Ok, there is a new benchmark that I want to put up, so can the regulars please download the following program : 
SiSoft Sandra 2001
so that the new figures can be put up, don't worry its only a 2 megs

These Benchmarks, will move to their own pages later, accompanied by graphs. I am busy trying out various packages to see whose graphs look the most appealing. At the moment I am leaning towards graphs that resemble firingsquad type graphs. I find that they aren't to big, and convey the message pretty succinctly.

Index :
- HOME
- System Setup

rants and raves :

Seeing that 3dfx is no more, you will be needing driver updates. Seeing that I am not currently busy with that, here are some links to ppl that are making them.

1. Tempest(the best I think)

Overclocking Gigabyte:
Just in case you were wondering what overclocking program I was using to overclock my Thunderbird, it is EasyTune III, but not the one I received from the CD of the motherboard, cause that one don't work. I had to download a newer version from Gigabytes web site. (just by the way, the Taiwan site SUCKS !!) rather go to the american one :
gigabyte.(MUCH faster)
One web site said that they could only overclock the chip to the recommended overclock, which 842 MHz, but I managed to push mine to 900 MHz (It was a rainy day, so my room was nice and cool, seeing that it is summer here in Cape Town now, I can only get up to 888 MHz)

Norton 2000

For this benchmark, the CPU and motherboard, play an important role. The amount and speed of the RAM does not seem to play much part, cause the numbers stayed the same for my machine when switching between 100 and 133 MHz. The only reason I can think of for the poor performance of the PIII 866, is the relatively unknown motherboard maker.
 
CPU and motherboard
Norton 2000 System Benchmark
PII 266@400 - Gigabyte
 191.8 
PIII 600 - Intel
279.4
K7 800@720 - Gigabyte
390.5
K7 800 - Gigabyte
433.4
PIII 866 - (unknown)
353.6
K7 800@872 - Gigabyte
473.9
Now for some game benchmarks: Some of these games are a bit old, but they are still enjoyable.

GL Hexen 2

Hexen 2 has 2 different modes in which you can play the game, either software, or in GL mode. The ATI Rage II could not run in GL mode, this is a driver problem. At the moment nothing can even come close to the performance of the Voodoo 2, not even non SLI mode ! (please note that this is a relative comment, pertaining to the video cards represented here ! ). I was expecting more from the TNT2 M64 though. Strangely enough even he had some troubles with GL mode.
CPU & video card 640x480 800x600
600 Rage Fury II 28.9* 20.4*
800 non SLI 73.5 47.3
800 SLI voodoo 2 125.7 87.1
866 TNT2 M64 -- 19.0

GL Quake 2

Once again we see the same trend. Except that here the 266@400 MHz beats the 866 PIII. Strange because they have the same video card. Level G made an observation that his video RAM is running @ 160 MHz and that of the 866 is @ 130 MHz. I was a bit reluctant to believe that 30 extra MHz would  make that much of a difference, but there you go (I personally believe it to be a driver issue). The numbers that I got for the 800 machine, were taken with enhanced dll's for 3DNOW instructions. (Only fair seeing that Quake 3 has built in SSE support)
CPU 640x480 800x600
600 28.6* 21.1*
800 non SLI 90 80
800 SLI 110 99
866 33.3 29.5
266@400 60 50
*Once again driver problems with ATI results in only software mode possible

Unreal Tournament

Well here the huge gap can be expected, not due to the fact of K7 dominance, but because the Unreal Engine was built for GLIDE specifically. (If you can remember when Voodoo 3 was released, the game was bundled with the card) Even with the 32 MB of the TNT's, they don't even come close.
CPU 640x480 800x600 1024x768
600 -- --
800 non SLI 34.29 - 139.42
ave: 84.5
9.13 - 98.75
ave: 59.23
800 SLI 36.27 - 154.7
ave: 77.6
30.27 - 116.51
ave: 70.36
driver problems
866 27,61 - 57,89
ave: 34,86
19,15 - 36.0
ave: 25,77
n/a
Due to problems with UT and ATI, the 600 machine cannot run in D3D, so we had to take software numbers. If you have the same card, and found a work around, please contact me.
The numbers above the average, is min and max framerate.

For the first time non SLI beats SLI by quite a couple of frames, I will have to redo this one a couple of times just to make sure of this result.(Its just that I have been having trouble creating demos with the UT, they dont seem to be recording properly)
= Note to Ralph, to due to your extreme slowness (hehe) you are not represented in some of the benchmarks !

Hosted by www.Geocities.ws

1