bkswaney's GeForce FX 5800 Ultra report

[5150 Joker]

Originally posted by ugapug
I don't know about your system, but with mine Unreal II runs like butter. 2xAA 4xAF 1280X1024 all settings maxed.
Specs:
Abit NF7-S rev 1.0
T-Bred 1700@190x11 (2112)
TwinX512 3200LL
PNY GeforceFX 5800 Ultra
WD Raptor SATA
WD 1000BB 100gig (w/2meg cache)

Upgraded from a Geforce4 Ti4600. Had an ATI 9700 Pro to compare to the Ultra. IQ seems quite similiar to me, but the colors on my Ultra with Digital Vibrance activated seem a bit more vivid.
Just my .02
And I like loud fans. It gives me an excuse to crank up the speakers :D

The game runs fine on my system with the same settings you listed but anything below 60 fps during a heavy action scene isn't acceptable to me. People have different standards.

[Rogozhin]

BKswaney

Would you mind doing a standard benchmark using the one in ut2003?

That way we can validate your claim. Just do it 1024x768 all options MAXED PLUS 2x AF and 2x AA, and use application on the AF, and I will use quality on my 9700, do flyby not botmatch.

Rogo

[Rogozhin]

Remember though that the highest tap trilinear AF a nv30 can do is 64 but on the 9700 and 9800 its 128 tap.

Rogo

[GlowStick]

3DMark Score: 5680 P4@2.66, R9800Pro@410/350, Performance Settings

Date: 2003-4-6
Res: 1024x768 32 bit
OS: Microsoft Windows XP
CPU: Intel Pentium 4 2660 MHz
GPU: ATI RADEON 9800 Series
411 MHz / 351 MHz
User: dumpspamhere@yahoo.com



3DMark Score: 5590 Northwood 2.53Ghz and Radeon 9800 Pro (both not ov...

Date: 2003-4-22
Res: 1024x768 32 bit
OS: Microsoft Windows XP
CPU: Intel Pentium 4 2533 MHz
GPU: ATI RADEON 9800 Series
378 MHz / 337 MHz
User: mmahtani@netvigator.com



3DMark Score: 5581 Radeon 9800 Pro OverClocked More

Date: 2003-4-17
Res: 1024x768 32 bit
OS: Microsoft Windows XP
CPU: Intel Pentium 4 2533 MHz
GPU: ATI RADEON 9800 Series
398 MHz / 357 MHz
User: glowstick@insightbb.com



3DMark Score: 5545 420 Inc.

Date: 2003-4-24
Res: 1024x768 32 bit
OS: Microsoft Windows XP
CPU: Intel Pentium 4 2591 MHz
GPU: ATI RADEON 9800 Series
398 MHz / 351 MHz
User: killerinstinct@adelphia.net

Hm so just everyone else in 3dmark03 is getting the same scores as me.

Rogozhin, your clocks were ameingly not 300/300, infact, they were stock wow : P

5150 joker, so you have no hidden secret of builing a system wow!

As haveing an intel motherboard my ram timeings are set spareingly for stability. I also use an Intel system because i needed Linux stability, the Athlons have a small flaw where the Extended page fileing dose not work (like an intel). It was documented at http://www.gentoo.org and the only fix they could find is to pass the option to the kernel to use the old page system.

You clearly want to say "OH i wont say whats wrong with your system because, oh im to lazy" Sounds odd, wait, you just dont know. There is nothing wrong with my pure Intel system.

The results are infront of my eyes. I have a Radeon 9800Pro, i see the results. WOW!

[Rogozhin]

AD HOC

sorry ;(

rogo

[5150 Joker]

Originally posted by GlowStick
3DMark Score: 5680 P4@2.66, R9800Pro@410/350, Performance Settings

Date: 2003-4-6
Res: 1024x768 32 bit
OS: Microsoft Windows XP
CPU: Intel Pentium 4 2660 MHz
GPU: ATI RADEON 9800 Series
411 MHz / 351 MHz
User: dumpspamhere@yahoo.com



3DMark Score: 5590 Northwood 2.53Ghz and Radeon 9800 Pro (both not ov...

Date: 2003-4-22
Res: 1024x768 32 bit
OS: Microsoft Windows XP
CPU: Intel Pentium 4 2533 MHz
GPU: ATI RADEON 9800 Series
378 MHz / 337 MHz
User: mmahtani@netvigator.com



3DMark Score: 5581 Radeon 9800 Pro OverClocked More

Date: 2003-4-17
Res: 1024x768 32 bit
OS: Microsoft Windows XP
CPU: Intel Pentium 4 2533 MHz
GPU: ATI RADEON 9800 Series
398 MHz / 357 MHz
User: glowstick@insightbb.com



3DMark Score: 5545 420 Inc.

Date: 2003-4-24
Res: 1024x768 32 bit
OS: Microsoft Windows XP
CPU: Intel Pentium 4 2591 MHz
GPU: ATI RADEON 9800 Series
398 MHz / 351 MHz
User: killerinstinct@adelphia.net

Hm so just everyone else in 3dmark03 is getting the same scores as me.

Rogozhin, your clocks were ameingly not 300/300, infact, they were stock wow : P

5150 joker, so you have no hidden secret of builing a system wow!

As haveing an intel motherboard my ram timeings are set spareingly for stability. I also use an Intel system because i needed Linux stability, the Athlons have a small flaw where the Extended page fileing dose not work (like an intel). It was documented at http://www.gentoo.org and the only fix they could find is to pass the option to the kernel to use the old page system.

You clearly want to say "OH i wont say whats wrong with your system because, oh im to lazy" Sounds odd, wait, you just dont know. There is nothing wrong with my pure Intel system.

The results are infront of my eyes. I have a Radeon 9800Pro, i see the results. WOW!


Once again you proved nothing. The fact remains I posted a score at 301/301 vs yours at a similar clock rate and got more than DOUBLE what you got. It should also be taken into consideration that you are using a R9800 which has the F-buffer and other optimizations my R9700 lacks yet you still get beat handily. Are you really that stupid?

[Rogozhin]

Glowstick

You were claiming that your radeon 9800 was getting 1550 marks, so what the heck is your problem? You are refuting your earlier post.

Those are decent scores for your system.

I never said that my 9700 non pro was stock. So the frequencies are still valid.

What's your proposition then?

Jeez

rogo

[5150 Joker]

Originally posted by Rogozhin
Glowstick

You were claiming that your radeon 9800 was getting 1550 marks, so what the heck is your problem? You are refuting your earlier post.

Those are decent scores for your system.

I never said that my 9700 non pro was stock. So the frequencies are still valid.

What's your proposition then?

Jeez

rogo


He was trying to show that nagus's claim of the R9700 beating the NV35 at 300/300 was a lie by setting his 9800's clock to 300/300 and getting a much lower 3DMark03 score (~1500). Turns out I did the same thing and yielded a MUCH higher score (>4.5k 3DMarks) than Glowstick which invariably proves he is an idiot.

[Rogozhin]

Glowstick

Rallisport challenge is heavily dependant on sound card if you use doppler effect. And there is a known bug with the lens flare option.

With lens flare turned off and all options on highest except for opponent car sounds, doppler effect, and weather, I am getting average fps of 50 and highs of 95.

You're doing something wrong in settings. If you need a walkthough then I can give you some pointers.

Also

You're either a fool or an idiot for posting your 9800 at 1550, I don't know which but if you're a fool it's better to keep your trap shut and not be validated than open it and remove all doubt.

rogo

[Rogozhin]

5150

thanks for the clarification.

i will go do the same just a sec.

Rogo

[GlowStick]

Originally posted by Rogozhin
Glowstick

You were claiming that your radeon 9800 was getting 1550 marks, so what the heck is your problem? You are refuting your earlier post.

Those are decent scores for your system.

I never said that my 9700 non pro was stock. So the frequencies are still valid.

What's your proposition then?

Jeez

rogo

Yes, we are testing cards underclocked at 300/300.

So you agree that my score of 5500's is "decent" i could probly get a few hundred more if i had agp8x and VERY aggressive ram timeings. But the 845PE chipset dose not support AGP8x and my ram timeings are fine as they are.

SO baiscly you are saying

My system is fine at its normal clock speeds. But when its underclocked my system is 'built wrong'.

Thx, just proved my point =]

[5150 Joker]

Originally posted by GlowStick
Yes, we are testing cards underclocked at 300/300.

So you agree that my score of 5500's is "decent" i could probly get a few hundred more if i had agp8x and VERY aggressive ram timeings. But the 845PE chipset dose not support AGP8x and my ram timeings are fine as they are.

SO baiscly you are saying

My system is fine at its normal clock speeds. But when its underclocked my system is 'built wrong'.

Thx, just proved my point =]

He thought your system was FUBAR due to the 1500 score w/out knowledge of you having underclocked the card. I maintain that you either a) purposely altered settings in the control panel to yield a lower score b) are a moron with a FUBAR system. So which is it? Surely my results cannot be skewed considering I have an older card, slower system with an older mainboard chipset.

[euan]

Originally posted by marcocom
ease up tiger.

the Radeon drivers are not SMP aware nor do they support multi-threading (they still operate in single CPU mode. pretty damn well too). why should they? when they fully support linux, they will most likely bring that online, i would guess. (for example i can run all id games and all games based on Q3 or Doom engine in SMP accelerated mode. silky nvidia ncreamy smoothness my smp brother)

the FireGL drivers were multi-threaded when developed by DiamondMultimedia on the IBM chipset, but im not sure about it on the ATi solution. its probably a large reason why the current firegl solution is so much slower in applications than the QuadroFX.

Hey I know the drivers aren't SMP Aware, and aren't multi-threaded. If you have ever sat down and tried to write a video driver (I have access to ATI's driver development program) then SMP and multi-threading are the last thing on your mind. :) Still it hardly sites as a reason for making it "impossible" to use.

Last time I played Quake 3 (on my old 8500) it also ran great in SMP mode, it got nearly 2x the fps. Last I heard ID had broke it in a patch, then fixed it then they disabled it.

At the moment it's really only the R300 that are CPU driven on Linux (as far as I can remember). I started writing a Radeon driver for the R200 a while back, but someone else beat our team with a very good attempt just as we were getting started. The Beos Driver has hardware acceleration for all the necessary functions up to the R300. What do you class as hardware accelerated? What the Xfree86 R300 is missing is video-video blitting, and system-to-memory blitting etc, hence it has to manually redraw the framebuffer every frame, which is very CPU intensive. I think ATI just don't want to open source their hardware. We haven't even got the Xfree R300 code to work properly in BeOS. Very strange. Ati's "driver developer site" is of no use either, and they don't provide assistance.

Someone else commented on the Radeon 64 DDR BSOD in win2k... well yeah no change there, nor for everyone else. My old Athlon 800 had the same problem with a 7200, and Geforce 2 as well as the Tyan Tiger MP system. The story repeated for the first 6 months of WinXP as well, yet the PC's magically cured themselves through windows "critical" updates, many new bios's, and driver updates. :angel2:

[GlowStick]

Originally posted by 5150 Joker
LOL now you can't even keep track of who said what. Good god you truly are a lost cause.

Ran out of ideas?

Since i Quoted Rogo, i was responding too this last 2 posts.

He was confused, about the clock speeds we were useing.

[GlowStick]

Originally posted by euan
Hey I know the drivers aren't SMP Aware, and aren't multi-threaded. If you have ever sat down and tried to write a video driver (I have access to ATI's driver development program) then SMP and multi-threading are the last thing on your mind. :) Still it hardly sites as a reason for making it "impossible" to use.

Last time I played Quake 3 (on my old 8500) it also ran great in SMP mode, it got nearly 2x the fps. Last I heard ID had broke it in a patch, then fixed it then they disabled it.

At the moment it's really only the R300 that are CPU driven on Linux (as far as I can remember). I started writing a Radeon driver for the R200 a while back, but someone else beat our team with a very good attempt just as we were getting started. The Beos Driver has hardware acceleration for all the necessary functions up to the R300. What do you class as hardware accelerated? What the Xfree86 R300 is missing is video-video blitting, and system-to-memory blitting etc, hence it has to manually redraw the framebuffer every frame, which is very CPU intensive. I think ATI just don't want to open source their hardware. We haven't even got the Xfree R300 code to work properly in BeOS. Very strange. Ati's "driver developer site" is of no use either, and they don't provide assistance.

Someone else commented on the Radeon 64 DDR BSOD in win2k... well yeah no change there, nor for everyone else. My old Athlon 800 had the same problem with a 7200, and Geforce 2 as well as the Tyan Tiger MP system. The story repeated for the first 6 months of WinXP as well, yet the PC's magically cured themselves through windows "critical" updates, many new bios's, and driver updates. :angel2:

Your right, i did notice alot of the probelms with my Tyan Tiger MP system go away after a critical update! I get the feeling it was the Athlon Pageing problem? I do rember a certin set of nvidia drivers would cause the pageing to go wild. Scotty at http://www.commanderkitty.com had an attack of the same probelm. Weird.

[Rogozhin]

Alright

Here is my 9700 non pro score at 300/300.

http://service.futuremark.com/compare?2k3=655484

rogo

4200

with cat 3.2s and dx9.0a

I still think your 1500 score is messed up glowstick.

rogo

[Rogozhin]

Why would you underclock it in the first place?

I noticed that many of your 2003 run throughs weren't finished. Did you hack the registry?

Rogo

[GlowStick]

Originally posted by Rogozhin
Why would you underclock it in the first place?

I noticed that many of your 2003 run throughs weren't finished. Did you hack the registry?

Rogo

Well the 3dmark03 score is just based on the game tests, the rest is nice to compare to other systems.

Ok, my results may be odd.

I can look into this, i think may be a problem with underclocking it with my utility.

[bkswaney]

Low FPS Resolution dm-antalus dom-suntemple dm-phobos2 dm-inferno ctf-face3 ctf-citadel dm-asbestos
1024x768 85.917206 76.167679 51.772167 59.515678 30.955866 51.655670 95.369110


Average FPS Resolution dm-antalus dom-suntemple dm-phobos2 dm-inferno ctf-face3 ctf-citadel dm-asbestos
1024x768 144.606400 153.680740 151.748825 125.110313 159.209457 124.730850 207.808014


High FPS Resolution dm-antalus dom-suntemple dm-phobos2 dm-inferno ctf-face3 ctf-citadel dm-asbestos
1024x768 684.653381 456.609161 602.087769 469.915924 666.131226 311.095612 566.900085




This test was run using the High Quality Option
This test was run using the Direct3D Rendering Engine
This test was run using the full version of Unreal Tournament 2003
This test was run using the Custom_MaxDetail.ini provided by [H]ard|OCP


EDIT: These test were done with 2x AA and 2x AF. :)

[Rogozhin]

Bkswankey

Thank you I will go run my tests after I grab some beer ;)

Rogo

[bkswaney]

Originally posted by Rogozhin
Bkswankey

Thank you I will go run my tests after I grab some beer ;)

Rogo


Cool. :)

I cannot wait for NV to release the new Det FX drivers.
I bet we see a nice speed and IQ bump. :D

I played UT for about 5 hours last night.
It run so sweet all night. :D

[Rogozhin]

BK

Can you just run the regular benchmark real quick since I don't have the [H]ard ONE? ;)

My flyby bench at 1024x768 all options maxed in game (plus sound options maxed) with 2x AA and 2x AF are

85.100517 in Flyby.

Thanks ;)

Rogo

[Rogozhin]

BK

Thanks ;)

I did the bench but don't have the hard one as mentioned earlier. IF you don't mind would just bench using the regular one?

Glad you dig your new card ;)

Rogo

[StealthHawk]

Originally posted by nagus
.... and the winner is ...


Radeon 9700PRO >> 4674 points in 3Dmark2003 @ 300/300 :D :D :D

poor NV35 ;)

Which doesn't really mean anything considering that the r9700 should be clocked at the same proportion from retail clocks as the NV35 is to make a valid comparison.

As a side note....does every thread that even meantions NV30 always have to go down the ****ter? Must....resist.....urge....to moderatate.

[Rogozhin]

Cowboy.

Nice article.

Those damn brits are so cutting ;)

rogo