultra.
Its the same thing as the ultra just less memory, and they look better(MSI).

Hiwayman

07-07-2003, 11:44 AM

Congrats Macci:) AND your safely back from the Army too;)

Things are returning to normal, I think. I'm going shopping as soon as I get back from holiday.......

ctgilles

07-07-2003, 12:43 PM

Macci, care to let me in on your OC software? Mine resets when opening 3DMark :(

macci

07-07-2003, 12:56 PM

Coolbits does work just fine.. 3DMark03 just reads the 2D clocks.. not the 3D (which you are actually using when running 3D).
But I'm too lazy to open up the dang control panel.. so I use good old powerstrip :)

macci

07-07-2003, 01:03 PM

OPP, I've soldered a GPU vmod there but havn't increased the voltage above default yet ;)

charlie

07-07-2003, 01:33 PM

MACCI!
real curious about he '01 scores, I know you say they're not that great...but it would be interesting nonetheless.
C

zakelwe

07-07-2003, 02:18 PM

Congrats to Macci for his scores. The Orb is down at the moment but I guess he is #1 in 3dmark03 and not #1 in 2001 due to low P4 speed ( realatively of course) affecting car and lobby ?

I'm running a 5200U FX tonight and although the scores are 50% of Macci's I'm having a lot of fun with it. I find that where the GF4 liked about LOD +3.5-4.0 the FX likes LOD 6.5-8. The 44.03's are better still for 2001 than the 44.65 as someone said, though the 44.65's let you have build 330 at roughly the same score for 3dmark03.

My opinion is that with rivatuner and nvidia's myriad of "optimised" drivers you can have a lot more variety and fun than with Ati.

Drifting off the point here, well done again to macci.

Regards

Andy

ctgilles

07-07-2003, 02:44 PM

Originally posted by charlie
MACCI!
real curious about he '01 scores, I know you say they're not that great...but it would be interesting nonetheless.
C

Ehm, my non ultra hits 20k6 with stock clocks! and 4080MHz P4 power ;) I don't call that 'bad'

so I use good old powerstrip
Ah, you only use new 3D clocks and leave the 2D clocks?

Severian

07-07-2003, 03:35 PM

GeforceFX 5900 (Ultra)s are not supposedly as good at 3DMark 2K1 as 2K3 simply because of the way they are designed.

In 2K1, only 1 of the 4 tests (Nature being the exception) are multi-textured, compared with 2K3 in which the cut is closer to half-half (or possibly even more multi-textured than not? I'm not sure on the details, haven't used 2K3 that much).

It's a noted fact that while ATi specialises in the Anti-Aliasing front, the GeforceFX is especially good as multi-texturing (being a 4x2 architecture).

For that reason, I doubt that the 2K1 scores, with the exception of nature, will ever be as good as a Radeon 9800 Pro, while in 2K3, regardless of drivers, it probably still beats the pants off it.

As for complaining about "mystic tweaks" and "illegal driver optomizations", except for hacking the Futuremark database, they're all valid entries that can supposedly be repeated by anyone. So in that sense, it really is a level playing field, so how can anyone complain?

Short of being government funded to run an industrial scale liquid nitrogen setup, I can't really see how anyone can have an unfair advantage. Advantage yes, unfair no.

retrospooty

07-07-2003, 04:04 PM

Originally posted by Severian

As for complaining about "mystic tweaks" and "illegal driver optomizations", except for hacking the Futuremark database, they're all valid entries that can supposedly be repeated by anyone. So in that sense, it really is a level playing field, so how can anyone complain?

I myself was not "complaining" nor do I think its illegal (perhaps it IS misleading to the unaware buyer) The point I was trying to get across, is just that if the improved scores come at a loss of visual quality the card itself is not as good as it appears. Therefore buyer beware.

As far as Macci's scores... well they speak for themselves, his scores were impressive before, now, and I am sure the next gen of hardware goodies that come out will also fall victim to his overclocking Jedi master skills :slobber: (and I mean that in a very complimentary way :D ) Seriosly, nicely done man !

GreenBeret

07-07-2003, 09:16 PM

The point I was trying to get across, is just that if the improved scores come at a loss of visual quality the card itself is not as good as it appears.

It's not just a loss of visual quality. The inserting static clipped planes and not clearing the back buffer cheats make it almost like running pre-rendered scenes...

macci

07-08-2003, 01:01 AM

Originally posted by charlie
MACCI!
real curious about he '01 scores, I know you say they're not that great...but it would be interesting nonetheless.
C
Just ran it with P4C800 and 2.4C@3608MHz (300FSB, 5:4 mem, 2-2-2-5) card at 555/951 aircooled and got this:

22825 (http://service.futuremark.com/compare?2k1=6732616)

Not that bad actually ;)

Marci

07-08-2003, 03:37 AM

Back on top again eh d00d?! Seems like every time I look at the Orb these days it's swapped from you to Opp to you to Opp...

So when's the race for 10k starting then? I wanna get a track-side ticket... (bench-side?)!

macci

07-08-2003, 07:04 AM

does it show the super cooling sub zero temps from your hardware?
ignore what I said earlier..

The temp display goes down to 0C and after that it starts from 255C (so 255C = -1C). With my R404 cooler I've got it down to 213C which equals to -43C (255 - 213 = 42, -1 - 42 => -43).

btw 630 core is not even close to max for this card ;)

caater

07-08-2003, 07:43 AM

Just wondering.. How much will You get with older build of 3dmark03 ? I know that orb shows newer build by default.. and even stops (or stopped?) accepting old scores..
but just, out of curiosity :)
Maybe 10k ? :)

macci

07-08-2003, 07:47 AM

I think that badboy from FM forums ran both 320 and 330 build with his 5900Ultra + 44.65 drivers and the difference was like 50Marks

Marci

07-08-2003, 07:50 AM

personally I'd love to see a run on 3DMark 99 MAX... but it just ain't gonna happen :(

charlie

07-08-2003, 10:32 AM

Originally posted by macci
Just ran it with P4C800 and 2.4C@3608MHz (300FSB, 5:4 mem, 2-2-2-5) card at 555/951 aircooled and got this:

22825 (http://service.futuremark.com/compare?2k1=6732616)

Not that bad actually ;)
At 3600mhz, that's not bad at all. The drivers obviously aren't optimized for a 3dmk01 type application. I'm sure someone will come out wiht a stronger driver set for these "older" applications.
C

QuadDamage

07-08-2003, 05:04 PM

i ran 3dmark 2k with msi non-ultra at 490/990 and got around 10k with 8xS AA and 8xAF / MAX Quality :D

ctgilles

07-08-2003, 11:59 PM

Originally posted by QuadDamage
i ran 3dmark 2k with msi non-ultra at 490/990 and got around 10k with 8xS AA and 8xAF / MAX Quality :D
Hehe, with anti detector patches I got 5k2 w/ 4080MHz and 475/475 (or so. ;))

SpicyHuevos

07-09-2003, 02:29 AM

Dj <------waves to macci:)
orb check sir;)

charlie

07-09-2003, 02:39 AM

DJ
I don't see it...
What did ya' get?
C

charlie

07-09-2003, 02:43 AM

Ahhhh...
There it is...
http://service.futuremark.com/compare?2k1=6737624
3DMark 2001, **24,256** @ 4373mHz
FX5900 ULTRA @ 549/964
Sweeeeet, DJ. Has it got 25K in it?
C
edit: DJ, What cpu is that, the 266fsb/4373 thing has me confused...

zakelwe

07-09-2003, 07:01 AM

Nice work DJ Spicy . :toast:

I wonder if macci doing a 600+ gpu speed will compensate for the 700Mhz he is down on DJ's rather speedy p4

I hope some of the extreme big guns can get the $$ for a 5900 ultra soon, too many vr-zone lot in the 5900 class top 2 :p

Regards

Andy

ctgilles

07-09-2003, 07:25 AM

In '03 yes, not in '01 :)
I'm getting one IF I can trade the RMA with my 9800 PRO

SpicyHuevos

07-09-2003, 08:52 AM

Glycol/alcohol@ -40f and a homemade ghetto waterchiller+ 18,500btu at 45mph.
If you leave the side of the case on it will blow it over;)

Btw worth mentioning!
I do have a Prometheia i have had now for about 6months.
It worked one month out of that 6. And im still waiting for chip con to to do something about it it's been over 5 months since i started the rma procedure. 2 shipping companies they had called to come get this for warranty freaked out when they learned freon gas was inside and declined transport.

The problem is with it i plug it in and turn it on Kapow house 30 amp breaker blows. replaced breaker same thing took to neighbors house same thing it blew his too.
Bascially the compressor is locked and wont kick over.
oh and i never so much as opened the side panel. So needless to say its stock.
p.s thanks Michal for that speedy customer support at chip-con
By the time they take care of it Mach 5 will be out;)

ctgilles

07-09-2003, 09:21 AM

Originally posted by MickeyMouse
what the hell is DJ cooling that P4 with at almost 4400mhz???

You have seen nothing yet :D
At Duracell a bit further from me they need to cool down, they make batteries :) well, those compressors/tubes/vents are as big as my room, I'll go take a piccie tomorrow :)

macci

07-10-2003, 12:31 PM

FX5900 supercooling pics (click pics for better view):

card and some R404 mounting gear:
http://www.solidhardware.com/macci/5900ultra/01_fx_mount_insu_s.jpg (http://www.solidhardware.com/macci/5900ultra/02_fx_mount_insu.jpg)

GPU and memchip details:
http://www.solidhardware.com/macci/5900ultra/04_fx_gpu_2_s.jpg (http://www.solidhardware.com/macci/5900ultra/05fx_gpu_2.jpg)
http://www.solidhardware.com/macci/5900ultra/03_fx_memchip.jpg

Cooling installed:
frontside:
http://www.solidhardware.com/macci/5900ultra/06fx_front_s.jpg (http://www.solidhardware.com/macci/5900ultra/07fx_front.jpg)
backside:
http://www.solidhardware.com/macci/5900ultra/09fx_backside_s.jpg (http://www.solidhardware.com/macci/5900ultra/10fx_backside.jpg)

Full system:
http://www.solidhardware.com/macci/5900ultra/11fullsystem_s.jpg (http://www.solidhardware.com/macci/5900ultra/12fullsystem.jpg)

http://www.solidhardware.com/macci/5900ultra/13fullsystem2_s.jpg (http://www.solidhardware.com/macci/5900ultra/14fullsystem2.jpg)

http://www.solidhardware.com/macci/5900ultra/15closeup_1_s.jpg (http://www.solidhardware.com/macci/5900ultra/16closeup_1.jpg)

Amoeba Assassin

07-10-2003, 01:20 PM

just curious, Dj, do you get the max score possible in PCMark02? Also, do you have a weak 9800 or something? You should #1:D

Anyways, very nice Macci, that is incredible & I like seeing Nvidia back at the top.

*Tips Hat*

SpicyHuevos

07-10-2003, 08:11 PM

Dj, do you get the max score possible in PCMark02? Also, do you have a weak 9800 or something
HAvnt tried pcmark yet with it.
Well no i dont have a weak 9800 i have 2 of them:)
Both do 538-543core mem 414-416.
Dont forget that with opps radeon speeds he could pretty much run it at 3500mhz and still be #1 in ati category because 2003 is
94% card imo.
I could clock it up to 4900mhz and i still wouldnt beat him at 540/411.:slobber:
Take this into consideration if you will
Im the only person in the top 10 using watercooling for cpu and a 80watt peltier on my video card. Instead of a vapochill or prommie;)
:smileysex

Amoeba Assassin

07-11-2003, 06:47 AM

Well that makes sense for 03, but what about 01. Can't you squeeze out another 6 pts, & take #1? Damn 18,500 btu??? 20k is just about having every appliance in your house run at once. What mem do you use, hyper x pc3500?

saaya

07-12-2003, 12:01 PM

macci whats your 2k1 score with that 5900?

jaawood

07-12-2003, 01:43 PM

macci

07-14-2003, 10:29 AM

MACCI is it just me or is that phasechage setup bigger than a vapochill unit/compressor/condensor and so on??
Yes its a bit more powerful than a Vapochill - it has danfoss 13cm3 R404 compressor ;)

btw updated score in the sig :)

Chong345

07-14-2003, 10:39 AM

Very nice scores macci. :toast:

Have you used any extra voltages on the GPU? Also how is the mod done?

Joey

macci

07-14-2003, 11:43 PM

Ok here is the GPU mod piccy for ya :)

http://www.solidhardware.com/macci/fx5900ultra/ultra5900_gpu.jpg
The VR pin with the wire is not makin contact with that powerconnector solderpad. The other VR pin is grounded to the powerconnector GND.
Measure voltage from the cap marked with blue.

SpicyHuevos

07-15-2003, 05:58 AM

Oh macci you caved!:eek:

ctgilles

07-15-2003, 06:35 AM

I was like whining for that pic a week :D
Thanks Macci :)

Chong345

07-15-2003, 09:08 AM

Thanks macci. :D

Joey

Chong345

07-15-2003, 02:51 PM

Question for you macci. What was the stock reading for the GPU voltage? Mine reads 1.345V and earlier was reading lower than that.

I just found a point that measures 2.71V (memory). Any ideas on that mod?

Joey

macci

07-15-2003, 10:22 PM

Mine was like 1.2V for 2D and 1.45V for 3D.

You can increase the 2.71V by adding a 20k VR to pin 5 of 6012 chip.

ALT-F13

07-17-2003, 03:08 AM

Macci
http://www.overclockers.ru/news/newsitem.shtml?category=1&id=1058417307 - news on russian overclocking site about your FX5900 (with pics from ur site).
And one question - why don't you cool memory? Maybe heatsinks could make some difference? Or I just don't understand something?:)

macci

07-20-2003, 10:49 PM

Originally posted by Niko
It's not just a loss of visual quality. The inserting static clipped planes and not clearing the back buffer cheats make it almost like running pre-rendered scenes...
Yes thats something they used to do with 44.03 drivers if I remember correctly. With the 44.67s things seem to be a bit different thou. Here's some info I found from Guru3D.com. Its about nVidia's drivers (http://content.guru3d.com/news.php?cat=1&perpage=40#194)

"After we tested the two drivers, we fired up the developer version of 3DMark03, ran Game Test 4 – Mother Nature, and enabled the free-camera mode. We wanted to see whether the same corner-cutting we observed with 44.03 was still present in this new driver. It wasn't. As far as we can tell, nVidia achieved these impressive performance gains without any "on-rail" optimizations."

ALT-F13, It might help yes but at the moment I think that the memory stays colder without em..the R404 cools down the PCB so well that the mem chips stay quite cool too.