Chkcpu wrote on 2025-10-03, 20:39:

Ciao Elio, […]
Show full quote

Ciao Elio,

I’m sorry to hear the 66MHz FSB trial didn’t work and the Camaro board still refuses to POST.

I made some progress analyzing this Compaq BIOS but POST step 2F has a lot of code and I haven’t figured out yet how it works.
Luckily, someone made a copy of this Compaq BIOS with Uniflash and posted it on TRW, so I have something to work with.

The top 16KB of this 256KB BIOS is the bootblock and contains readable code and data. The rest of the BIOS is not recognizable as code and is probably compressed or encrypted. I don’t recognize any known structures in this BIOS, so it is probably not AMI but Compaq’s own product.

All POST code up to step 2F is in the bootblock so I was able to disassemble that. When I follow the code from the start address to POST_2F, I see the follow POST steps in this order:
00, 08, 09, 01, 02, 03, 04, 05, 0D, 14, 15, 10, 11, 12, 13, 0E, 42, 43, 0F, 28, 2A, 2D, 42, 43, 2E, 2F.
This coincides nicely with what you found, only steps 28 and 2D are missing from your list.
In POST steps 42 and 43 the RAM is tested and all POST code up till step 0F runs from ROM and POST_0F copies the bootblock to RAM. Step 28 jumps to this code in RAM and from step 2D onwarts the POST runs from RAM, including a second call to steps 42 and 43 to test another part of memory. So also steps 2E and 2F run from RAM.
If one of the two steps 43 reports a memory error, the code jumps straight to POST 47 and halts, and never reaches steps 2E and 2F.

Because of this initial analysis, I agree that the RAM is fine and that the problem is either the BIOS or some hardware failure.
I haven’t found any checksum code yet and I don’t know if this BIOS has a recovery mode, so shorting 2 address lines on the BIOS chip to trigger a BIOS checksum error may not work on this BIOS.

I will study the POST_2F code further and hope to find a clue there.
Cheers, Jan

Ciao Jan
Thank you very much for your analysis on the BIOS, and in fact as I thought, the first part of this Compaq BIOS is working, at least to the point where it detects the RAM, here maybe I can try to see the post codes, starting without RAM, in this case it stops at 47, this should confirm the sequence obtained from both, even if in my list, there are not those two codes that you reported, or it does not show them (I don't know why), or it is possible that when pressing the button, it inadvertently pressed a second time (?).
On the origin of this BIOS, that is, if it is a modified AMI, this seemed very likely to me, but looking at what 😨 they correspond to, they seem more likely to those of the link (put in my last comment), so maybe as you rightly suggest, it could be a BIOS made entirely by Compaq, maybe using an AMI inspiration scheme, so maybe the 2F is in fact an error, corresponding to "Write to diagnostic byte", unfortunately I still don't understand what it means.
So the problem of the Recovery BIOS would remain, if it is as you say, that is, that there may not be a Recovery BIOS, if it is forced by joining the two corresponding PINs of the BIOS chip, it would be a big problem, because there would be nothing else to do, in fact perhaps the only solution would be to remove the chip (with the welder), after adding a socket on the MB, reprogram and try again.
Last chance, it would be perhaps 🤔 the easiest, if there was an extension cable, for 8 PIN chips there are tweezers, which attach to the welded chip, for PLCC32 chips I don't think there is anything like that, it seems to me that the only solution would be to build it yourself, and then connect the programmer, I think the battery 🔋 CR2032 of the motherboard should be removed, I don't know the reason why you can't find anything like that, maybe for those who deal with reprogramming the BIOS chips, it's not complicated to remove and replace the chip (welding it), but I don't think that in a service center, they do it on purpose to make the restoration uneconomical, maybe the reprogramming could be done using something else (?), I don't think there is a special port like SPI ASUS, or who knows if Compaq used USB, or even the Serials or the Parallel.

Doing some research 🔍, I found the PC it should come from, it would be a Compaq PC, model Presario 7350, the 7350 had AMD K6-2 475 MHz CPU, 40 GB disk 🇬🇧, and 64 MB of RAM, for the rest having both audio and video integrated, no PCI cards were needed, in fact the slots were free, although I believe that at least one was occupied by an internal modem (PCI), but I'm not sure 🤔, because there were external modems, and that PC was used to connect to the internet, yes today it seems a bit strange, but it was a PC to allow entry to the web, maybe it was a cheap PC (?), but there were other faster models, the 7360 for example was a 500 MHz, while the 7370 was a 533 MHz, the latter would be the upgrade I had in mind.
Unfortunately 😣 as seen, until now, even changing CPU, RAM or jumper position, the PC does not go beyond the Post Code 2F 2E, so I would rule out a problem due to a defective CPU, also because usually only — —, so no post code flows.
Even if I find the pdf manual, I don't think I would find useful indications for the resolution of this problem, although there are usually some solutions to simple startup problems, this case I think it would be a problem to solve by taking the PC to service.
Looking at the BIOS screens, it looks similar to that of my AST Bravo Pro (Pentium Pro 200), and I think it is based on the Phoenix BIOS, which was used in Intel production motherboards, but I don't know if it actually is, or if it is just similar to the view, while the post codes could refer to something else.

As Socket 7, in addition to the Compaq Camaro, I have on the bench a similar PCChips M560 Rev:4.1A, the motherboard is working, except for the L2 cache on board, which should be 512 KB, there are two UT6164C32Q-6 chips, but there are no jumpers to select the L2 cache, it means that there is a bridge, and the selection is fixed.
There is an RTC chip VIA VT82885N, which must be fixed, the battery is missing 🪫, for this reason, turning it off it does not store the BIOS settings and the changes, fixing it is relatively simple, but I don't think the L2 cache is not working, it depends on the lack of the battery in the RTC chip, it is a problem that I imagine is due to something else, unfortunately the cache chips are welded, and it is not possible to exclude one or the other, to see if maybe 🤔 one of the two works, in that case you would have 256 KB, which anyway are always better than 0 KB.
The BIOS is an AMI, Relase 11/26/1997S, and the chipset is rebranded as TX PRO, in practice it should be an ALI M1531, maybe 🤔 not the best for this kind of MB, but anyway at most the FSB reaches 83 MHz, but if you settle for 400 MHz with FSB 66, and if you have a CPU with integrated L2 cache, the on-board one of 512 KB becomes L3, in this case the performance with L3 cache of 512 KB or with 0 KB, should be quite similar, if not even the same.
Currently with the cache enabled I read on the post card PCI, the 71 AC codes, just out of curiosity, I could write down the previous codes, of course I already know that the problem is due to the on-board L2 cache (defective?), I have the same problem on another PCChips M549, initially I would suspect L2 cache chips, one or both could be not working, but sometimes there could be a different problem, such as a non-powered pin, for an interrupted track or a faulty SMD component, maybe first it would be the case to do some checks, because to manage to make the L2 cache work, it will work best, I have also seen some that replace the chips with others of double capacity, in this case getting 1 MB of L2 cache, honestly I don't know if this requires the movement of the bridge, in the missing jumper area to select the amount of cache.

PC@LIVE wrote on 2025-10-04, 13:06:

Doing some research 🔍, I found the PC it should come from, it would be a Compaq PC, model Presario 7350, the 7350 had AMD K6-2 4 […]
Show full quote

Doing some research 🔍, I found the PC it should come from, it would be a Compaq PC, model Presario 7350, the 7350 had AMD K6-2 475 MHz CPU, 40 GB disk 🇬🇧, and 64 MB of RAM, for the rest having both audio and video integrated, no PCI cards were needed, in fact the slots were free, although I believe that at least one was occupied by an internal modem (PCI), but I'm not sure 🤔, because there were external modems, and that PC was used to connect to the internet, yes today it seems a bit strange, but it was a PC to allow entry to the web, maybe it was a cheap PC (?), but there were other faster models, the 7360 for example was a 500 MHz, while the 7370 was a 533 MHz, the latter would be the upgrade I had in mind.
Unfortunately 😣 as seen, until now, even changing CPU, RAM or jumper position, the PC does not go beyond the Post Code 2F 2E, so I would rule out a problem due to a defective CPU, also because usually only — —, so no post code flows.
Even if I find the pdf manual, I don't think I would find useful indications for the resolution of this problem, although there are usually some solutions to simple startup problems, this case I think it would be a problem to solve by taking the PC to service.
Looking at the BIOS screens, it looks similar to that of my AST Bravo Pro (Pentium Pro 200), and I think it is based on the Phoenix BIOS, which was used in Intel production motherboards, but I don't know if it actually is, or if it is just similar to the view, while the post codes could refer to something else.

Ciao Elio,

About the Camaro board, I’ve now completed the analysis on the POST_2F code. There is a lot of work done by this POST step.

It starts to copy the whole BIOS to RAM and then puts the Bootblock into Shadow RAM. It then scans the BIOS copy in RAM for the headers of the compressed sections. These headers all start with a 0xF1D64342 signature, and I’ve counted 7 of them so there must be 7 compressed modules in this BIOS.

Then the decompression engine in the Bootblock is set to work to decompress all modules to their allocated places in RAM. Finally it jumps into one of the decompressed modules to continue with the rest of POST.

Because this whole process is done from the Bootblock copy in shadow RAM, and all the previous hardware tests were successful, I believe the RAM must be OK.
Looking at all the decompression work in POST_2F, I think there is a high probability that the BIOS is corrupted, maybe by a bad flash, and only the Bootblock is still intact.

To bad the BIOS chip is soldered to the board and you need a hot-air station to desolder it. Maybe time to invest in one if you want to fix boards like these. You can then also put a PLCC32 socket on the board. 😉

I really hope you can fix this nice board.
Greetings, Jan

Chkcpu wrote on 2025-10-06, 18:41:

Ciao Elio, […]
Show full quote

PC@LIVE wrote on 2025-10-04, 13:06:

Doing some research 🔍, I found the PC it should come from, it would be a Compaq PC, model Presario 7350, the 7350 had AMD K6-2 4 […]
Show full quote

Doing some research 🔍, I found the PC it should come from, it would be a Compaq PC, model Presario 7350, the 7350 had AMD K6-2 475 MHz CPU, 40 GB disk 🇬🇧, and 64 MB of RAM, for the rest having both audio and video integrated, no PCI cards were needed, in fact the slots were free, although I believe that at least one was occupied by an internal modem (PCI), but I'm not sure 🤔, because there were external modems, and that PC was used to connect to the internet, yes today it seems a bit strange, but it was a PC to allow entry to the web, maybe it was a cheap PC (?), but there were other faster models, the 7360 for example was a 500 MHz, while the 7370 was a 533 MHz, the latter would be the upgrade I had in mind.
Unfortunately 😣 as seen, until now, even changing CPU, RAM or jumper position, the PC does not go beyond the Post Code 2F 2E, so I would rule out a problem due to a defective CPU, also because usually only — —, so no post code flows.
Even if I find the pdf manual, I don't think I would find useful indications for the resolution of this problem, although there are usually some solutions to simple startup problems, this case I think it would be a problem to solve by taking the PC to service.
Looking at the BIOS screens, it looks similar to that of my AST Bravo Pro (Pentium Pro 200), and I think it is based on the Phoenix BIOS, which was used in Intel production motherboards, but I don't know if it actually is, or if it is just similar to the view, while the post codes could refer to something else.

Ciao Elio,

About the Camaro board, I’ve now completed the analysis on the POST_2F code. There is a lot of work done by this POST step.

It starts to copy the whole BIOS to RAM and then puts the Bootblock into Shadow RAM. It then scans the BIOS copy in RAM for the headers of the compressed sections. These headers all start with a 0xF1D64342 signature, and I’ve counted 7 of them so there must be 7 compressed modules in this BIOS.

Then the decompression engine in the Bootblock is set to work to decompress all modules to their allocated places in RAM. Finally it jumps into one of the decompressed modules to continue with the rest of POST.

Because this whole process is done from the Bootblock copy in shadow RAM, and all the previous hardware tests were successful, I believe the RAM must be OK.
Looking at all the decompression work in POST_2F, I think there is a high probability that the BIOS is corrupted, maybe by a bad flash, and only the Bootblock is still intact.

To bad the BIOS chip is soldered to the board and you need a hot-air station to desolder it. Maybe time to invest in one if you want to fix boards like these. You can then also put a PLCC32 socket on the board. 😉

I really hope you can fix this nice board.
Greetings, Jan

Ciao Jan
Thank you for all the tests and suggestions, that you did to help me understand something more, undoubtedly there is a problem, and your deductions seem to confirm, that there should be no HW failure.
For my part, I have performed other tests, to understand if the card is blocked, or remains waiting for instructions, when on the PCI post card there is 2F 2E, I tried to reset, and the codes restart until they reach the usual 2F 2E, this means that the card does not lock, but it would be waiting for something (?), previously I tried to press various keys at startup (F1 F2 etc...), but I got nothing.
Another thing 😨 that I did, is to start without RAM, on the post card it shows 47 43, then I pressed the PCI post card button, and I wrote down the following codes:
42 F4 0E 13 12 11 10 15 14 0d 05 04 03 02 01 09 08 00.
The result is quite similar, to the one with the RAM installed, obviously here there is a lower number of codes, but precisely this in my opinion, confirms that the RAM is working, which you also agree with.
In light of all this additional info, so it would seem correct what I had given for very likely, that is, that the BIOS is partially damaged, perhaps 🤔 for an attempted upgrade, or for interrupting the update too soon.
Of course, it would not have been a problem, if the chip was not welded, at this point, I need to understand if it is worth removing it, or if it is possible to self-build an MF extension for the programmer.
Removing the BIOS chip with hot air, it wouldn't be too difficult, for those who have some experience, even if in the various YouTube videos some Vogons, do it without major problems, for my part, I wouldn't want to risk damaging the MB, like skipping some tracks, I think that first I should practice on scrap cards, but it takes a lot of free time, so at least in a short time it won't be possible.
Perhaps the easiest thing to do is to build the MF extension cord, if the problem is only the BIOS to rewrite, it would be relatively simple and fast, moreover I could reuse it in other Motherboards with welded BIOS chips, for example some Siemens, I think it can be the simplest and cheapest solution, assuming it works.
Greetings
Elio

Just so I don't get bored, I fished a Chaintech 5TDM2 M101, Intel TX chipset, AT format with the possibility of using ATX power supplies, I had written about it a while ago, unfortunately 😣 despite some attempts in the past, it doesn't seem to give any sign of life, I checked the voltages and they seem ok, using a Pentium MMX 200 I read in the VRM just over 2.8V, which would be the right voltage, I would rule out a problem with the BIOS because I had reprogrammed it, you can think that the problem is somewhere else, I could suspect some electrolytic capacitors, which although they are not swollen, they could be Malfunctioning or at the end of life, it wouldn't surprise me but maybe it would be appropriate to try to remove some of them to see what state they are in. (?!)
Before reprogramming the BIOS, it had displayed a code on the post card, C0 — I don't think there were other codes before, because the second one is—, so it started starting but stops immediately, the BIOS I remember being AWARD, in short I should be 100% sure, that the problem is not the BIOS, I downloaded the one I downloaded from the TRW website, I think there are tested and working files there, so even trying to reprogram with another version of BIOS, I shouldn't get any results.
Unfortunately that MB uses automatic CPU recognition, that is, there are no jumpers to select VCORE and FSB CPUs, this on the one hand is very convenient, but on the other hand, it makes the doubt that if there is a problem in the recognition, the card may not start.
In short, even here there may not be an HW problem, but perhaps more likely a HW (or SW?) defect, the system for automatic CPU recognition, should be integrated in the BIOS, and it should be somehow possible, from some reference PIN on the S.7, that is, one would say that VCORE use, or if it is AMD or Intel CPU or other.

Looking for info on some S.7, I found a possible secret setting, for an S.7 ACORP A-5VIA3P motherboard, currently on that board with VIA VPX chipset there is a Cyrix MII 333 (263 MHz 75X3.5), the performance is quite slow, I have tried other Intel CPUs, and I have never seen it go like other PCs (the same), probably by eye I would say at least 30% less, exactly it could be more or less, but I should do some bench, maybe 🤔 with those of Phil's I would find if there is something in particular, that can slow it down, for example the cache or the lack of driver.
Let's say that would be the fastest CPU (Cyrix) I can install, in fact if what I saw on TRW is correct, it might be possible to install faster CPUs, I would have found an interesting 🧐 secret setting of the VCORE, by using a second jumper, you could get a 2.2V VCORE, this would be right for AMD K6-2 CPUs and also the K6 266 and 300, in addition to the latest versions of the Cyrix MII, in short it would be possible to install a K6-2-400 or even a 450.
Of course I haven't done anything yet, because I would like to be sure that it doesn't create other problems, first, I should be sure that the regulator is switching, after the double jumper actually works, because it could be referred to a later version of the same MB.
In case it works, there could be a BIOS problem, currently there is the latest version dated 1997, easily a K6-2 would be recognized as simple K6, and the displayed speed could be lower, for example K6-300 instead of K6-2-400.
Surely it would work faster, but I don't know if some functions introduced with the K6-2, would actually be used, in case you don't use them, clearly the BIOS should be modified, to update it, and add support to the new CPUs, I just wonder if there are other combinations, for the double jumper VCORE, it would be interesting if there was a selection for 2.4 or 2.5V, or even something less than 2.2V, it would be a nice surprise.

I return to the Chaintech 5TDM2 M101 190, I tried to power it directly from the ATX connector, despite this attempt no code is seen, so I would say that both work (AT and ATX), then however looking at the card, which is in excellent condition, I can rule out that there could be a bad contact.
Looking for damage in some track, there is a slight scratch near the drive connector, but honestly even if there was an interrupted track, I don't think it prevents the PC from starting, if anything it could give some problems with the drive, in short I would say that eventually it would be checked, only if there were problems with the drive, but certainly not now.
I check with the magnifying glass, I checked that the RAM pins are all straight, and in fact it is so, no pin is crooked and no one shows signs of oxidation, even here though, I think that even without RAM the PC should show codes, and then stop, in the Compaq Camaro it stops at 47, here being a BIOS AWARD the code should be different.
So I check the pins of the S.7, all the pins seem to be in order and in good condition, no pin shows twist, they are all well centered with respect to the hole, basically the copper pin divides the hole in two, so I would conclude by saying that everything is fine ✌️.
Reading the manual, the card has initial settings, which it uses at the first start, for CPUs with two voltages, they would be 110 MHz and 2.8V Core (3.3V I/O), the value of 110 is a bit strange, because it should be 2X55, but among the supported CPUs there are not those with FSB below 60, also there is apparently no support for K6 VCORE 2.2V, to be sure you should check from BIOS, if there are other values, in addition to those commonly used.
Finally, I tried without unsoldering an electrolytic capacitor, the values are strange, but it is possible that they are influenced by other components, they should be removed and checked one by one, the brands used are not known for quality, a dozen are G-CAP 1500uF 6.3V, and these would be the ones that could give some problems, there are other smaller ones but I don't think it's necessary to check them, they are in peripheral areas, so not important at least for starting the PC.

PC@LIVE wrote on 2025-10-08, 10:00:

Looking for info on some S.7, I found a possible secret setting, for an S.7 ACORP A-5VIA3P motherboard, currently on that board […]
Show full quote

Looking for info on some S.7, I found a possible secret setting, for an S.7 ACORP A-5VIA3P motherboard, currently on that board with VIA VPX chipset there is a Cyrix MII 333 (263 MHz 75X3.5), the performance is quite slow, I have tried other Intel CPUs, and I have never seen it go like other PCs (the same), probably by eye I would say at least 30% less, exactly it could be more or less, but I should do some bench, maybe 🤔 with those of Phil's I would find if there is something in particular, that can slow it down, for example the cache or the lack of driver.
Let's say that would be the fastest CPU (Cyrix) I can install, in fact if what I saw on TRW is correct, it might be possible to install faster CPUs, I would have found an interesting 🧐 secret setting of the VCORE, by using a second jumper, you could get a 2.2V VCORE, this would be right for AMD K6-2 CPUs and also the K6 266 and 300, in addition to the latest versions of the Cyrix MII, in short it would be possible to install a K6-2-400 or even a 450.
Of course I haven't done anything yet, because I would like to be sure that it doesn't create other problems, first, I should be sure that the regulator is switching, after the double jumper actually works, because it could be referred to a later version of the same MB.
In case it works, there could be a BIOS problem, currently there is the latest version dated 1997, easily a K6-2 would be recognized as simple K6, and the displayed speed could be lower, for example K6-300 instead of K6-2-400.
Surely it would work faster, but I don't know if some functions introduced with the K6-2, would actually be used, in case you don't use them, clearly the BIOS should be modified, to update it, and add support to the new CPUs, I just wonder if there are other combinations, for the double jumper VCORE, it would be interesting if there was a selection for 2.4 or 2.5V, or even something less than 2.2V, it would be a nice surprise.

Ciao Elio,

You keep presenting interesting boards. Luckily this Acorp A-5VIA3P does work! 😀

The undocumented 2.2V Vcore setting is indeed nice. When experimenting with other jumper combinations trying to find more hidden Vcore voltages, just use your Pentium MMX and set it to 2x50MHz. At this 100MHz speed, the Pentium MMX should keep running, even at 2.0V Vcore.

Unfortunately this board uses a linear Vcore regulator, so running a K6-2/400 at 2.2V will overheat and fry this regulator! A better option would be to use a K6-2+, provided the board has a hidden 2.1V or 2.0V Vcore setting. Or use an interposer.

The latest 12/02/97 Rev C2 BIOS does support the K6-2 up to 400MHz, however it will indicate it as a regular K6(tm) and doesn’t support the K6-2 Write-Allocation. I believe there is an Unicore BIOS upgrade for this board with full K6-2(+)/III(+) support. I will check.

Cheers, Jan

Chkcpu wrote on 2025-10-09, 14:04:

Ciao Elio, […]
Show full quote

PC@LIVE wrote on 2025-10-08, 10:00:

Looking for info on some S.7, I found a possible secret setting, for an S.7 ACORP A-5VIA3P motherboard, currently on that board […]
Show full quote

Looking for info on some S.7, I found a possible secret setting, for an S.7 ACORP A-5VIA3P motherboard, currently on that board with VIA VPX chipset there is a Cyrix MII 333 (263 MHz 75X3.5), the performance is quite slow, I have tried other Intel CPUs, and I have never seen it go like other PCs (the same), probably by eye I would say at least 30% less, exactly it could be more or less, but I should do some bench, maybe 🤔 with those of Phil's I would find if there is something in particular, that can slow it down, for example the cache or the lack of driver.
Let's say that would be the fastest CPU (Cyrix) I can install, in fact if what I saw on TRW is correct, it might be possible to install faster CPUs, I would have found an interesting 🧐 secret setting of the VCORE, by using a second jumper, you could get a 2.2V VCORE, this would be right for AMD K6-2 CPUs and also the K6 266 and 300, in addition to the latest versions of the Cyrix MII, in short it would be possible to install a K6-2-400 or even a 450.
Of course I haven't done anything yet, because I would like to be sure that it doesn't create other problems, first, I should be sure that the regulator is switching, after the double jumper actually works, because it could be referred to a later version of the same MB.
In case it works, there could be a BIOS problem, currently there is the latest version dated 1997, easily a K6-2 would be recognized as simple K6, and the displayed speed could be lower, for example K6-300 instead of K6-2-400.
Surely it would work faster, but I don't know if some functions introduced with the K6-2, would actually be used, in case you don't use them, clearly the BIOS should be modified, to update it, and add support to the new CPUs, I just wonder if there are other combinations, for the double jumper VCORE, it would be interesting if there was a selection for 2.4 or 2.5V, or even something less than 2.2V, it would be a nice surprise.

Ciao Elio,

You keep presenting interesting boards. Luckily this Acorp A-5VIA3P does work! 😀

The undocumented 2.2V Vcore setting is indeed nice. When experimenting with other jumper combinations trying to find more hidden Vcore voltages, just use your Pentium MMX and set it to 2x50MHz. At this 100MHz speed, the Pentium MMX should keep running, even at 2.0V Vcore.

Unfortunately this board uses a linear Vcore regulator, so running a K6-2/400 at 2.2V will overheat and fry this regulator! A better option would be to use a K6-2+, provided the board has a hidden 2.1V or 2.0V Vcore setting. Or use an interposer.

The latest 12/02/97 Rev C2 BIOS does support the K6-2 up to 400MHz, however it will indicate it as a regular K6(tm) and doesn’t support the K6-2 Write-Allocation. I believe there is an Unicore BIOS upgrade for this board with full K6-2(+)/III(+) support. I will check.

Cheers, Jan

Ciao Jan
Thanks for the compliments and suggestions, yes the card works, even if it has never shone for performance, quite modest for PCs like that, maybe it depends on the VIA VPX chipset, which should be the equivalent of the i430VX, or needs some driver, to get the maximum in performance.
First of all, I have to verify that indeed the version in my possession, is the same (as a jumper) to that of the manual with secret settings, it could be a previous version, but from memory I don't remember, if it was not possible to use the double jumper, the only way as you rightly suggest, would be the use of an interposer to select voltages of 2.2V or lower.
Actually I believe that the regulators are linear, and I think it is evident from the large heatsink present in one of the two regulators, but the circuit seems quite robust to me, it withstands a frequency of 263 MHz without giving problems, but in fact I have never tried faster CPUs, so I could be very close to the maximum.
Interesting, that the K6-2+ can be used instead of the normal one, but I believe that a low-VCORE K6-2 can also be used for laptops, if I remember correctly, there is one of about 400 MHz, which should be 9W, obviously 🙄 that then you have to find a way to get a VCORE lower than 2.0V, this if possible would avoid the use of an interposer (at least I think), provided that then the frequency of about 360 or 400 MHz, does not produce too much heat.
Very interesting 🧐, that the Pentium MMX can work at just over 2.0V, if you use it at only 100 MHz, it's a test I've never done, in the past I've only tried it at lower VCORE, from 2.2V up, but with low voltage it starts DOS, but using some bench programs, some didn't work.
It would be interesting 🧐 to know if there is a later version of the BIOS, currently as you can see from the image I have the C2 version of 1997, with this the Cyrix MII 333, it is seen as Cyrix 686MX 266, ditto with the Cyrix MII 300, so I wouldn't be surprised if the K6-2, was seen as a normal K6.
I think that C2 version has other limitations, like that of the disk capacity, which could be about 8 GB 🇬🇧, but maybe this would not be a difficult problem to solve, it remains to be seen if for example, the slowness of the PC is due to the transfer of the disk, for this I would have to use a special diagnostic program (?), I don't know if there is something suitable among the Phil's benches.
Greetings
Elio

Ciao Elio,

The latest Acorp A-5VIA3P 12/02/97 Rev C2 BIOS has indeed a 32GB HDD limit, and also suffers from the UDMA mode bug.
But I found the Unicore upgrade BIOS for this A-5VIA3P (VPX) board. Here is a copy.

This BIOS supports all Socket 7 CPUs ever made, fixes the 32GB HDD limit and has a fix for the UDMA mode bug as well.

I hope you will find some more hidden Vcore settings.

Cheers, Jan

Chkcpu wrote on 2025-10-10, 19:26:

Ciao Elio, […]
Show full quote

Ciao Elio,

The latest Acorp A-5VIA3P 12/02/97 Rev C2 BIOS has indeed a 32GB HDD limit, and also suffers from the UDMA mode bug.
But I found the Unicore upgrade BIOS for this A-5VIA3P (VPX) board. Here is a copy.

The attachment 2A5LDSMB.zip is no longer available

This BIOS supports all Socket 7 CPUs ever made, fixes the 32GB HDD limit and has a fix for the UDMA mode bug as well.

I hope you will find some more hidden Vcore settings.

Cheers, Jan

Ciao Jan
Thank you very much for the latest version of the BIOS, if I can these days, I will try it to see if updating, I can see some improvement, I believe that the UDMA bug can slow down the PC, I hope that this problem is solved the PC is better.
Before updating, I'll do some bench to see if and how much better after the update, regarding the K6-2 CPUs, I could try for example at only 300 MHz, using multi 6X and FSB 50, or I could try with the current settings, multi 3.5X and FSB 75 then 262.5 MHz, but the doubt remains about the voltage 😬 VCORE, I should go below 2.8V up to 2.2V, for this I should check on my MB, if it is possible to use the secret settings.
To get other VCORE, you should do some tests, for this reason I want to try the jumper with resistance, measure if it goes down and how much the VCORE, in that case I could do some experiments, as I had done on the SOYO SY-5BT, finding the intermediate values between 2.2V and 2.8V, then I would have to do the experimentation for the voltages below 2.2V, but I needed space on the bench to work on other MB, and I postponed.
Only the problem of the linear regulator remains, not long ago, someone said they solved the 3.3V instead of the 5V, it would be interesting if it does not require particular changes, in practice if I understand correctly, more heat is produced, when there is a difference in voltage 😬 greater, using the 3.3V instead the difference would be much smaller.
Greetings
Elio

I'll see if I can recover ❤️‍🩹, the PC with MB ACORP and Cyrix MII-333, I remember that for a short time it was my best PC, it happened at the end of the 90s, but it didn't last long because I switched in a short time to the Celeron 400 (also changing the motherboard), however that MB had not been chosen by me, I had bought a Powerboard VX, which worked very well for a short period, initially I had a K6-200 (on loan), then changed to a Cyrix MII-300, unfortunately 😣 with the arrival of the summer heat, the linear regulators must have heated up too much, maybe 🤔 an additional fan would have been needed, unfortunately he never came back from the service, and in return I got this ACORP with VPX chipset, which seems to be more robust, the linear regulators have worked, without ever giving problems, that PC has become for a few years, the stock PC without any problems, except a slowness due to I don't know what 😨.
In this regard, I could make a direct comparison with another motherboard, same chipset and using the same CPU, I later bought a ZIDA 5SVA-E, because I wanted to see if there was a difference in speed, apparently not 👎, maybe this one could be faster, but I don't think much, so I think I'll make a parallel comparison between motherboards, using two identical CPUs, in this case I would use two Cyrix MII-300 (3.5X 66), and via CF memory card with DOS, I will run some bench, this maybe I should do it before updating the BIOS of the MB ACORP.
The purpose of this comparison is to understand if the VIA VPX chipset can be improved in common use performance, and if it is slow, what is this slowing due to, I remember that on the VX the performance was much better, between AMD K6 200 and Cyrix MII-300, there was little speed difference (visually), while the Cyrix MII-300 on the VIA VPX, was definitely slow, sometimes it seemed to have stuck, it actually did something, maybe the buffers were full (?).

In the afternoon I downloaded the file, which contains the updated BIOS for the ACORP A-5VIA3P, I took a look at the content, and I saw that it recognizes, practically all the S.7 CPUs, of course you have to make sure that the VCORE is correct, if the board could not go down below a certain voltage, and considering that the regulators are linear, the use of an interposer would be ideal.
Undoubtedly if in addition to the support of new CPUs, there is also some bug fixes, I think it is worth using it, and replace the one currently present, I don't know it will make the motherboard faster, and if it does how much faster it will be, but if there are no problems, updating the BIOS could be the 😨 best thing.
Honestly, I don't know if the BIOS of this kind of motherboards, that is Socket 7, can be interchangeable, that is, if they have the same chipsets and various chips (I/O etc...), they can work on motherboards from different manufacturers.
The idea I've had over the years, would be that up to 233 MHz or a little more, you can use the Pentium MMX, if instead you want to go faster, with at least one K6-2 you can reach 400 or 450 MHz, using instead motherboards with FSB 100 or higher, you could get well over 500 MHz, and with 0.18u CPUs, you could also exceed 600 MHz, personally I have never tried to go to more than 550 MHz, also because I think there is no such difference, as to risk a CPU difficult to find, unfortunately as I wrote some time ago, K6+ with FSB 133 were not put on the market, assuming that then at that frequency no other problems are created.
So as soon as I can I prepare the ACORP VPX PC, and start to do some benches, initially with the Cyrix MII-333, then if I can lower the VCORE to 2.2V or about 2.5V, I'll switch to a K6-2 300, I don't want to exaggerate with the frequency, I'll start from the 263 MHz one (3.5X 75), just to see if it works, and if the linear regulators heat up (?), if I wanted later I could switch to 300 MHz (6X 50), but I don't know if it can create operating problems.

First of all, I recovered the PC with MB ACORP A-5VIA3P, before starting it I opened it, to see how it is configured.

The PC is in a case 🏘, classic AT Mini Tower, externally there are, three LED power HD turbo, two reset and turbo switches, plus a key lock, finally a switch for ignition with positions 0 and 1, for off and on.

Inside there is a 200W AT power supply, two 5.25" drives for CD and Floppy player, two 3.5" drives, one for the 1.44 MB floppy, the other empty, plus an additional 3.5" place, occupied by the WD Caviar 21600 HD, where Windows should be loaded, I think with ME memory, so I move on to the PCI cards present.

There are no ISA cards, I have all three free slots, the reason is that I only used three of the four PCI slots, if necessary I can use the one that remained free.

The video card is from Compaq, an S3 Virge GX I think 1 or 2 MB, more likely 2 MB, there are only two memory chips.

The other two PCI slots, are for a 10/100 LAN card with 3Com chip, and for a PCI sound card, with Ensoniq ES1370 chip, it should be the one before the purchase by SB.

There is no indication of which version the MB is, but from the silkscreens, the voltage jumper seems identical to the one with the secret settings, I would say that there are excellent possibilities that it works, using two jumpers I could be able to lower the VCORE to 2.2V.

I'll try this later, for now I do some checks, visually I don't see any inductor near the regulators, this would confirm that they are linear regulators, one has the acronym UTC D45H2, the other looks the same, so I would say without a doubt linear regulators.

I just finished trying to start the PC with MB ACORP A-5VIA3P and Cyrix MII-333, the first boot worked up to a certain point, then a blue Windows screen appeared, and at this point I don't think anything could be done, for this reason I turned off, and turned on the PC.

Starting I noticed a couple of interesting things, the first concerns the RAM, there is only one 128 MB bench, I would say that it is beyond what is necessary, considering the type of PC and Windows 98SE or ME (?), I hope they should run slightly better, compared to others with less RAM.

One thing I had never noticed, is the temperature of the second linear regulator, the one with the largest heat sink, is much hotter than the other, but considering that the temperatures in this period are cooler, it should be within the operating limits, also because if it exceeded the critical temperature, I would expect a shutdown or reset of the PC, and for the moment it has never happened.

It is also true, that in this configuration, with MII-333 I used it very little, on rare occasions, most of the work I did it when there was a MII-300, today it is a little difficult, to think of doing something with a PC so dated, even if many problems you had at the time, can be solved with technologies then not available or then expensive, let's see some problems:

For the RAM, if I wanted I could use more, change the bench with a 256 MB one, honestly I don't see the need, but certainly at the end of the century, a PC like that had between 16 and 32 MB of RAM, to have more than 32 MB it cost, then of course those who could afford it were free to expand it.

A second problem was that of the CPU change, if for example you had a 233 MHz (any), you could not change it with other CPUs, except the Cyrix MII-333, because the minimum VCORE was 2.8V, today instead thanks to the recent discovery on TRW (Thank you very much), I could instead think of installing a K6-2 400 or 450, even wanting a 500.

Unfortunately, even if it is possible to use 2.2V VCORE (via double jumper), having the linear regulators, I don't think it can work beyond a certain frequency, probably around 300 MHz, a second problem would be that of the (synchronous) frequency of the PCI, currently with 75 FSB I have 37.5 for the PCI, looking at the pdf of the frequency generator, which is also on TRW, there are six frequencies available, 50 55 60 66 75 and 83, but the same are also available in asynchronous mode, that is, the PCI would always go to 32 MHz, when the FSB is set between 50 and 83 MHz, So in total there are twelve possibilities.

This can slow down when the FSB is at 75 or 83 MHz, but it should improve when we have FSB 50 55 or 60, I don't know if the memories work at the same frequency as the FSB, I guess so, however this asynchronous mode must be tried, and it doesn't necessarily work, but it could solve any problems, using high FSBs like 75 or 83.

Finally a problem that has arisen, unfortunately the old WD Caviar 21600 disk, seems to be close to retirement, it works but after blocking with Windows blue screen, it says "disk not booting", a sign that some bytes must have been lost, I should try a floppy SYS, maybe it could be the MBR, but for the moment I disconnect it, and I go with the IDE-CF adapter with DOS, just to start the bench.

Then from there once I take note of all the scores, I update the BIOS with the one Jan found for me (Thank you very much).

Finished now to do the benchmarks with the PC MB ACORP A-5VIA3P and Cyrix MII-333, first of all with the FSB 75 it was not possible to make one, probably the synchronous operation, creates problems with the drives, or with the IDE-CF adapter, it is possible that the disk can be recoverable, but in this regard I have not done any test, in fact it is not a standard frequency, and I do not know if the WD disks have problems with the 75 MHz, however to make a long story short, I reduced the FSB to 66 MHz simulating a Cyrix MII-300 (233 MHz), and at this point, I ran Phil's bench.

First of all, I report the info on the BIOS AWARD readable when starting the PC:

ACORP 5VIA3P VER:C2 12/02/97

12/02/97-580VPX-SMC669-2A5LDSMBC-00

CPU Type: 6X86MX

CPU Clock: PR266 (instead of 333 or 300)

Let's move on to Phil's benches:

2) 296.1 3) 271.0 4) 40.4 5) 65.5 6) 22.7

A) 297.57 b) 79.88 c) 37.8 d) 17.0 e) 32.7

N) on CF card with DOS 6.22 it does not work, while with DOS 5.0 it goes but reads only 64 MB of RAM, maybe a limit of 5.0 (?), the score is 164.67

Finally I ran a diagnostic program SYSCHK 2.43, and the score obtained is a "RATING" of 410.7

At this point, I would say that I should only save the BIOS and then update it with the one that Jan linked me, maybe there could be variations in some bench.

But maybe first I should, try with FSB 75 asynchronous (PCI 32 MHz), I think it would solve a lot of problems, especially with the drives, maybe it won't go much faster, but the PC with about 30 MHz more, could get an estimable improvement of about 10%, then we'll see if it will be like that, but I think some will remain maybe the same, because the scores improve using more powerful VGA.

I did a test with the asynchronous 75 MHz FSB, that is, with PCI 32 MHz, unfortunately 😣 it didn't work, maybe the problem is caused by something not directly connected to the drives, but I don't know if the PCI actually works at 32 MHz, but the IDE drives always work at 37.5 MHz, so the asynchronous operation is only needed for PCI cards, while everything else continues to work at 1/2 of the FSB.
However, there would be an alternative, which I have never tried, would be the use of multi 4X instead of 3.5X, of course with FSB 66, the final frequency would be 266 MHz, about 3 MHz more, but everything depends on the MB, that I know on the Cyrix MII CPU the use of the multi 4X is provided, and in this ACORP 5VIA3P there is a JBF3 jumper to close, to get multi 4X or 4.5X, I think it also works with multi 5.0X and 5.5X, but on the pdf manual found online, the maximum is 4.5X, maybe it would be the limit of the MB, since the regulators are linear, however beyond the 4X the multi only work on the AMD type K6, even on the CXT there is no need to close the JBF3, but just select the jumpers on 2X to get a multi 6X.
So I would say that I will also do this test, FSB 66 and multi 4X, let's see how it goes, theoretically it should go at the same speed as PR333, so if it works I should see a small difference, measurable in about 10%, in any case interesting 🧐 because currently the CPU is under the normal operation clock.
Another thing 😨 interesting 🤨, looking at the pdf manual, there is a JBS jumper, I didn't know what it was for, nor what it was, in the past I tried to activate the "Linear Burst" from the BIOS, the PC crashed, the reason is that the JBS Jumper must be closed, so I have to check if it is open, I guess so, so I will try again to activate the "Linear Burst" from the BIOS, closing the JBS jumper, if it works I should gain something in speed, maybe 🤔 it may not be very visible, but from the bench I could see if there is a difference.

I performed some tests, the results are not what I hoped for, the first one I performed is the multiplier change, the settings for the 4X are the same as the 2X except for the JBF3 (closed or open).
Currently the JBF3 is open, so I try to set the jumpers for the 2X, and start the PC, the CPU is detected as 6X86MX PR166, but the real frequency is 133 MHz, all in all it works, although slowly obvious 🙄
With this setting I run the Phil's benches, some scores I expected to be much lower, but most are relatively close to those of the MII-300, for the record they are these:
2) 187.0 3) 104.7 4) 29.1 5) 42.5 6) 17.4
a) 240.16 b) 65.06 c) 24.6 d) 11.3 e) 21.9

The next step is to try the multi 4X, so I close the JBF3, and start the PC, touch the dis of the linear controller, I feel that it heats up quickly, it seems to work I would say, on the screen I read as CPU, MMX-33 MHz, but it crashes to the recognition of the drives, I try again, so I can imagine 💭 that it works, but there is some problem, of course it could be a problem of non-recognition of the CPU (?), but in the past in similar situations, the PC has started anyway.
The next attempt I can make would be the following, put the FSB at 60 and try again with the multi 4X, I would get a frequency of 240 MHz, so quite similar to the one currently in use (233), if it works it means that the multi 4X works, but there is something that creates problems even at 66 MHz, since with the initial settings, 3.5X and 75 MHz I had problems with the drives, it is possible that the IDE CF to 75 MHz adapter has operating problems.
I think I just have to try, the active Linear Burst, and then I'll see to update the BIOS with the one provided by Jan, if the Linear Burst could improve performance by about 5%, who knows if the correct BIOS can make something else work better, we'll see from the bench if there will be differences.

Ciao Elio,

I did a quick check on your original Acorp A-5VIA3P 12/02/97 Rev C2 BIOS, and it doesn’t support the Cyrix 6x86MX/MII when set to a x4 multiplier. I see this a lot in these late 1997/ early 1998 BIOSes.

The Cx6x86MX changes its CPU Model ID in the Device Identification Register 0 (DIR0) for each supported multiplier setting. The BIOS uses this register to identify the CPU and its multiplier setting, but the Cyrix ID table in your Rev C2 BIOS doesn’t go further than the ID for x3.5.
So when set to x4, the BIOS doesn’t recognize the CPU and can’t initialize it correctly.

The Unicore BIOS upgrade does support Cyrix x4 multiplier correctly. 😀

Cheers, Jan