Again, whats your 3DMark 2000 results?

[Sanjuro Makabe]

WOW dZeus!
8000 x 600 huh....hehe....I guess it should be 800x600...hehe...no offense...

I am getting 3229 in 3dMark2k


------------------
Sanjuro Makabe

P!!!550E @ 803Mhz
ASUS P3V4X 1004.02, 4in1 4.22, AGP 4.03
128Mb PC100 C2 Ram
Matrox G400Max PD5.55
SBLive! LW3.0
WinMe 4.90.2525

[pepe]

my score with PIII650@728
256 ram pc100
gigabyte 2000 mobo
g400 max 10% oc
3D MARK 2000 default :3359

[Arsa]

When I was lucky enough to finish it I got 0.
Athlon 700, G400 32Mb DH, K7V.

[Rik]

No-one seems to have posted workable Win2000 results, so here's what I have managed to get:-

First, the goods:
PIII-500 @ 560, Asus P2B, G400 non-max @ 155mhz, 128mb.

All test are the default (1024, 16bit).
Win2000 with 5.04 drivers, DX8 beta:

3DMark Result: 2579 3D marks
CPU Speed: 171 CPU 3D marks
Game 1 - Helicopter - Low Detail: 53.9 FPS
Game 1 - Helicopter - Medium Detail: 35.1 FPS
Game 1 - Helicopter - High Detail: 15.1 FPS
Game 2 - Adventure - Low Detail: 55.8 FPS
Game 2 - Adventure - Medium Detail: 33.9 FPS
Game 2 - Adventure - High Detail: 21.1 FPS
Fill Rate (Single-Texturing): 273.4 MTexels/s
Fill Rate (Multi-Texturing): 275.0 MTexels/s
High Polygon Count (1 Light): 1726 KTriangles/s
High Polygon Count (4 Lights): 1554 KTriangles/s
High Polygon Count (8 Lights): 1379 KTriangles/s
8MB Texture Rendering Speed: 195.6 FPS
16MB Texture Rendering Speed: 175.7 FPS
32MB Texture Rendering Speed: 125.2 FPS
64MB Texture Rendering Speed: Not enough AGP memory FPS
Bump Mapping (Emboss, 3-pass): 75.8 FPS
Bump Mapping (Emboss, 2-pass): 98.5 FPS
Bump Mapping (Emboss, 1-pass): 145.8 FPS
Bump Mapping (Environment): 80.1 FPS

For comparison, Win98SE with 5.52 driver:

3DMark Result: 2643 3D marks
CPU Speed: 171 CPU 3D marks
Game 1 - Helicopter - Low Detail: 52.5 FPS
Game 1 - Helicopter - Medium Detail: 34.8 FPS
Game 1 - Helicopter - High Detail: 16.0 FPS
Game 2 - Adventure - Low Detail: 59.2 FPS
Game 2 - Adventure - Medium Detail: 36.1 FPS
Game 2 - Adventure - High Detail: 21.6 FPS
Fill Rate (Single-Texturing): 277.5 MTexels/s
Fill Rate (Multi-Texturing): 276.1 MTexels/s
High Polygon Count (1 Light): 1986 KTriangles/s
High Polygon Count (4 Lights): 1803 KTriangles/s
High Polygon Count (8 Lights): 1612 KTriangles/s
8MB Texture Rendering Speed: 229.7 FPS
16MB Texture Rendering Speed: 212.1 FPS
32MB Texture Rendering Speed: 134.6 FPS
64MB Texture Rendering Speed: Not enough AGP memory FPS
Bump Mapping (Emboss, 3-pass): 101.4 FPS
Bump Mapping (Emboss, 2-pass): 129.5 FPS
Bump Mapping (Emboss, 1-pass): 231.6 FPS
Bump Mapping (Environment): 96.2 FPS


And Win98SE with the 5.55 driver:

3DMark Result: 2908 3D marks
CPU Speed: 190 CPU 3D marks
Game 1 - Helicopter - Low Detail: 60.5 FPS
Game 1 - Helicopter - Medium Detail: 39.9 FPS
Game 1 - Helicopter - High Detail: 17.6 FPS
Game 2 - Adventure - Low Detail: 60.9 FPS
Game 2 - Adventure - Medium Detail: 39.6 FPS
Game 2 - Adventure - High Detail: 23.9 FPS
Fill Rate (Single-Texturing): 277.5 MTexels/s
Fill Rate (Multi-Texturing): 276.1 MTexels/s
High Polygon Count (1 Light): 2093 KTriangles/s
High Polygon Count (4 Lights): 1914 KTriangles/s
High Polygon Count (8 Lights): 1709 KTriangles/s
8MB Texture Rendering Speed: 230.1 FPS
16MB Texture Rendering Speed: 222.4 FPS
32MB Texture Rendering Speed: 156.4 FPS
64MB Texture Rendering Speed: Not enough AGP memory FPS
Bump Mapping (Emboss, 3-pass): 101.4 FPS
Bump Mapping (Emboss, 2-pass): 129.6 FPS
Bump Mapping (Emboss, 1-pass): 231.8 FPS
Bump Mapping (Environment): 96.5 FPS

Win2000 performance with the 5.04 driver was thus not too bad for me, at least until the advent of the 5.55 Win98 driver.

[archangle]

dZeus,

Thats all that matters, as long as your happy with the performance (eg it dosn't ruin your gamming experince) then you'll enjoy playing gammes

Personally I don't mind if other poeple get higher results, some are getting high 6000's with GTS 2 systems. The reason I posted this is to see if anyone is beating 3500, my current target for a G400 system. I doubt if I'll get it without at least a 1Ghz CPU, well theres always this xmas

[dZeus]

I don't mind there are people that have higher scores, I do mind if there's something wrong with my system. That's all I wanted to figure out. All I wanted to know is scores of 3Dmark in win2k on a g400, with people that do not have a SSE/3Dnow! capable CPU. I wanted to know how much the scores were dependend on these extra instructionsets. Then I would know what to expect from a CPU upgrade. Fortunately next week I will

[paulcs]

Does anyone here think 3DMark2000 might be optimized for nVidia drivers (or visa versa)? Scores with the G400 seem low compared to the TNT2 Ultra, whereas the reverse was true of the 3DMark99. I've been suspicious of it relative to the G400 for a while.

It just doesn't seem to scale well from manufacturer to manufacturer, and I'm not sure how good an indicator it is of realworld performance.

I'm also wondering about what people think about 3DMark2000's CPU scalability. It's almost too good.

Paul
paulcs@flashcom.net

[GT98]

Heres my numbers for 3DMark 2000:

2959 for a Celeron2@877 with the 5.52 Drivers from matrox in Win98SE

Scott


------------------
Abit BH6 with a Celeron 2-566@877mhz,128mb RAM,G400 MAX,SB Live!, Optiquest V95 19in montor, Asus 40x CD-ROM, Aopen 5x DVD-ROM,HP9110i 8x4x32 CD-RW,SupraMax 56k modem, Win98 SE on Western Digital 30.7GB, Win2000 on a WD 13 GB HDD

[dZeus]

3D mark 2000 is a farce....

I guess nVidia sponsers MadOnion or something similar.... in 3Dmark2K HW T&L gives a far bigger performance boost that it would do in a real-world game.

CPU-score is not even measured, but looked up in a table or something similar. If you change the speed of you CPU with softFSB after you started 3Dmark2K, it doesn't change it's CPUMARK score.

To put it plain and simple:
3Dmark2K sucks as a inter-computer benchmark. It is only usefull for very rough indication, and to look if any BIOS tweaks/different drivers and such have an effect on the score. And it looks quite cool.... but that's everything

[Nuno]

Paul:
I take 3dmark2k with a BIG grain of salt. It´s by no means a real world performance indicator. Both by good and bad reasons.

The GOOD: It has an *excelent* SIMD implementation, all the apps/games should be otimized like that. After all it´s suposedly based in a game engine. How hard can that be?
That´s why PIII/Athlons left celeron/PII in the dust scoring 50%+ higher at the same clock speeds. In real world SIMD optimization implementation only gain ~20%.

The BAD: It´s a little Nvidia biased. It has lots more polygons that any current game. It has T&L that very few games actually implement. I said "a little" because as I stated it does a very above-average job doing T&L with SIMD instructions. And to answer your question, that´s why it scales so well with faster cpu´s, even with T&L cards. That´s really a lot of geometry calculation to do.

It´s Intel biased. The Athlon at the same clock speed has more raw fpu power than PIII and the ehanced 3dNow! instructions are rougly SSE equivalent, performance wise. They could have put a little more effort in the athlon dll.

It could be actually flawed. I saw several reports on the net of the cpu score being derived of the cpu speed in mhz when the program starts. Like start 3dmark, bench, you got 170 cpu marks. DOWN the clock speed with fsb. Bench. 170 3dmarks again. Close the program and run it again maintaining the lower clock speed. Bench. 150 cpu marks??

PS: Oooops, didn´t see the message above. Basically, I agree

[This message has been edited by Nuno (edited 07 June 2000).]

[archangle]

I agree, 3DM2K is very dodgy. I use it to compare like systems eg G400 & Athlon and my own against tweeks etc.

I only own the Q3Demo and don't play it much so I rarley benchmark with it.

Just O/Ced my Athlon 500@715 (110x6.5) could tell it was faster but it overheated before 3DM2K could finish (Hey, I am only using a single 500Mhz rated fan!!!). Looking at a new fan every day now. Should be able to get 800 (100x8) or 770 (110x7).

That 3DM2K of 3500 is getting closer

[g_goncalo]

where can i "download" the result browser?
Maggi@home - i would like to make a few questions about you new cell. Can you send me you e-mail?
g_goncalo@hotmail.com

[Jazza]

At 1024 16bit I got 2413 with the .52 drivers, then I installed Marvel .55 and got 2617, which was nice

--
Jazza

P3 500 (not O/C)
Gigabyte BXDS
256Mb Ram
Mator 13.5Gb IDE
Fujusi 15Gb IDE
Quatum 1Gb SCSI
Creative 5x DVD
Yamaha 4416s CDRW
Iomega Zip
SB Live
G400 Max
Win98se

[archangle]

I'm getting everyone over at the 3DFIles board to do benching to see the difference SSE, 3DNow!, Athlon & T&L make in 3DMark 2K. Can you all go over and add your results.
http://board.3dfiles.com/Forum1/HTML/048710.html

Cheers,

archangle

[WarFlail]

2991 (default benchmark test)

-=W=-


------------------
My system specs:
Intel Pentium III 650MHz
Asus P2B motherboard
128MB PC-100 SDRAM
Matrox G400 32MB SH @ 184.50MHz, running ParaKnowYa's driver pack v1.5
Creative Sound Blaster Live! Value with latest drivers
Quantum Fireball ST 6.4GB & 4.3GB
I/O Magic 8X DVD-ROM (region-free)
Kenwood 42X TrueX CD-ROM
CTX PR711 17" Monitor, currently running 1024x768x32 @ 85Hz
Windows 98 (not SE)
3DMark2000 default benchmark score: 2991