Both systems are now finished.

GLQuake @ 23.6 FPS AVG with sound @ 800x600.3DFX
GLQuake @ 26.4 FPS AVG without sound @ 800x600.3DFX

◙ Cyrix 5x86 133MHz
◙ 512K Cache
◙ 64MB FPM RAM
◙ Biostar MB-8433UUD
◙ Matrox Millenium G200-8MB
◙ Creative AWE64 GOLD-28MB
◙ 3DFX VOODOO2-12MB
◙ 20GB SCSI Hard Disk
◙ SCSI Plextor CD-ROM
◙ Sony 17" TFT LCD SDM-S73
◙ Honeywell-SUH keyboard
◙ Microsoft 2-Button PS/2 Mouse 2.0
◙ Gravis PC Gamepad
◙ Roland SC55-ST MIDI Device
◙ JBL GO Portable Bluetooth Speaker Red
◙ JBL GO Portable Bluetooth Speaker Black
◙ Full Tower case with MHz LED display.



GLQuake @ 30.6 FPS AVG with sound @ 800x600.3DFX
GLQuake @ 33.9 FPS AVG without sound @ 800x600.3DFX

◙ Intel PODP5V83 83MHz
◙ 512K Cache
◙ 64MB FPM RAM
◙ Elitegroup UM8810P-AIO
◙ Matrox Millenium II-4MB
◙ Yamaha YMF-718
◙ Creative Wave Blaster
◙ 3DFX VOODOO2-12MB
◙ 10GB SCSI Hard Disk
◙ SCSI Plextor CD-ROM
◙ Sony 17" TFT LCD SDM-S73
◙ Honeywell Litetouch Keyboard
◙ Microsoft 2-Button Mouse
◙ X-Fighter Thrustmaster Joystick
◙ Roland SC55 MIDI Device
◙ JBL GO Portable Bluetooth Speaker Red
◙ JBL GO Portable Bluetooth Speaker Black
◙ Full Tower case with MHz LED display.

Is the mouse issue still present in Win98? What about in WinNT4? I still encourage you to test with LSSER=off w/different Cyrix CPUs. I found some CPUs handle the special features better than others.

Did you try the POD83 in Windows? You should also run chkcpu to ensure that L1 is enabled with WB mode.

Too bad about the Voodoo2 SLI. Usually, when I upgrade graphics cards, even with the CPU is the bottleneck, I see some fractional percent increase, but it looks like that is not the case here.

rgart wrote:

Both systems are now finished.

Be careful with this type of talk, it will come back to byte you!

No the mouse issue is not present anymore. LSSER=off was causing it. I could recreate the problem every time and then resolve it every time by removing the line. Yes I'm using the POD83 in Windows 95/Windows 98. I was really disappointed with the Voodoo 2 in SLI, I was expecting *some* performance increase even if it was 1 or 2 fps.

I found another POD83 cheap and local, when It arrives I'll try it at 100MHz but besides that I wont be making any changes 😀

L1 in WB mode, check!

The only other thing I did before I closed up the cases *for good* was to install two case fans.

Each tower received a F12 and F8 Arctic case fan for silent operation; the larger F12 to circulate air and cool the Voodoo 2 and Matrox card and the smaller F8 to remove hot air from the case.

rgart wrote:

No the mouse issue is not present anymore. LSSER=off was causing it. I could recreate the problem every time and then resolve it every time by removing the line.

Then the issue is likely related to your specific CPU. You mentioned that the issue also occurred on the MB-8433UUD. I have every revision of this board and have not run into this problem. The ability to use certain performance and control registers, such as LSSER, have shown frequency-dependence and voltage-dependence. In other cases, some CPUs just don't like certain features irregardless of external factors, however, I found this cause to be rare. I suggest you test alternate Cyrix 5x86 CPUs, alternate operating frequencies, and voltages up to 3.75 V to further isolate the problem.

Remeber, LSSER was shown to have an average performance boost of 8%, but some benchmarks indicated up to a 47% improvement.

rgart wrote:

besides that I wont be making any changes 😀

Heard that one before.

rgart wrote:

GLQUAKE:
My Cyrix 5x86-133/4x :
◙ width 800 -height 600 -bpp 16 -waveonly -nomouse -nosound +timedemo demo1 = 26.3 fps

rgart wrote:

Both systems are now finished.
GLQuake @ 28.0 FPS AVG without sound @ 800x600.3DFX
◙ Cyrix 5x86 133MHz

Why the difference between then and now? With LESSER=1, the results should have become worse, not better.

feipoa wrote:

rgart wrote:

GLQUAKE:
My Cyrix 5x86-133/4x :
◙ width 800 -height 600 -bpp 16 -waveonly -nomouse -nosound +timedemo demo1 = 26.3 fps

rgart wrote:

Both systems are now finished.
GLQuake @ 28.0 FPS AVG without sound @ 800x600.3DFX
◙ Cyrix 5x86 133MHz

Why the difference between then and now? With LESSER=1, the results should have become worse, not better.

The LSSER=OFF instability ONLY occurred on my Cyrix 5x86 120MHz Stepping 1 Revision 3, not the Cyrix 5x86 133MHz Stepping 0 Revision 5 you quoted above.

Thanks for pointing out my mistake, I have retested and edited that post with the correct value which was 26.4 FPS and not

+timedemo demo1

-identical Voodoo 2 cards-
-identical SCSI cards-
-identical OS (Windows 95 C 4.0 Build 1111)
-identical DirectX 6.1 (4.06.02.0436)

Cyrix 5x86 133MHz + MB-8433UUD

GLQUAKE 800x600 with sound (no parameters set)

1st run : 23.4
2nd run : 23.4
3rd run : 23.6
4th run : 23.6
5th run : 23.6

GLQUAKE 800x600 without sound (-nojoy -nonet -nomouse -nosound -nocdaudio)

1st run : 26.3
2nd run : 26.4
3rd run : 26.4
4th run : 26.4
5th run : 26.4

With console open +timedemo1 +sound +transparent effect : 18.1

Intel PODP5V83 83MHz + UM8810P-AIO

GLQUAKE 800x600 with sound (no parameters set)

1st run : 30.6
2nd run : 30.4
3rd run : 30.6
4th run : 30.6
5th run : 30.6

GLQUAKE 800x600 without sound (-nojoy -nonet -nomouse -nosound -nocdaudio)

1st run : 33.9
2nd run : 33.9
3rd run : 33.9
4th run : 33.9
5th run : 33.9

With console open +timedemo1 +sound +transparent effect : 23.6

For those interested in my version numbers:

Cyrix 5x86 133MHz

Intel PODP5V83 83MHz

[/b]

Last lot of benchmarks:

DOS 7.10 TESTS

BTB_EN=ON, FP_FAST=ON, LSSER=OFF for Cyrix enhancement gains and all else OFF in the Peter Moss utility for stability throughout all tests. (so obviously DOS settings)

All systems optimal bios settings: 2,1,1,1 with no wait states + all systems in WB mode.

(All systems 64MB RAM)

Cyrix 5x86 120MHz @ 40MHz FSB - UM8810P-AIO with G200

Speedsys: 67.64
3Dbench 1.0c : 103.0
Pcpbench /vgamode : 23.1
Pcpbench 100 640x400 : 9.4
Pcpbench 103 800x600 : 6.2
Pcpbench 105 1024x768 : 4.3
Quake 1.06 SW : 17.1

Cyrix 5x86 133MHz @ 33 MHz FSB - MB-8433UUD with G200

Speedsys: 75.50
3Dbench 1.0c : 90.6
Pcpbench /vgamode : 23.7
Pcpbench 100 640x400 : 8.4
Pcpbench 103 800x600 : 5.3
Pcpbench 105 1024x768 : 3.6
Quake 1.06 SW : 17.8

Intel PODP5V83 83MHz @ 33MHz FSB - UM8810P-AIO with Millenium II

Speedsys: 61.40
3Dbench 1.0c : 84.4
Pcpbench /vgamode : 22.5
Pcpbench 100 640x400 : 8.5
Pcpbench 103 800x600 : 5.3
Pcpbench 105 1024x768 : 3.8
Quake 1.06 SW : 21.3

The Quake DOS results for the POD83 are pretty good, however my IBM 5x86-133c gets 19.8 fps, which really gets close to closing the gap and rendering the Pentium of little benefit. This is why a POD at 100 MHz is far more interesting. Too bad none of your POD's can work at 100 MHz in the non-modified state. Of the 3 POD83's I have, one works at 100 MHz, unmodified. Another one works fine at 100 MHz with the diode mod and I trust it more than the unmodified unit.

It would be interesting to see a an Am5x86-160 installed on your UM8810P-AIO motherboard.

If I had another mini tower, I would put a POD100 in there with the MB-8433UUD.

Fair to say the 5x86-120 is a better gaming chip because the fsb is higher? Or do we need more tests?

The DOS Quake results were better on the Cyrix 5x86-133 compared to the Cyrix 5x86-120, however the pcpbench and 3dbench results were better on the 120. Rgart, could you run DOOM timedemo3 on both systems?

feipoa wrote:

The Quake DOS results for the POD83 are pretty good, however my IBM 5x86-133c gets 19.8 fps, which really gets close to closing the gap and rendering the Pentium of little benefit. This is why a POD at 100 MHz is far more interesting. Too bad none of your POD's can work at 100 MHz in the non-modified state. Of the 3 POD83's I have, one works at 100 MHz, unmodified. Another one works fine at 100 MHz with the diode mod and I trust it more than the unmodified unit.

It would be interesting to see a an Am5x86-160 installed on your UM8810P-AIO motherboard.

If I had another mini tower, I would put a POD100 in there with the MB-8433UUD.

Your IBM 5x86 @ 66MHz FSB kills all challengers feipoa.

I'm going to play through GLQuake on the POD with Voodoo 2. Beautiful, Bright, Vibrant and 7 fps faster with sound or without than the Cyrix 5x86 133MHz.

I'll definitely have a play with the AM5x86 @ 160MHz sometime, I actually pulled a couple of those CPU's out of their static bags prior to trying the POD but when I looked at the UM8810P-AIO manual there were no jumper settings so I moved on to the POD 83MHz which had documented settings.

I'm also waiting for another POD 83MHz to arrive so hopefully that runs at 100MHz. If you ever feel exceptionally generous and want to diode MOD my CPU's.... 😀

I had a few hours to play before work today. Here are the doom results you requested:

Doom 1.666 (+timedemo demo3)

Cyrix 5x86 120MHz @ 40MHz FSB - UM8810P-AIO with Millentium II

7001 gametics 4389 realtics = 55.8 fps
7001 gametics 4387 realtics = 55.8 fps

Cyrix 5x86 133MHz @ 33 MHz FSB - MB-8433UUD with G200

7001 gametics 4838 realtics = 50.6 fps
7001 gametics 4839 realtics = 50.6 fps

Without a doubt Doom is visibly faster on the 120MHz @ 40FSB, it's the same with SVGA modes in pcpbench which I notice too.

I picked up some PCI video cards on eBay for about $5 - $8 each. I don't know much about them and haven't done a whole lot of googling yet, anything you guys could add I would really appreciate. I know the 3DLabs/Integraph cards are for CAD programs and require Pentium/Windows NT ( according to the manual) I have seen Win95 drivers floating around though. Will they support OpenGL and possibly GL Quake?

This card works fine with the Cyrix 5x86 120MHz + UM8810P-AIO mobo. It has a RAGE XL chip but seems to identify itself as a HP Remote Insight Lights-Out Edition II and has a 200MHz CPU onboard and a BIOS.

Reasonable results.

3Dbench 1.0c : 103.0
Pcpbench /vgamode : 22.7
Pcpbench 100 640x400 : 9.4
Pcpbench 103 800x600 : 6.2
Pcpbench 105 1024x768 : 4.3
Quake 1.06 SW : 16.9
Doom 1.666: 54.6

My full tower case seems to accommodate these long-ass full length cards perfectly and the ends slide into these grooves.

Intergraph Intense 3D 2200 & 3600 16Mb PCI Video Cards

The two Intergraph cards above did NOT like the Cyrix 5x86 120 + UM8810 motherboard as I had expected.

Pleasing results from the little Hercules Stingray Pro with the Cyrix 5x86 120MHz + UM8810P-AIO

3DBench: 103.3
pcpbench vgamode: 23.1
pcpbench 100 640x400: 10.6
pcpbench 103 800x600: 6.7
pcpbench 105 1024x768: didn't work with univbe 5.1a or without. (didnt try 6.7)

Did those full length cards work in any socket 3 system?

Yes the first full length card pictured. HP Remote Insight Lights-Out Edition II. It has a rage XL chipset.

I'll try the others in the MB-8433UUD in a couple of days when I have time.

rgart wrote:

This card works fine with the Cyrix 5x86 120MHz + UM8810P-AIO mobo. It has a RAGE XL chip but seems to identify itself as a HP R […]
Show full quote

This card works fine with the Cyrix 5x86 120MHz + UM8810P-AIO mobo. It has a RAGE XL chip but seems to identify itself as a HP Remote Insight Lights-Out Edition II and has a 200MHz CPU onboard and a BIOS.

Reasonable results.

3Dbench 1.0c : 103.0
Pcpbench /vgamode : 22.7
Pcpbench 100 640x400 : 9.4
Pcpbench 103 800x600 : 6.2
Pcpbench 105 1024x768 : 4.3
Quake 1.06 SW : 16.9
Doom 1.666: 54.6

Did the WIndows display drivers install?
What GLQuake score does it get?

The Intergraph cards didn't work in the MB-4833UUD or UM8810P-AIO socket 3 motherboards.

The HP Insight Lights-Out II appears to be just a basic Rage XL chip with the 200MHz CPU and other jazz dedicated to proprietary remote access for Windows NT Compaq Proliant servers. It benchmarked fine in DOS with similar results to the RAGE 128 VR and RAGE PRO.

I didn't bother loading up Windows 95. It probably will mean *another* fresh install when Windows 95 complains.

Tested an Am5x86 CPU @ 160 MHz in the UM8810P-AIO board. Good results.

Speedsys : 59.93
3DBench2 : 102.4
pcpbench /vgamode : 25.8
pcpbench 640x400 : 11.6
pcpbench 800x600 : 7.3

Quake 1.06 Shareware (nosound): 17.7
Doom 1.666 Shareware (nosound): 63.25
GL Quake 0.97 with Quake 1.09 - 800x600 (nosound): 26.3
GL Quake 0.97 with Quake 1.09 - 800x600 (nosound): 26.4