Something new about OPTI chipset compatibility

Partially success with BIOSes from Biostar MB-1575PCT

It gives POST codes and BIOS messages on screen with ISA videocard, but freezes if if press DEL to enter BIOS SETUP

I've organized the news of my "project" on the Spring Circle SF586 motherboard.

Today 2MB of cache arrived from China.

The cache isn't 10ns, but 15ns. Chips are remarked but all are working.

It overclocks worse than my previous 512KB Winbond chips.

Well, trading speed for capacity and stability. Previous 512KB I had overclocked to 12.8ns but on the edge of stability (the floppy disk was already starting to fail). With this new cache, it only takes near 15ns but it's stable and faster overall.

CTCM's with 2MB of cache is driving me nuts - it says there's no L2 cache.

And CACHECHK thinks it's 15ns RAM.

The new cache started at 10-11 ns. But the test hangs.

I read the chipset documentation, there's also a catch with cache interleaving.

Like, if you want higher speeds, set it to 1 MB; if you set it to 2 MB, it disables cache interleaving.

Okay, it's a little faster and more stable.

And my favorite section: "Turbo Mode Operation"

Motherboard can operate in the following modes:

1) mode with the Turbo button pressed (CPU at 100 MHz and everything at maximum settings) - speed is naturally equal to a P100 computer.

2) mode with the Turbo button released (CPU at 50 MHz and everything slows down to 50 MHz, although it's slow, but the L1 and L2 caches are present) - speeds below 5x86 133 MHz and even below the Pentium 60 MHz

3) When I press CTRL ALT - on the keyboard (the CPU reports 100 MHz, disables all L1 and L2 caches, and apparently adds waits) - the speed is about the same as a 386DX40 without L2 cache.

While writing this, I thought I should try simultaneously depressing the Turbo button and pressing CTRL ALT - on the keyboard. I tried it - according to SPEEDSYS, the speed dropped to the level of some 286 computer.

The current configuration:

1) SPRING CIRCLE SF586 motherboard with 2048 KB cache

2) INTEL Pentium-S 100 MHz processor

3) RAM: 2 x 32 MB TRANSCEND SIMM 72-pin FPM 60 nS

4) Pine PT-627B VLB multi-card

5) DIAMOND STEALTH 2000 PRO 4 MB PCI graphics card

6) Future Domain TMC-810 PCI SCSI controller

I boot via IDE from a 256 MB CF card and use 1 GB IBM SCSI hard drive.

P.S. Yesterday, I used the AWARD BIOS editor to slightly tweak my existing stable BIOS (open hidden PCI controls and adjust things up for usability).
I'll try to fix the Y2K issue. If it work, I'll publish the firmware on TRW.

Waiting for a VLB video card and a 40 MHz crystal generator to continue and study how things work with them.

I had time and tested all the turbo combinations.

50MHz mode - I selected CTRL ATL - with the buttons - 4.2 fps in 3dbench
50MHz mode - I selected CTRL ATL + with the buttons - 27.7 fps in 3dbench
100MHz mode - I selected CTRL ATL - with the buttons - 8.5 fps in 3dbench
100MHz mode - I selected CTRL ATL + with the buttons - 55.5 fps in 3dbench

Basically, the machine runs as 286/12, 386sx/20, P1/50 (which is about 486/100), P1/100

For the truly hardcore, you can enable 25MHz FSB mode. I think you can even reach XT 😉

AND OF COURSE I TRIED IT)))

25MHz mode - I selected CTRL ATL - with the buttons - 2.4 fps in 3dbench (should we consider this something like the TurboXT?)
25MHz mode - I selected CTRL ATL + with the buttons - 16.3 fps in 3dbench (is that more like 386DX/40?)

Bonus - patched BIOS for SF586 (Y2K BIOS bug fix, LBA mode up to 8GB limit)

Babasha wrote on 2025-12-26, 16:51:

Bonus - patched BIOS for SF586 (Y2K BIOS bug fix, LBA mode up to 8GB limit)

What BIOS version is this based on?

My SF586 is in the mail and I'll be experimenting with it soon.

Beerfloat wrote on 2025-12-26, 17:34:

Babasha wrote on 2025-12-26, 16:51:

Bonus - patched BIOS for SF586 (Y2K BIOS bug fix, LBA mode up to 8GB limit)

What BIOS version is this based on?

My SF586 is in the mail and I'll be experimenting with it soon.

its based on 4.50G - VIP3A-ZS1 - https://theretroweb.com/motherboard/bios/sf58 … 9b420397457.zip

meanwhile 4.50G - (P3B01A) - https://theretroweb.com/motherboard/bios/fore … d5800131675.zip
is got some pros (more PCI options and supports 4 IDE drives detection) and cons (its not so fast and compatible)

I have been wondering.
Is this Forex chipset a renamed OPTi Python or Cobra chipset?

bracecomputerlab wrote on 2025-12-29, 02:43:

I have been wondering.
Is this Forex chipset a renamed OPTi Python or Cobra chipset?

I'm pretty sure it is not. Both the Opti Cobra and Python chipsets feature 208-pin PQFP AT bus controllers (the 82C596 and 82C546 respectively), with their other components being 160-pin PQFP system controllers (82C597, 82C547), a 100-pin integrated peripheral controller (82C206, various brands), and optionally a 208-pin VLB to PCI bridge (the Opti 82C822).

On the other hand, the Forex chipset found on the SF586 consists of 3 160-pin PQFP chips along with the 100-pin 82C206. This can't be Cobra or Python.

As far as matching Opti chipsets go the Forex 58C601/58C602 duo do share package sizes with the Opti 486/Pentium writeback chipset (82C571 with 82C572, both 160-pin).
That's a 1993 chipset which could support either 486 or P60/P66 Pentium CPUs, flexibility that comes by virtue of being limited to a 32-bit memory bus.
As such, it could be found on some of those entertaining Socket 3/4 VLB combo boards like the TMC/MyComp/MyNix/Megastar PAT45PV and the Acer Vi7, which I would love to have also.

Someone must've been stuck with some chip surplus because the 82C571/82C572 also ended up on a few Socket 5/3.3v P54C boards, including the TMC PCI54PL and PCI58PL, where they are accompanied by Opti's venerable 82C822 VLB to PCI Bridge.

While the Forex 82c601/82c602 could be a match for the Opti 82c571/82c572, the Forex 58C613 can not be an Opti 82c822 because it has too few pins - 160 instead of 208.
But it might just be a VIA VT82C505 VLB to PCI bridge instead. That's a 160-pin IC that can be interfaced to non-VIA chipsets according to VIA's notes.

Just thinking out loud. I should have my board in a few days and look forward to finding out more.

Beerfloat wrote on 2025-12-29, 13:08:

I'm pretty sure it is not. Both the Opti Cobra and Python chipsets feature 208-pin PQFP AT bus controllers (the 82C596 and 82C54 […]
Show full quote

bracecomputerlab wrote on 2025-12-29, 02:43:

I have been wondering.
Is this Forex chipset a renamed OPTi Python or Cobra chipset?

I'm pretty sure it is not. Both the Opti Cobra and Python chipsets feature 208-pin PQFP AT bus controllers (the 82C596 and 82C546 respectively), with their other components being 160-pin PQFP system controllers (82C597, 82C547), a 100-pin integrated peripheral controller (82C206, various brands), and optionally a 208-pin VLB to PCI bridge (the Opti 82C822).

On the other hand, the Forex chipset found on the SF586 consists of 3 160-pin PQFP chips along with the 100-pin 82C206. This can't be Cobra or Python.

As far as matching Opti chipsets go the Forex 58C601/58C602 duo do share package sizes with the Opti 486/Pentium writeback chipset (82C571 with 82C572, both 160-pin).
That's a 1993 chipset which could support either 486 or P60/P66 Pentium CPUs, flexibility that comes by virtue of being limited to a 32-bit memory bus.
As such, it could be found on some of those entertaining Socket 3/4 VLB combo boards like the TMC/MyComp/MyNix/Megastar PAT45PV and the Acer Vi7, which I would love to have also.

Someone must've been stuck with some chip surplus because the 82C571/82C572 also ended up on a few Socket 5/3.3v P54C boards, including the TMC PCI54PL and PCI58PL, where they are accompanied by Opti's venerable 82C822 VLB to PCI Bridge.

While the Forex 82c601/82c602 could be a match for the Opti 82c571/82c572, the Forex 58C613 can not be an Opti 82c822 because it has too few pins - 160 instead of 208.
But it might just be a VIA VT82C505 VLB to PCI bridge instead. That's a 160-pin IC that can be interfaced to non-VIA chipsets according to VIA's notes.

Just thinking out loud. I should have my board in a few days and look forward to finding out more.

You are right its not Cobra or Python Pentium chipset but a something OPTI related - some BIOS'es for Cobra or Python partially works with SF586
I think its OPTI 486/PODP chipset clone or derivative made for Forex
With its 32bit memory access and settings for 486 inside BIOS... i think so

Babasha wrote on 2025-12-29, 13:18:

You are right its not Cobra or Python Pentium chipset but a something OPTI related - some BIOS'es for Cobra or Python partially […]
Show full quote

Beerfloat wrote on 2025-12-29, 13:08:

I'm pretty sure it is not. Both the Opti Cobra and Python chipsets feature 208-pin PQFP AT bus controllers (the 82C596 and 82C54 […]
Show full quote

bracecomputerlab wrote on 2025-12-29, 02:43:

I have been wondering.
Is this Forex chipset a renamed OPTi Python or Cobra chipset?

I'm pretty sure it is not. Both the Opti Cobra and Python chipsets feature 208-pin PQFP AT bus controllers (the 82C596 and 82C546 respectively), with their other components being 160-pin PQFP system controllers (82C597, 82C547), a 100-pin integrated peripheral controller (82C206, various brands), and optionally a 208-pin VLB to PCI bridge (the Opti 82C822).

On the other hand, the Forex chipset found on the SF586 consists of 3 160-pin PQFP chips along with the 100-pin 82C206. This can't be Cobra or Python.

As far as matching Opti chipsets go the Forex 58C601/58C602 duo do share package sizes with the Opti 486/Pentium writeback chipset (82C571 with 82C572, both 160-pin).
That's a 1993 chipset which could support either 486 or P60/P66 Pentium CPUs, flexibility that comes by virtue of being limited to a 32-bit memory bus.
As such, it could be found on some of those entertaining Socket 3/4 VLB combo boards like the TMC/MyComp/MyNix/Megastar PAT45PV and the Acer Vi7, which I would love to have also.

Someone must've been stuck with some chip surplus because the 82C571/82C572 also ended up on a few Socket 5/3.3v P54C boards, including the TMC PCI54PL and PCI58PL, where they are accompanied by Opti's venerable 82C822 VLB to PCI Bridge.

While the Forex 82c601/82c602 could be a match for the Opti 82c571/82c572, the Forex 58C613 can not be an Opti 82c822 because it has too few pins - 160 instead of 208.
But it might just be a VIA VT82C505 VLB to PCI bridge instead. That's a 160-pin IC that can be interfaced to non-VIA chipsets according to VIA's notes.

Just thinking out loud. I should have my board in a few days and look forward to finding out more.

You are right its not Cobra or Python Pentium chipset but a something OPTI related - some BIOS'es for Cobra or Python partially works with SF586
I think its OPTI 486/PODP chipset clone or derivative made for Forex
With its 32bit memory access and settings for 486 inside BIOS... i think so

Oh yeah I've been reading your investigations with great interest. And dionb's as well. Really nice.

For all I know the Forex chipset could still be its own thing. I agree that it likely isn't.

If it were really derived from the Opti 82C571/572 then some things start to make sense:
- The board working at least partially with these BIOSes from a near relative Opti chipset
- dionb's board booting with a single SIMM, a sign of a 32-bit memory bus (though there was also a SiS 64-bit capable chipset that could boot a Pentium in 'half bank' mode)
- 128MB mem is officially supported by the chipset
- The 571/572 supports a maximum of 1MB L2 cache so we could pretty much stop trying to get 2048KB to work
- The chipset really only supports VLB at a fixed 33Mhz, asynchronous with the front side bus
Both 60 and 66Mhz FSB CPUs get 33Mhz VLB. Other bus speeds were unknown when the chipset came out and are unpredictable. Cobra was the same. Python introduced flexible VLB.
- Although 32-bit memory is bad the Pentium still retains a 64-bit data path to L2 cache, which should help
- We should try some 571/572 board BIOSes

Beerfloat wrote on 2025-12-29, 15:02:

Oh yeah I've been reading your investigations with great interest. And dionb's as well. Really nice. […]
Show full quote

Babasha wrote on 2025-12-29, 13:18:

You are right its not Cobra or Python Pentium chipset but a something OPTI related - some BIOS'es for Cobra or Python partially […]
Show full quote

Beerfloat wrote on 2025-12-29, 13:08:

I'm pretty sure it is not. Both the Opti Cobra and Python chipsets feature 208-pin PQFP AT bus controllers (the 82C596 and 82C54 […]
Show full quote

I'm pretty sure it is not. Both the Opti Cobra and Python chipsets feature 208-pin PQFP AT bus controllers (the 82C596 and 82C546 respectively), with their other components being 160-pin PQFP system controllers (82C597, 82C547), a 100-pin integrated peripheral controller (82C206, various brands), and optionally a 208-pin VLB to PCI bridge (the Opti 82C822).

On the other hand, the Forex chipset found on the SF586 consists of 3 160-pin PQFP chips along with the 100-pin 82C206. This can't be Cobra or Python.

As far as matching Opti chipsets go the Forex 58C601/58C602 duo do share package sizes with the Opti 486/Pentium writeback chipset (82C571 with 82C572, both 160-pin).
That's a 1993 chipset which could support either 486 or P60/P66 Pentium CPUs, flexibility that comes by virtue of being limited to a 32-bit memory bus.
As such, it could be found on some of those entertaining Socket 3/4 VLB combo boards like the TMC/MyComp/MyNix/Megastar PAT45PV and the Acer Vi7, which I would love to have also.

Someone must've been stuck with some chip surplus because the 82C571/82C572 also ended up on a few Socket 5/3.3v P54C boards, including the TMC PCI54PL and PCI58PL, where they are accompanied by Opti's venerable 82C822 VLB to PCI Bridge.

While the Forex 82c601/82c602 could be a match for the Opti 82c571/82c572, the Forex 58C613 can not be an Opti 82c822 because it has too few pins - 160 instead of 208.
But it might just be a VIA VT82C505 VLB to PCI bridge instead. That's a 160-pin IC that can be interfaced to non-VIA chipsets according to VIA's notes.

Just thinking out loud. I should have my board in a few days and look forward to finding out more.

You are right its not Cobra or Python Pentium chipset but a something OPTI related - some BIOS'es for Cobra or Python partially works with SF586
I think its OPTI 486/PODP chipset clone or derivative made for Forex
With its 32bit memory access and settings for 486 inside BIOS... i think so

Oh yeah I've been reading your investigations with great interest. And dionb's as well. Really nice.

For all I know the Forex chipset could still be its own thing. I agree that it likely isn't.

If it were really derived from the Opti 82C571/572 then some things start to make sense:
- The board working at least partially with these BIOSes from a near relative Opti chipset
- dionb's board booting with a single SIMM, a sign of a 32-bit memory bus (though there was also a SiS 64-bit capable chipset that could boot a Pentium in 'half bank' mode)
- 128MB mem is officially supported by the chipset
- The 571/572 supports a maximum of 1MB L2 cache so we could pretty much stop trying to get 2048KB to work
- The chipset really only supports VLB at a fixed 33Mhz, asynchronous with the front side bus
Both 60 and 66Mhz FSB CPUs get 33Mhz VLB. Other bus speeds were unknown when the chipset came out and are unpredictable. Cobra was the same. Python introduced flexible VLB.
- Although 32-bit memory is bad the Pentium still retains a 64-bit data path to L2 cache, which should help
- We should try some 571/572 board BIOSes

As i remember i already try all 571/572 BIOS-es with same success - they partially works but with ISA video only

All of which reminds me I also need to take time to try out and compare the board below.

It's socket 4 not 5 but should give some indication how much the potential 32-bit memory access on Forex 58C601/58C602 (Opti 571/572?) really holds it back vs 596/597 Cobra.
Might not turn out to be a huge deal in practice.

Beerfloat wrote on 2025-12-29, 17:40:

All of which reminds me I also need to take time to try out and compare the board below. […]
Show full quote

All of which reminds me I also need to take time to try out and compare the board below.

It's socket 4 not 5 but should give some indication how much the potential 32-bit memory access on Forex 58C601/58C602 (Opti 571/572?) really holds it back vs 596/597 Cobra.
Might not turn out to be a huge deal in practice.

The attachment air_586mi_1.11_sm.jpg is no longer available

As i can see on most users screenshots of SPEEDSYS of OPTI Cobra and Python chipsets they are faster in memory speed over Forex SF586

Babasha wrote on 2025-12-29, 19:01:

Beerfloat wrote on 2025-12-29, 17:40:

All of which reminds me I also need to take time to try out and compare the board below. […]
Show full quote

All of which reminds me I also need to take time to try out and compare the board below.

It's socket 4 not 5 but should give some indication how much the potential 32-bit memory access on Forex 58C601/58C602 (Opti 571/572?) really holds it back vs 596/597 Cobra.
Might not turn out to be a huge deal in practice.

The attachment air_586mi_1.11_sm.jpg is no longer available

As i can see on most users screenshots of SPEEDSYS of OPTI Cobra and Python chipsets they are faster in memory speed over Forex SF586

Synthetic benchmarks will undoubtedly suck. But let's get some more game results.

I'm also hoping to find/cobble together Pentium MMX and K6-2/3 BIOS support with a voltage interposer.

My board arrived and I'll be playing around with it in the next few days. The previous owner soldered a BF0/BF1 mod switch onto it.
I'll have to take a closer look at these cache chips because other pictures I've seen use a different configuration.

Beerfloat wrote on 2026-01-03, 19:41:

My board arrived and I'll be playing around with it in the next few days. The previous owner soldered a BF0/BF1 mod switch onto it.
I'll have to take a closer look at these cache chips because other pictures I've seen use a different configuration.

I see your board on eBay twice(?)
Last time in .bg seller list - there is no cache tag chip so l2 cache no go
And early in .de seller list - he use it with 2mb l2 cache and set up with pentium 200mhz (non-mmx) and idt winchip 200mhz

Babasha wrote on 2026-01-03, 20:36:

I see your board on eBay twice(?) Last time in .bg seller list - there is no cache tag chip so l2 cache no go And early in .de […]
Show full quote

Beerfloat wrote on 2026-01-03, 19:41:

My board arrived and I'll be playing around with it in the next few days. The previous owner soldered a BF0/BF1 mod switch onto it.
I'll have to take a closer look at these cache chips because other pictures I've seen use a different configuration.

I see your board on eBay twice(?)
Last time in .bg seller list - there is no cache tag chip so l2 cache no go
And early in .de seller list - he use it with 2mb l2 cache and set up with pentium 200mhz (non-mmx) and idt winchip 200mhz

Haha yeah you know more about this board than I do.
This one indeed came in from Bulgaria. Not a few of the boards I'm interested in seem to end up coming from Eastern Europe.

200 Mhz non-MMX confirmed working is good as I should have a couple of those somewhere.
Thanks for the headsup on the tag chip, I should have a few spares around.

Test my SF586 with DIAMOND STEALTH 32VL VER.B and with external quartz generator 40MHz

Some results:

1) CPU and cache/RAM always works with internal 33MHz oscilator - just ok
2) VLB and PCI buses can work in asynchonous mode with 40MHz clock - good
3) VLB stable at 40MHz with VLB multicard (PINE PT-627) and VLB videocard (DIAMOND STEALTH 32VL) its best video and HDD speed - very good
4) PCI dont work or unstable at 40MHz - i think VLB-PCI bridge just locks at 40MHz - bad

So i can get fastest VLB-only system at 40MHz or very stable and fast enough VLB/PCI system at 33MHz (with internal oscilator or 33MHz external quartz generator)
My choice is very stable and fast enough VLB/PCI system at 33MHz

I think thats all with my experiments and testing
Will try to make some kind or manual for other users of SF586 😀

Motherboard slowly moves to my market-place listing 😉

Beerfloat wrote on 2025-12-31, 08:53:

I'm also hoping to find/cobble together Pentium MMX and K6-2/3 BIOS support with a voltage interposer.

Hi Beerfloat,

The SF586 05/18/95 (VIP3A-ZS1) BIOS is already on my “to be patched” list and I expect to start working on it later this month.

Note that the CPU support in this BIOS is very limited. I only see Pentium (P54) and Cyrix 6x86 (M1) support. No Pentium MMX, Cx6x86MX, K5, K6, or Winchip C6/2 support.
So I don’t know yet if Pentium MMX and K6-2/III support can be added to such an old BIOS. But Winchip support should be possible.
I will know more when I’ve completed the disassembly and analysis of this BIOS.

I will keep you posted.
Cheers, Jan