I did confirm the second board is also at BIOS Rev. 0305 (from 1995). I think I'm all set though - someone has graciously offered to burn me two of these DIP-28 chips with the latest BIOS for these boards. I'll let everyone know how it pans out. I also picked up a Write-Back version of the Intel DX4-100 (SK096) that I'll be testing against the AMD DX4-S 100 Write Back processor.

While I wait for these chips to arrive over the next few weeks, I think I'll start a build with the PCI/I-486SP3G and benchmark it. Anyone know if this one allows a software BIOS flash or will I run into the same issue? I checked the Asus web page and it doesn't reference anything about using a EPROM programmer like the other board:

http://www.asus.com.tw/Products/BIOS/bios-486.html

PCI/I-486SP3G

awsg0302.zip
PCI/I-486SP3G M/B, BIOS ver. 0302.

awsg0304.zip
PCI/I-486SP3G M/B, BIOS ver. 0304.

Update NCR BIOS to 3.07.00.
Support Diamond Stealth 64 Video VRAM.

It really depends on which BIOS ROM that was used. For any one model of 486 board, the makers could have used an EEPROM or an EPROM. The surest way to determine this is to a) look at the jumpers on the motherboard to see if it is set for EEPROM or EPROM, or b) look at the part number on the BIOS ROM chip and google-it.

It was a wise decision to benchmark the PCI/I-486SP3G. Theoretically, the memory throughput should be twice as fast as the other 486 boards you are playing with, however it is my suspicion that additional cycles are inserted which counteract the effectiveness of memory interleaving. For whatever stable/optimal memory/cache wait states in CMOS you come up with, please share the 3DBench, Doom, Quake, Speedsys, and Cachechk results.

Should have the benchmarks going this weekend. I'll update the thread with my PCI/I-486SP3G findings.

Artex wrote:

Should have the benchmarks going this weekend. I'll update the thread with my PCI/I-486SP3G findings.

Sorry guys- I got stuck doing the 'honey do' list this weekend but I'll try to get this board benchmarked later this week or over the weekend.

Artex wrote:

bristlehog wrote:

I have got a STG G540 programmer, what exactly you want to burn, Artex?

Hey Bristlehog - It sounds like I need to purchase a M27C512 EPROM Chip and have it programmed with the latest BIOS for my Asus VL/I-486SV2GX4 Rev. 2.1 board. I believe it's the 0402 bios.

I'd say a Winbond W27C512 or W27E512 should work fine and be good EEPROM replacements for those old M27C512 EPROMS.

I've been successful replacing a CM-32L control ROM with the EEPROMS above. The original ROM was a AM27C512.

Artex wrote:

Sorry guys- I got stuck doing the 'honey do' list this weekend but I'll try to get this board benchmarked later this week or over the weekend.

Any chance you could prioritize benching of the S3 vs ET4000/W32p cards when you get the time?

Yup. Picking up a 22 inch Mitsubishi Diamondtron tomorrow then benchies will ensue. Sorry for the delay. Warmer weather means an expanded 'honey do' list for me. :p

Artex wrote:

Yup. Picking up a 22 inch Mitsubishi Diamondtron tomorrow then benchies will ensue. Sorry for the delay. Warmer weather means an expanded 'honey do' list for me. :p

Great 😀 Not many people have benched VLB cards up against one another on the same system.The only lists I know about are on thandor.net and this one in Hungarian:
http://logout.hu/cikk/vesa_buszos_vga_kartyak … /bevezetes.html

The Thandor one says that a 1MB W32p is equal to a S3 Vision864 1MB on a DX2/66, but I wonder if the CPU put some limitations here.

Atleast SiS471 wins as the quickest VLB chipset: http://logout.hu/cikk/socket3_vlb_platformtes … rat/teljes.html

Ok, here we go! I benched three VLB cards today on my Asus VL/I-486SV2GX4 Rev. 2.0 (SiS 471-407) with Am486 DX4-100 (Write Through Mode). I'm currently awaiting some freshly flashed EPROM chips to arrive with the latest Asus BIOS so that I can run the processor in Write-Back mode. I'll also test my Intel DX4-100 (80486DX4100) (SK096) in Write-Back mode to see how it compares. For now I used two Tseng ET4000/W32P-based cards and one S3 Vision 864 card. I was going to try my Cirrus Logic 5428 card but the system won't boot with it.

AMD Am486 DX4-100 (A80486DX4-100SV88)
ASUS VL/I-486SV2GX4 Rev. 2.0
16MB RAM (FPM)
256KB L2 Cache
CF Card on VLB Promise IDE Controller

The Test Bench

SuperScape
S3 Vision 864 - #9 (9GXE64) Revolution Ticket to Ride (1MB): 635
Tseng ET4000 W32P - Diamond Stealth 32 (2MB): 635
Tseng ET4000 W32P - STB Lightspeed 50/50 (2MB): 635

PC Player Bench
S3 Vision 864 - #9 (9GXE64) Revolution Ticket to Ride (1MB): 13.4
Tseng ET4000 W32P - Diamond Stealth 32 (2MB): 13.4
Tseng ET4000 W32P - STB Lightspeed 50/50 (2MB): 13.4

DOOM
S3 Vision 864 - #9 (9GXE64) Revolution Ticket to Ride (1MB): 2108 realtics
Tseng ET4000 W32P - Diamond Stealth 32 (2MB): 2091 realtics
Tseng ET4000 W32P - STB Lightspeed 50/50 (2MB): 2090 realtics

Quake
S3 Vision 864 - #9 (9GXE64) Revolution Ticket to Ride (1MB): 8.6 fps
Tseng ET4000 W32P - Diamond Stealth 32 (2MB): 8.7 fps
Tseng ET4000 W32P - STB Lightspeed 50/50 (2MB): 8.6 fps

S3 Vision 864 - #9 (9GXE64) Revolution "Ticket to Ride" (1MB)

Tseng ET4000 W32P - Diamond Stealth 32 (2MB)

Tseng ET4000 W32P - STB Lightspeed 50/50 (2MB)

feipoa wrote:

It really depends on which BIOS ROM that was used. For any one model of 486 board, the makers could have used an EEPROM or an EPROM. The surest way to determine this is to a) look at the jumpers on the motherboard to see if it is set for EEPROM or EPROM, or b) look at the part number on the BIOS ROM chip and google-it.

It was a wise decision to benchmark the PCI/I-486SP3G. Theoretically, the memory throughput should be twice as fast as the other 486 boards you are playing with, however it is my suspicion that additional cycles are inserted which counteract the effectiveness of memory interleaving. For whatever stable/optimal memory/cache wait states in CMOS you come up with, please share the 3DBench, Doom, Quake, Speedsys, and Cachechk results.

I didn't forget about you! I will fire up this board later today or tomorrow to get it benched as well.

Anonymous Coward wrote:

sliderider wrote:

Top Vl-Bus video card to me is Mach64 4mb, but good luck finding one. A Mach32 will be easier to find and cheaper and give you most of the performance, so represents a much better value.

What's the appeal there exactly? It's rather so so in DOS, and the extra 2MB in windows is borderline useless.

Just saw this but I disagree. 2MB vs 4MB is important. If the chipset supports it, 4 MB opens up 1024x768x24/32 and 1280x1024x16

4 to 6MB probably won't gain much but 2MB is bordering on too little for a late 486 system

Artex wrote:

I didn't forget about you! I will fire up this board later today or tomorrow to get it benched as well.

Great! Could you compare the Asus PCI/I-486SP3G against the Biostar MB-8433UUD?

feipoa wrote:

Artex wrote:

I didn't forget about you! I will fire up this board later today or tomorrow to get it benched as well.

Great! Could you compare the Asus PCI/I-486SP3G against the Biostar MB-8433UUD?

You bet. Anything specific benchmark wise?

First determine fastest stable memory and cache timings. Then test some standard titles like:

DOS
Norton Sysinfo
Speedsys
Cachechk [Read: cachechk -t6 -d], [Write: cachechk -t6 -d -w]
3Dbench
Doom Timedemo3
Pcpbench
Quake timedemo1 @320200 and @640x480

Windows98SE
Ziff-Davis CPUMark99 (stand-alone version)
Ziff-Davis WinBench99 FPU Mark
WinTune98
Sandra99
GLQuake @640x480 or less
Quake II OpenGL @640x480 or less

For Quake II in OpenGL to work, you need a Diamond RIVA TNT, Voodoo 2/3/Banshee, or ATI Rage 128 VR. If you have a Matrox G200, you can use driver version 5.06 or 5.07 in Windows NT 4.0. I do not recall which graphic cards work with GLQuake, presumably more than this.

Test whatever you can.

Holy cow! I'll need some time... Right now both setups are DOS only.

Since you are comparing the two motherboards to each other directly, you could also use Win95 if it is more convenient. Perhaps using Win95 is preferred as others members have suggested that it might yield better results than Win98. I've yet to verify this hypothesis with benchmark data though. If you only have 256KB cache, I suggest using either 64 MB of memory with L2 cache set to write-through, or 32 MB of memory with L2 set to write-back. For the PCI/I-486SP3G, Anonymous Coward noted that you need to fill all the memory slots to witness the benefit of memory interleaving.

Cool benchmarks there!

With the DX4, I find that the IntelDX4 is the way to go if you want 100 MHz. It is, clock for clock, faster compared to the AMD 5x86. Even on the latest 486 boards. On older boards the IntelDX4 is likely to have better / more support.

The 5x86 133 is faster, but not by much. It will certainly not make lagging games smooth.

Yeah, I think I'm gonna hold off on some of the benchmarks on the ASUS VL/I-486SV2GX4 until I get the new EPROM chips that allow me to run these processors in write-back mode. Those should hopefully be here within a couple of weeks. I'm gonna switch gears for now and start benchmarking the Asus PCI/I-486SP3G against the Biostar MB-8433UUD. I haven't powered up the Asus board to see what its BIOS revision looks like for processor support.

Artex wrote:

Yeah, I think I'm gonna hold off on some of the benchmarks on the ASUS VL/I-486SV2GX4 until I get the new EPROM chips that allow me to run these processors in write-back mode. Those should hopefully be here within a couple of weeks. I'm gonna switch gears for now and start benchmarking the Asus PCI/I-486SP3G against the Biostar MB-8433UUD. I haven't powered up the Asus board to see what its BIOS revision looks like for processor support.

Seems like a good idea. Thanks though for providing the VLB benches. Cool to see that the later S3 Vision 864 (and most likely 964) is on par with the W32p. Great to know it's an alternative!