it's annoying to see the variation with driver version, with my stock FX 5900XT on a P4 with the 845PE it get around 12.5k with the latest driver, 13.8 with 93. something and 14.8k with 53.03...
(all on XP SP3 and with the default driver quality settings).

oh well, GMA 950 is not so bad on 3dmark 2001!

Just got my new CPU, haven't done any overclocking to it yet, or gotten my XMP 4133Mhz Ram setup yet:

Here are my Scores:

Config: Abit KT7A @ 150-MHz FSB/RAM, CL2 everything on Turbo, Athlon XP-M @ 1800-MHz, 1,28VCore, GeForce 4 Ti 4200 (Albatron, simple layout) AGP x8 @ default (GPU 250-MHz/ MEM 513-MHz) and of course 43.45 Drivers.

Intellisample @ HQ

Intellisample @ HP

without changing any setting my score miraculously raised to this 😳

After the strange Rise Of The G-Raider I can proudly submit this Score (overclocked GPU)

11089 Points are imho pretty good. Windows 98 at it's best

As it is a benchmark and is meant to test your system performance, there are no strict minimum requirements. However, our offici […]
Show full quote

As it is a benchmark and is meant to test your system performance, there are no strict minimum requirements. However, our official minimum requirements are 128 MB of RAM, a 500 MHz Intel-compatible processor and a modern 3D accelerator with 32MB of video memory.

The 128MB of memory is an absolute requirement, but you may be able to run it with a slower CPU and an 8 or 16 MB graphics card. Beware - it is definitely not meant for low-end systems.

The benchmark is meant for the games of the future and will require high-end hardware. A new Pentium III, Pentium 4, Celeron (SSE), Athlon or Duron system is recommended, as is a Hardware T&L compatible 32MB 3D accelerator. A fully DX8 compatible card wouldn’t hurt either, although unfortunately, there aren’t that many out there.

the slowest cpus that can run 3dmark01 are idt winchip and cyrix 6x86.
slowest video cards that can finish 3 tests at lowest resolution(640*480*16bit) are sis6326 and s3trio3d.
slowest video card that can finish 3 tests at default resolution(1024*768*32bit) is matrox g200-16mb.
slowest video card that can run all 4 tests is r9200se.
also, i think running on 64mb machine is possible, especially with a tidied up win98.

That's interesting, I've wondered before about the Matrox G200.

ETA: (reviewing the thread) man, Imageshack really did a number on the legacy internet.

Errius wrote:

That's interesting, I've wondered before about the Matrox G200.

ETA: (reviewing the thread) man, Imageshack really did a number on the legacy internet.

it would surely run, but can also depend on os and driver version. i think g200's driver had never been very stable till the end, although slightly better than s3 savage3.

Modern PC, Windows 98 SE, and not so modern GPUs 😜

I have heard people saying in the forums many times that Geforce MX4000 is a really bad card.
I have had Geforce 2 MX in my P3 1000Mhz retro pc for a while now and I decided to try MX4000 card that I got for free last year for the first time. Both cards are passively cooled so no noise and both are agp cards.
3dmark01 score for Gerforce 2 MX: 2438
3dmark01 score for Geforce MX4000: 3277

Which one would you keep in the system?

Baoran wrote:

Which one would you keep in the system?

Neither. Get a Voodoo 3 3000 AGP for compatibility or a GF4 Ti4200 for performance. Even a GF3 Ti200 would be a major upgrade.

(I suppose if I had to choose, I would keep the MX4000)

KT7AGuy wrote:

Baoran wrote:

Which one would you keep in the system?

Neither. Get a Voodoo 3 3000 AGP for compatibility or a GF4 Ti4200 for performance. Even a GF3 Ti200 would be a major upgrade.

(I suppose if I had to choose, I would keep the MX4000)

I think voodoo 3 would be bit redundant because the pc already has dual voodoo 2 cards, so the reason for trying to find a good nvidia card to complement the voodoo 2 cards was to keep dos compatibility while getting something faster for bit newer directx games.
My voodoo 2 SLI only got 888 points in 3dmark01 on that system.

Baoran wrote:

I think voodoo 3 would be bit redundant because the pc already has dual voodoo 2 cards, so the reason for trying to find a good nvidia card to complement the voodoo 2 cards was to keep dos compatibility while getting something faster for bit newer directx games.
My voodoo 2 SLI only got 888 points in 3dmark01 on that system.

Aha! I wasn't aware that you had Voodoo 2 cards in SLI. You're right; that makes a Voodoo 3 redundant.

The MX4000 is better than your GeForce 2 MX card. However, it's still pretty weak and won't take full advantage of your 1ghz CPU. In this case, I recommend one of these instead, ranked from low-to-high performance:

GeForce 3 Ti200
GeForce 4 MX460
GeForce 3
Geforce 4 Ti4200

Prices on these cards are still reasonable.

This video may interest you:
Why did nobody buy the GeForce4 MX 460?
https://www.youtube.com/watch?v=_qWvc94DjJE

Also, if your primary interest is DOS compatibility, then check out this spreadsheet:
https://gona.mactar.hu/DOS_TESTS/

Integrated graphics: Sandy Bridge vs Haswell

Haswell is only 2 generations ahead of Sandy Bridge, but the difference in IGP performance is absolutely MASSIVE!

i7-4790 (Haswell) 3DMark01:

i7-2600 (Sandy Bridge) 3DMark01:

Haswell 3DMark2005:

Sandy 3DMark2005:

Standard Def Steve wrote:

Haswell is only 2 generations ahead of Sandy Bridge, but the difference in IGP performance is absolutely MASSIVE!

Less massive as if you were to check the HD 3000 IGP with 12 execution units, for example in the i7-2600K. The i7-2600 and most of the other desktop Sandy Bridge chips use the 6-EU HD 2000.

System specs:
- ECS P6S5AT with SiS635T chipset
- Pentium III-S 1400 MHZ Tualatin (SL6BY).
- 512 MB DDR RAM 2-2-2-6
- Win98SE (Version 4.10, Build: 2222) with the latest service pack (SP.2.1b).
- Direct X 8.1.( 4.08.01.0901).
- SiS AGP GART Driver 1.07

GeForce 4 Ti 4600 128MB (44.03)

GeForce 6 6800 GT 256MB (81.98)

VooDooMan wrote:

GeForce 6 6800 GT 256MB (81.98)

That is probably one of the highest scores I've seen for a stock-clocked PIII! Any chance you could overclock the CPU to 1575MHz? I'd love to see how that 635T board stacks up against my carefully tuned QDI board clock-for-clock. 😁

EDIT: Forgot to mention that I overclocked my 6800GT to 385/540 to hit my score.

Standard Def Steve wrote:

VooDooMan wrote:

GeForce 6 6800 GT 256MB (81.98)

That is probably one of the highest scores I've seen for a stock-clocked PIII! Any chance you could overclock the CPU to 1575MHz? I'd love to see how that 635T board stacks up against my carefully tuned QDI board clock-for-clock. 😁

EDIT: Forgot to mention that I overclocked my 6800GT to 385/540 to achieve that score.

Wouldn't P3 cpu be the bottleneck with such a new video card?

Standard Def Steve wrote:

VooDooMan wrote:

GeForce 6 6800 GT 256MB (81.98)

That is probably one of the highest scores I've seen for a stock-clocked PIII! Any chance you could overclock the CPU to 1575MHz? I'd love to see how that 635T board stacks up against my carefully tuned QDI board clock-for-clock. 😁

EDIT: Forgot to mention that I overclocked my 6800GT to 385/540 to hit my score.

NO PROBLEM! 😁 I will do that today 😀 All of my SL6BY do 150 FSB at ease without adding voltage so no worries 😉 The problem for me is to find a good batch that will do 166 FSB... Then this motherboard with DDR 333 at 2-2-2-6 and the correct AGP divider will turn into real monster ;P

EDIT: Please notice, there is almost no difference between 6 6800 GT 256MB (81.98) and 5950 Ultra (44.03) on this mobo when 5 FX is on 150 FSB:

Baoran wrote:

Wouldn't P3 cpu be the bottleneck with such a new video card?

Yes, it is pointless to use such a powerful video card with this CPU, I installed it only to check the benchmark results. What holds this GPU( on this config) even more is the drivers... 44.03 are much faster than 81.89. Personally I think that Ti 4600 is the perfect match for Tualatin 1400 MHz 😉 I had 2 builds like that - one on ASUSU TUSL2C, and one on ABIT VH6T, but now that I've discovered that both of those motherboards suck in comparison to ECS P6S5AT I want to make some changes 😜

Standard Def Steve wrote:

VooDooMan wrote:

GeForce 6 6800 GT 256MB (81.98)

That is probably one of the highest scores I've seen for a stock-clocked PIII! Any chance you could overclock the CPU to 1575MHz? I'd love to see how that 635T board stacks up against my carefully tuned QDI board clock-for-clock. 😁

EDIT: Forgot to mention that I overclocked my 6800GT to 385/540 to hit my score.

Here are the results 😉

SL6BY@150 FSB, GeForce 6800 GT Default CLOCKS (81.98)

SL6BY@150 FSB, GeForce 6800 GT OC 385/540 (81.89)

I must admit that you have WON this time! 😉 But there are reasons why I've lost:
1 thing is that my ram runs at CL2,5 your is CL2 ;P
2 thing you did your test on WinXP, I did on Windows98se 😉

To make this battle "fair" I have to find better RAM and install WindowsXP 😀 but this is going to take some time...

When it comes to lowering CAS latency, I assume I could score around 200 points more at CL2 😉
I made a test some time ago: I checked the results on CL2 vs CL2,5 using Ti4600 - the difference was slightly over 100 points. In this case 6800 GT is much faster than ti4600 so i expect around 200 points more at 2-2-2-6 but I might be wrong. Still I hope that one day I will be able to alter tRAS to set it to 5 with appropriate memories 😀

I have no idea what difference in performance switching from Win98 to XP would make with 6800 GT...

That's awesome. Thanks for redoing the benchmark at 1575MHz. No doubt you're going to beat my score once you get CL2 memory. And yeah, the 6800GT is definitely more of an XP-focused card, so you'll probably gain a bit from that, too.

And here I thought no one would even come close to challenging my DDR Tualatin 3DMark scores. 😀