Depends. If you use an unsupported CPU with the correct voltage configuration, the POST screen if it appears will not properly identify the CPU and the BIOS would not enable the K6 Write Allocation. That feature could be later enabled by software after the POST was successful. You should disable both "Video RAM Cacheable" and "System BIOS Cacheable" for stability with K6plus CPUs if the BIOS offers these settings.

There's a place where there are unofficial BIOS updates that brings the K6plus support for motherboards that didn't got them: http://web.inter.nl.net/hcc/J.Steunebrink/k6plus.htm. However there isn't one for that "PC Compatibility Card" there.

I have a computer using a AMD K6-2+/450 CPU on a Gigabyte GA-586ATV motherboard with a voltage adapter, however in my case I had to use an older BIOS version with the patcher you mentioned and some modifications with MODBIN as the latest one for my motherboard caused a reboot loop with the K6-2+ during the POST screen no matter what I do regardless whether I use the one that BIOS Jan offers on his website or the official one from Gigabyte website.

Interesting. Unfortunately, I've yet to discover any useful way to get in the middle of the boot process of the card earlier or any kind of debug output/log for what's going on. I haven't attempted to attach MacsBug (old classic Mac debugger) to it yet. So, I suppose it's possible the CPUs are both just inoperable. But, that seems unlikely. They're both seemingly in practically brand new condition.

The OrangePCi application doesn't even attempt to switch the video over to control of the PC Compatibility card's VGA output when either K6-III+ is installed. That's normally nearly instantaneous with the WinChip and Pentium MMX installed, so something is failing very early in the boot process. I wondered if the K6-III+ maybe required special initialization that wasn't backward compatible with the K6-III that might cause a machine to not even get far enough to POST? The presence of the special PowerNow! power management features in-particular seemed like they might be such an impediment.

But, it sounds like from your description that any machine which supported the K6-III should at least get to the point of initializing the CPU, etc. to get to the POST screen if you swapped in a K6-III+ instead?

Nathan_A wrote on 2021-08-22, 19:26:

Interesting. Unfortunately, I've yet to discover any useful way to get in the middle of the boot process of the card earlier or any kind of debug output/log for what's going on. I haven't attempted to attach MacsBug (old classic Mac debugger) to it yet. So, I suppose it's possible the CPUs are both just inoperable. But, that seems unlikely. They're both seemingly in practically brand new condition.

The OrangePCi application doesn't even attempt to switch the video over to control of the PC Compatibility card's VGA output when either K6-III+ is installed. That's normally nearly instantaneous with the WinChip and Pentium MMX installed, so something is failing very early in the boot process. I wondered if the K6-III+ maybe required special initialization that wasn't backward compatible with the K6-III that might cause a machine to not even get far enough to POST? The presence of the special PowerNow! power management features in-particular seemed like they might be such an impediment.

But, it sounds like from your description that any machine which supported the K6-III should at least get to the point of initializing the CPU, etc. to get to the POST screen if you swapped in a K6-III+ instead?

When I was attempting to use an older BIOS one of these old versions didn't even POST at all with the K6-2+ CPU on it. I also have a K6-III (non plus) CPU and got the same issues with that motherboard so it is very unlikely for me that the PowerNow! feature is the impediment.

The only difference besides the PowerNow! feature and the core voltage requirement is the CPU family and model between K6-III and K6plus as the latter ones are like K6-III for mobile computers.

Gmlb256 wrote on 2021-08-22, 19:34:

When I was attempting to use an older BIOS one of these old versions didn't even POST at all with the K6-2+ CPU on it. I also have a K6-III (non plus) CPU and got the same issues with that motherboard so it is very unlikely for me that the PowerNow! feature is the impediment.

The only difference besides the PowerNow! feature is the CPU family and model between K6-III and K6plus as the latter ones are like K6-III for mobile computers.

Well, I have a K6-III non-plus on the way. We'll see what shape it's in when it arrives. So, I'll be able to see if that works. Space is tight and cooling options are scant given that this thing is basically a PC on a 7" PCI card, so I was really hoping to be able to use the K6-III+ for both its lower voltage requirements as well as its potential to trivially reach 500 Mhz (83.33x6) with relatively meager cooling.

It sounds like something I really need to do is find a way to test these K6-III+ CPUs in a more standard setup to make sure they actually work at all. Unfortunately, I don't have anything remotely resembling a standalone Socket 7 or Super Socket 7 machine around to do that with.

This is a longshot, but what are the odds that there's anyone in or around the Portland, OR area in the US that has a system handy that these CPUs could be tested in to make sure they work at all?

I'd also consider shipping them to someone for testing, but if someone is local, that would be easier and cheaper.

as long as the VRM will support the proper voltage, you could modify the bios and add a Microcode Patch so it will detect the cpu.
I don't know much about those SBCs essentially thats what they are anyways. It might be a rabbit hole not worth going down anyways as you could face some bottle necks and even if you got the cpu to run it would be crippled compared to a supersocket 7 board.

Warlord wrote on 2021-08-23, 07:53:

as long as the VRM will support the proper voltage, you could modify the bios and add a Microcode Patch so it will detect the cpu.
I don't know much about those SBCs essentially thats what they are anyways. It might be a rabbit hole not worth going down anyways as you could face some bottle necks and even if you got the cpu to run it would be crippled compared to a supersocket 7 board.

Oh, it's crippled as-is given the SiS 5598 integrated video, and this entire vintage project I'm undertaking started out as a rabbit hole not worth going down. 😉

But, I have hopes some day of eventually getting an OrangePC 660 card, which was a Super Socket 7 card (max 112 Mhz FSB and a RIVA 128 chip) that I'll want to fiddle with getting to work, so I figure the lesser 620 card is as good a way as any to learn the ropes about what it will take to max out the CPU these weird little cards can handle.

A great vintage project indeed, and nice you could extract the BIOS image!

I've checked this Award SiS 5598 BIOS for CPU support and as you would expect, it has no K6-2+/K6-III+ support.
It does support the K6-2 and K6-III up to 400MHz. So if it boots with a K6-III, but not with the K6-III+, you know the BIOS is to blame.

On these older Award BIOSes, I often could get a K6-2+/-III+ to boot when I set he multiplier to x3. The POST screen would then indicate somthing odd like "Unkown -S CPU at 66MHz". When this worked, I used a DOS util like TSC.EXE to increase the multiplier, but this wasn't always stable.

You can find more details about this work-around at:
http://www.steunebrink.info/S1590-K6-III+.htm

Jan.

Edit: updated my website link.

Nathan_A wrote on 2021-08-23, 16:13:

Warlord wrote on 2021-08-23, 07:53:

as long as the VRM will support the proper voltage, you could modify the bios and add a Microcode Patch so it will detect the cpu.
I don't know much about those SBCs essentially thats what they are anyways. It might be a rabbit hole not worth going down anyways as you could face some bottle necks and even if you got the cpu to run it would be crippled compared to a supersocket 7 board.

Oh, it's crippled as-is given the SiS 5598 integrated video, and this entire vintage project I'm undertaking started out as a rabbit hole not worth going down. 😉

But, I have hopes some day of eventually getting an OrangePC 660 card, which was a Super Socket 7 card (max 112 Mhz FSB and a RIVA 128 chip) that I'll want to fiddle with getting to work, so I figure the lesser 620 card is as good a way as any to learn the ropes about what it will take to max out the CPU these weird little cards can handle.

I don't think it is the chipset that is the problem.

I have a PC Chips M571 Rev3.2 board that has the same chipset, I have it patched with a Steunebrik BIOS which allows for k6-III and K6-3+ support, however my board doesn't officially support the + chips even with the patched BIOS, at least not with the revision board I have. It is essentially down to a voltage problem and I can only get the voltage as low as 2.2v.

However, I was resilient and with a socket 370 cooler with a copper base and a 40mm fan pointed at the VRM I was able to get a K6-3+ to run at 83x6 for 500mhz 100% stable. The chip doesn't even really get warm.

But be aware that the SiS5598 doesn't officially support 83mhx FSB, it onlt supports 75mhz and 83 is actually overclocking it. With the right ram it should work fine however.

Link to my thread if you are interested.

PC Chips M571 K6-3+ Support - UPDATED WITH BENCHMARKS

Chkcpu wrote on 2021-08-23, 19:57:

A great vintage project indeed, and nice you could extract the BIOS image! […]
Show full quote

A great vintage project indeed, and nice you could extract the BIOS image!

I've checked this Award SiS 5598 BIOS for CPU support and as you would expect, it has no K6-2+/K6-III+ support.
It does support the K6-2 and K6-III up to 400MHz. So if it boots with a K6-III, but not with the K6-III+, you know the BIOS is to blame.

On these older Award BIOSes, I often could get a K6-2+/-III+ to boot when I set he multiplier to x3. The POST screen would then indicate somthing odd like "Unkown -S CPU at 66MHz". When this worked, I used a DOS util like TSC.EXE to increase the multiplier, but this wasn't always stable.

You can find more details about this work-around at:
http://web.inter.nl.net/hcc/J.Steunebrink/S1590-K6-III+.htm

Jan.

Interesting. I'll try out 66x3 @ 2.1v just to see if it gets any further through the initialization. How were you able to determine so quickly what the supported CPUs & frequencies were in the image I posted? I'm asking because I want to also extract the OrangePC 5xx series and Orange PC 660 BIOS images to interrogate them for similar details.

As a separate matter, if you're amenable, I'd very much like to work with you to see we can get this little OrangePC 620 fella working.

Incidentally I happen to also have a Nubus-based OrangePC 290 that I've been fiddling with. It uses a different piece of software for control, but it's setup more or less the same way. Namely, the BIOS is also encoded in the resource fork of the application. It came with an AMD Am486DX2-80, but I've managed to also get it to boot with: an AMD Am586-133 (even at 160Mhz for a bit), an Intel DX4-100 (also stable at 120Mhz), and a Cyrix 5x86-100GP (also stable at 120Mhz). It uses an old 386/486 AMI BIOS image for an OPTi chipset w/ 128k L2 cache and a WD90C30 VGA chip. Supposedly the chipset supports write-back mode, but it doesn't appear to be enabled by default, and there's no setting in the BIOS to enable it. I also haven't been able to get the 50Mhz bus speed working at all. Lastly, the POST screen incorrectly reports every CPU frequency above 75Mhz and decides that the Cyrix CPU is a Cyrix 486SLC as well as has weird layout artifacts of the POST table with the Cyrix CPU. Though otherwise works fine.

All these OrangePC cards also have their CMOS fed to them at boot by the host application. There's a default CMOS setting stored in the resource fork of the control application which then gets written to a "Preferences" file on first launch (the Mac version of something like registry settings or a config file), so that when you make CMOS changes after the card has bootstrapped they get written back to this file and used in subsequent boots. So, when you want to "clear your CMOS" you just have to delete this file on the host machine.

Hi Jasin Natael,
Thanks for chipping in with your SiS5598 chipset experiences. I read your M571 K6-3+ adventures with great interest! 😀

Nathan_A wrote on 2021-08-24, 00:18:

Interesting. I'll try out 66x3 @ 2.1v just to see if it gets any further through the initialization. How were you able to determ […]
Show full quote

Interesting. I'll try out 66x3 @ 2.1v just to see if it gets any further through the initialization. How were you able to determine so quickly what the supported CPUs & frequencies were in the image I posted? I'm asking because I want to also extract the OrangePC 5xx series and Orange PC 660 BIOS images to interrogate them for similar details.

As a separate matter, if you're amenable, I'd very much like to work with you to see we can get this little OrangePC 620 fella working.

Incidentally I happen to also have a Nubus-based OrangePC 290 that I've been fiddling with. It uses a different piece of software for control, but it's setup more or less the same way. Namely, the BIOS is also encoded in the resource fork of the application. It came with an AMD Am486DX2-80, but I've managed to also get it to boot with: an AMD Am586-133 (even at 160Mhz for a bit), an Intel DX4-100 (also stable at 120Mhz), and a Cyrix 5x86-100GP (also stable at 120Mhz). It uses an old 386/486 AMI BIOS image for an OPTi chipset w/ 128k L2 cache and a WD90C30 VGA chip. Supposedly the chipset supports write-back mode, but it doesn't appear to be enabled by default, and there's no setting in the BIOS to enable it. I also haven't been able to get the 50Mhz bus speed working at all. Lastly, the POST screen incorrectly reports every CPU frequency above 75Mhz and decides that the Cyrix CPU is a Cyrix 486SLC as well as has weird layout artifacts of the POST table with the Cyrix CPU. Though otherwise works fine.

All these OrangePC cards also have their CMOS fed to them at boot by the host application. There's a default CMOS setting stored in the resource fork of the control application which then gets written to a "Preferences" file on first launch (the Mac version of something like registry settings or a config file), so that when you make CMOS changes after the card has bootstrapped they get written back to this file and used in subsequent boots. So, when you want to "clear your CMOS" you just have to delete this file on the host machine.

Hi Nathan_A,
Back in the day I did a lot of Socket 7 BIOS patching for my K6plus support project, resulting in "The Unofficial K6-2+ / K6-III+ page" at http://www.steunebrink.info/k6plus.htm
I specialized in the 1997/1998/1999 Award BIOSes and can usually find their known limits now within 5 minutes. 😉
However, Award made a lot of code changes over time and patching such a BIOS for K6plus support requires a detailed analyses and takes me about 4 hours work...
(A lot quicker than the 3 months of spare time it cost me to patch my first Socket 7 BIOS) 🙁

I don't know if I'm able to help with the other OrangePC cards but yes, I do like to work with you on the 620 card.
I've already determined that the OPC_620 BIOS can be patched for K6plus support, so I can start with that.

Cheers, Jan

Edit: updated the link to my website.

Chkcpu wrote on 2021-08-24, 14:26:

Hi Jasin Natael, Thanks for chipping in with your SiS5598 chipset experiences. I read your M571 K6-3+ adventures with great inte […]
Show full quote

Hi Jasin Natael,
Thanks for chipping in with your SiS5598 chipset experiences. I read your M571 K6-3+ adventures with great interest! 😀

Nathan_A wrote on 2021-08-24, 00:18:

Interesting. I'll try out 66x3 @ 2.1v just to see if it gets any further through the initialization. How were you able to determ […]
Show full quote

Interesting. I'll try out 66x3 @ 2.1v just to see if it gets any further through the initialization. How were you able to determine so quickly what the supported CPUs & frequencies were in the image I posted? I'm asking because I want to also extract the OrangePC 5xx series and Orange PC 660 BIOS images to interrogate them for similar details.

As a separate matter, if you're amenable, I'd very much like to work with you to see we can get this little OrangePC 620 fella working.

Incidentally I happen to also have a Nubus-based OrangePC 290 that I've been fiddling with. It uses a different piece of software for control, but it's setup more or less the same way. Namely, the BIOS is also encoded in the resource fork of the application. It came with an AMD Am486DX2-80, but I've managed to also get it to boot with: an AMD Am586-133 (even at 160Mhz for a bit), an Intel DX4-100 (also stable at 120Mhz), and a Cyrix 5x86-100GP (also stable at 120Mhz). It uses an old 386/486 AMI BIOS image for an OPTi chipset w/ 128k L2 cache and a WD90C30 VGA chip. Supposedly the chipset supports write-back mode, but it doesn't appear to be enabled by default, and there's no setting in the BIOS to enable it. I also haven't been able to get the 50Mhz bus speed working at all. Lastly, the POST screen incorrectly reports every CPU frequency above 75Mhz and decides that the Cyrix CPU is a Cyrix 486SLC as well as has weird layout artifacts of the POST table with the Cyrix CPU. Though otherwise works fine.

All these OrangePC cards also have their CMOS fed to them at boot by the host application. There's a default CMOS setting stored in the resource fork of the control application which then gets written to a "Preferences" file on first launch (the Mac version of something like registry settings or a config file), so that when you make CMOS changes after the card has bootstrapped they get written back to this file and used in subsequent boots. So, when you want to "clear your CMOS" you just have to delete this file on the host machine.

Hi Nathan_A,
Back in the day I did a lot of Socket 7 BIOS patching for my K6plus support project, resulting in "The Unofficial K6-2+ / K6-III+ page" at http://web.inter.nl.net/hcc/J.Steunebrink/k6plus.htm
I specialized in the 1997/1998/1999 Award BIOSes and can usually find their known limits now within 5 minutes. 😉
However, Award made a lot of code changes over time and patching such a BIOS for K6plus support requires a detailed analyses and takes me about 4 hours work...
(A lot quicker than the 3 months of spare time it cost me to patch my first Socket 7 BIOS) 🙁

I don't know if I'm able to help with the other OrangePC cards but yes, I do like to work with you on the 620 card.
I've already determined that the OPC_620 BIOS can be patched for K6plus support, so I can start with that.

Cheers, Jan.

Oh that was you? I'm currently using your AP5T BIOS, thanks for that 😀

Chkcpu wrote on 2021-08-23, 19:57:

A great vintage project indeed, and nice you could extract the BIOS image! […]
Show full quote

A great vintage project indeed, and nice you could extract the BIOS image!

I've checked this Award SiS 5598 BIOS for CPU support and as you would expect, it has no K6-2+/K6-III+ support.
It does support the K6-2 and K6-III up to 400MHz. So if it boots with a K6-III, but not with the K6-III+, you know the BIOS is to blame.

On these older Award BIOSes, I often could get a K6-2+/-III+ to boot when I set he multiplier to x3. The POST screen would then indicate somthing odd like "Unkown -S CPU at 66MHz". When this worked, I used a DOS util like TSC.EXE to increase the multiplier, but this wasn't always stable.

You can find more details about this work-around at:
http://web.inter.nl.net/hcc/J.Steunebrink/S1590-K6-III+.htm

Jan.

Didn't know that you were here at VOGONS!

Chkcpu wrote on 2021-08-24, 14:26:

Hi Jasin Natael,
Thanks for chipping in with your SiS5598 chipset experiences. I read your M571 K6-3+ adventures with great interest! 😀

Hello!

I thank you very much for the all of the work you have done over the years, I've used your patched BIOS' on more than one occasion.
So many of us literally couldn't be using these CPUs without your work, so thank you!

I'm glad you got some interest out of my thread.

Gmlb256 wrote on 2021-08-24, 14:53:

Didn't know that you were here at VOGONS!

I joined recently.
After helping people via e-mail and visiting here as guest for many years, it was high time I joined the VOGONS community and make my contributions directly. 😉

Chkcpu wrote on 2021-08-24, 17:08:

Gmlb256 wrote on 2021-08-24, 14:53:

Didn't know that you were here at VOGONS!

I joined recently.
After helping people via e-mail and visiting here as guest for many years, it was high time I joined the VOGONS community and make my contributions directly. 😉

PM you about the status of my Gigabyte motherboard.

Jasin Natael wrote on 2021-08-23, 20:41:

But be aware that the SiS5598 doesn't officially support 83mhx FSB, it onlt supports 75mhz and 83 is actually overclocking it. With the right ram it should work fine however.

Link to my thread if you are interested.

PC Chips M571 K6-3+ Support - UPDATED WITH BENCHMARKS

I ran that same board at 83mhz for a decade your pci bus is out of spec and the asynchronous mode is in op but who cares?

I know at least one guy that modified his board for 108mhz FSB (hard wired) and everything ran great just need solid ram.

Also I have found that as long as the board knows what a k6 is and the voltage is in spec
the normal k6-2 seems to run fine even if the speed is displayed wrong and the board thinks it’s a normal k6

rmay635703 wrote on 2021-08-24, 17:46:

I ran that same board at 83mhz for a decade your pci bus is out of spec and the asynchronous mode is in op but who cares? […]
Show full quote

Jasin Natael wrote on 2021-08-23, 20:41:

But be aware that the SiS5598 doesn't officially support 83mhx FSB, it onlt supports 75mhz and 83 is actually overclocking it. With the right ram it should work fine however.

Link to my thread if you are interested.

PC Chips M571 K6-3+ Support - UPDATED WITH BENCHMARKS

I ran that same board at 83mhz for a decade your pci bus is out of spec and the asynchronous mode is in op but who cares?

I know at least one guy that modified his board for 108mhz FSB (hard wired) and everything ran great just need solid ram.

Also I have found that as long as the board knows what a k6 is and the voltage is in spec
the normal k6-2 seems to run fine even if the speed is displayed wrong and the board thinks it’s a normal k6

Yes mine seems to be working fine as well, but it did require the right ram and some tinkering to get it stable.

My K6-3+ is detected as a standard k6-III non plus. But works just fine.

Quick update. So, my non-plus K6-III CPUs arrived today and I was able to immediately get them working in the OrangePC 620 card. They're K6-III 450/AHX 2.4v CPUs according to the imprint on the heat spreader. No idea if that's actually accurate given how many incredibly weird counterfeiting operations there are out there in the world for things like this, but the card definitely gets to POST. One oddity is that the POST screen thinks that the CPU is running at 256 Mhz, or at least is reporting it as such. I've seen this BIOS properly report my overclocked Pentium MMX at 262 Mhz, so I don't know what's going on there with the K6.

I was able to get the non-plus K6-III to run at 400 Mhz and 450 Mhz at both 2.4v and undervolted to 2.2v because the thing gets HOT! and the cooler on it is pitiful at the moment.

I also attempted the 66x3 @ 2.1v on both my K6-III+'s, but it didn't do anything different than hang as usual.

So... either there's something going on with the BIOS that's preventing my "plus" CPUs from getting to basic initialization and POSTing, or they were perhaps fried by the previous owner? Anybody got a good way for me to test the "plus" CPUs? I don't have a supported PC handy to try them in. Maybe Free Geek here in Portland will have a system around for testing.