Next step - to clean the beast and add some SCSI!

Original Sony IDE CDROM do not read discs (i will try to recover it later), but my destination is SCSI. Motherboard have BIOS support for NCR 53C810 PCI SCSI cards so... try to install this card with Toshiba XM5301 SCSI CDROM from 1994!

Cache upgrade and some BIOS settings.

Originally motherboard was equiped with 128KB L2 cache (perpect size for 386DX systems, but not for 486DX2).

Upgrade it up to 256KB (motherboard supports 512-1MB size if you have such chips).
Play with L2 cache strategy - Write-Back gives 3-5% CPU speedup, but miss the 15-20% of RAM speed. Chose the Write -Back for CPU L1 cache and Write-Thru for L2 cache (get the 44MB/sec memory bandwidth and 24.5MB memory throughput in SPEEDSYS 4.78).

Babasha wrote on 2024-09-10, 19:11:

Upgrade it up to 256KB (motherboard supports 512-1MB size if you have such chips).
Play with L2 cache strategy - Write-Back gives 3-5% CPU speedup, but miss the 15-20% of RAM speed. Chose the Write -Back for CPU L1 cache and Write-Thru for L2 cache (get the 44MB/sec memory bandwidth and 24.5MB memory throughput in SPEEDSYS 4.78).

Maybe your BIOS has 7+1 tag mode?
With write-through supported, I wouldn't bother with modifying the board to add a Dirty RAM, though.

For WB CPUs like Intel DX4-100 (WB) you need the latest BIOS (4.27). This BIOS has the 7+1 TAG option, but as soon as L1 WB is working it disables the dirty TAG bit - hence the performance drop. Same goes for Am5x86 btw. (BIOS 4.26 and it´s ESCOM variant also don´t do the trick.)

Overclocking to 40 MHz (or 50 MHz) makes no sense also, because the BIOS increases waitstates heavily as soon as you leave the 33MHz zone. CPU performance goes up, memory and cache performance are still disappointing.
I have tried some alternative BIOSes (there are only very few around) and Dirty TAG was working. But those BIOS versions lacked support for PCI, Super-IO, etc.).
I´m sure this is just some tiny BIOS flaw that could be fixed...

majestyk wrote on 2024-09-11, 05:29:

For WB CPUs like Intel DX4-100 (WB) you need the latest BIOS (4.27). This BIOS has the 7+1 TAG option, but as soon as L1 WB is w […]
Show full quote

For WB CPUs like Intel DX4-100 (WB) you need the latest BIOS (4.27). This BIOS has the 7+1 TAG option, but as soon as L1 WB is working it disables the dirty TAG bit - hence the performance drop. Same goes for Am5x86 btw. (BIOS 4.26 and it´s ESCOM variant also don´t do the trick.)

Overclocking to 40 MHz (or 50 MHz) makes no sense also, because the BIOS increases waitstates heavily as soon as you leave the 33MHz zone. CPU performance goes up, memory and cache performance drop far below the 33MHz scores.
I have tried some alternative BIOSes (there are only very few around) and Dirty TAG was working. But those BIOS versions lacked support for PCI, Super-IO, etc.).
I´m sure this is just some tiny BIOS flaw that could be fixed...

Tnx! I read about the 4.27 and beta BIOSes will try to burn new ROM chip with 4.27
33MHz bus is my ideal clock! I dont like 40-50MHz and overclocking.

At 33 MHz it´s a rockstable and very versatile system.
My speedsys results are:
Intel DX4-100 WT: 66.66 / 43.9 / 23.9
Intel DX4-100 WB: 80.84 / 28.02 / 13.57
Just for the record - with an Intel DX4-100WB @ 120 it´s 97.46 / 35.40 / 17.21(!)

Stting cache timings to turbo, fast or medium makes no difference at all with BIOS 4.27.

It´s a pity, since with dirty-TAG bit enabled L2 and RAM thrroughput would look far better.

With the beta-BIOS there´s no PCI support and L2 cache sizes larger than 256K are not detected correctly which makes this "VIP" mainboard a joke.

majestyk wrote on 2024-09-11, 07:54:

At 33 MHz it´s a rockstable and very versatile system. My speedsys results are: Intel DX4-100 WT: 66.66 / 43.9 / 23.9 Intel DX4- […]
Show full quote

At 33 MHz it´s a rockstable and very versatile system.
My speedsys results are:
Intel DX4-100 WT: 66.66 / 43.9 / 23.9
Intel DX4-100 WB: 80.84 / 28.02 / 13.57
Just for the record - with an Intel DX4-100WB @ 120 it´s 97.46 / 35.40 / 17.21(!)

Stting cache timings to turbo, fast or medium makes no difference at all with BIOS 4.27.

It´s a pity, since with dirty-TAG bit enabled L2 and RAM thrroughput would look far better.

With the beta-BIOS there´s no PCI support and L2 cache sizes larger than 256K are not detected correctly which makes this "VIP" mainboard a joke.

Tnx! Set the Write-Back - Enable and Combine Alter an Tag bits - Enable and its the best and fastest for me.

What BIOS version is it? Can you run "ctcm" to see weather dirty TAG is active or not?

When I jumper my DX4 for 66 MHz (2 x 66) dirty TAG IS activated, but as soon as it runs at 2 x 40, 3 x 33 or 3 x 40 it gets turned off.

majestyk wrote on 2024-09-11, 11:20:

What BIOS version is it? Can you run "ctcm" to see weather dirty TAG is active or not?

When I jumper my DX4 for 66 MHz (2 x 66) dirty TAG IS activated, but as soon as it runs at 2 x 40, 3 x 33 or 3 x 40 it gets turned off.

The attachment 2x66vip_io.JPG is no longer available

My BIOS is 4.26GN2(ES2) 12/07/94
Ha! Need to save it and upload to Theretroweb.

Will check cache with ctcm next day.

Sound card (perriod) correction.
Change the Aztec BX II ISA8 (nice old sound card, but more correct for 286/386 systems) with Aztech Sound Galaxy NX II Extra ISA16 sound.
New card compatible with Soundblaster 2.5, Adlib, Covox and Disney Sound Source.
Unusual listen to Duke3D where the digital sound is Covox and Adlib music.

majestyk wrote on 2024-09-11, 11:20:

What BIOS version is it? Can you run "ctcm" to see weather dirty TAG is active or not?

When I jumper my DX4 for 66 MHz (2 x 66) dirty TAG IS activated, but as soon as it runs at 2 x 40, 3 x 33 or 3 x 40 it gets turned off.

The attachment 2x66vip_io.JPG is no longer available

CTCM test - done!)))

New BIOS V. 4.27 on its way to me and 2nd MB of video RAM. Hope install they this week.
Only thing to be done is to investigate secondary IDE channel "glitch".
I think 486-VIP-IO IDE channels not equal. PRIMARY IDE chanel is "VESA IDE" connected to VLB BUS VT82C486G chip (support only TWO IDE devices with high speed).
SECONDARY IDE channel connected to ISA BUS with support of less-performance devices support (CDROM etc). Its a usual scheme for early PRI/SEC IDE controllers where high-performance IDE devices connected to PRIMARY channel and slow devices uses SECONDARY channel. Or... SECONDARY IDE channel is ATAPI-protocol only without correct support of ATA IDE HHDs?

Sweet build 😃

Case is a dream come true - love the design.

VRAM and BIOS updates time.

Burn the new BIOS ver. 4.27 and check it - no speed changes, new BIOS settings RAM PARITY (i dont need) and new CPU name during POST (P24D-S 66MHz).
Check the SECONDARY CHANNEL IDE - no news, so its an ATAPI (CDROM) devices only.
Will return original BIOS ver. 4.26GN (ES2).

Upgrade videocard with additional 1MB - nice)))

Babasha wrote on 2024-09-09, 23:23:

Video: Avance Logic 2301 PCI 1MB (no drivers, VGA-only 640x480 in 16 colors)

This early PCI graphics card is known to be budget optimized.
I've zipped my driver directory for you.

Lets look closer to SECONDARY IDE CHANNEL and its "glitch" when HHD's detected in BIOS but not in DOS.
My idea was - its a realization or BIOS limitation (can not say anything about BIOS).
Lets start PowerIDE! diag...

Yeap!
PRIMARY IDE CHANNEL is VLB (controlled by main chipset VT82C486G chip).
SECONDARY IDE CHANNEL is ISA (controlled by external chip). On this channel you can use ATAPI devices thru DOS drivers (CDROMs, ZIPs...). Maybe this is concept idea of motherboard to use it only with ATAPI devices but not with HDDs?
Tnx God we have DOS driver for HDD's in Adaptec PowerIDE!

Lets edit CONFIG.SYS and reboot!
Ahahaha! After that PowerIDE! successfully detect SECONDARY IDE channel and HDD on it in DOS (fdisk, format and other utilities work with it). Check it copy/move/create/delete operations wighout any fails)))
And continue with Windows 95...
Install PowerIDE! Win95 driver. I DO NOT remove original PRIMARY standard ESDI/IDE driver just install PowerIDE! driver, reboot... and my Win95 completely fail((( (no HDD controllers found after reboot).
I look in .inf file and find its removes HDD controllers from registry before install. Slighly change the .inf file (delete some lines with HDD controllers remove). Reinstall it once more and MIRACLE!!!)))

So the PRIMARY IDE (IRQ14) channel works as standard ESDI/IDE controller with Microsoft driver and SECONDARY IDE (IRQ15) works with Adaptec PowerIDE driver as virtual SCSI device.
Made a quick and dirty test... will continue it later.

PS. Patched PowerIDE! in attach 😉

Final test and memory optimization.

Speedsys final scores in pure DOS.

Install QEMM97 and got 624KB of free DOS memory with all drivers loaded IDE with IDE ASPI, SCSI CAM, SCSI CDROM, MSCDEX and other Microsoft stuff (setver, ifshlp, dblbuff, country and display/mode localizations).
Not bad as for me.

majestyk wrote on 2024-09-11, 07:54:

At 33 MHz it´s a rockstable and very versatile system. My speedsys results are: Intel DX4-100 WT: 66.66 / 43.9 / 23.9 Intel DX4- […]
Show full quote

At 33 MHz it´s a rockstable and very versatile system.
My speedsys results are:
Intel DX4-100 WT: 66.66 / 43.9 / 23.9
Intel DX4-100 WB: 80.84 / 28.02 / 13.57
Just for the record - with an Intel DX4-100WB @ 120 it´s 97.46 / 35.40 / 17.21(!)

Stting cache timings to turbo, fast or medium makes no difference at all with BIOS 4.27.

It´s a pity, since with dirty-TAG bit enabled L2 and RAM thrroughput would look far better.

With the beta-BIOS there´s no PCI support and L2 cache sizes larger than 256K are not detected correctly which makes this "VIP" mainboard a joke.

Some success with dirty BIOS hack)))

FIC 486-VIP-IO try to patch BIOS 4.27