I found an undocumented jumper setting for this board. 😀

For Cyrix CPUs, jumpers JP1 and JP2 have only documented settings for multipliers of 2x and 3x (1-2, 2-3 and 1-2, 1-2 respectively) The other 2 combinations for JP1 and JP2 (2-3, 2-3 and 2-3, 1-2) are undefined for Cyrix CPUs.

So of course I had to try them. 😀

Bad news first: with JP1=2-3 and JP2=1-2, the motherboard does not POST at all at any FSB with a 6x86.

And good news: JP1=2-3, JP2=2-3 is read by the Cyrix 6x86 as 1x. 😀 So the CPU speed is the same as the FSB. This allowed me to underclock the 6x86 down to 50-83MHz. Yes, it is stable at 83FSB x1.

stamasd wrote on 2022-01-22, 14:48:

I found an undocumented jumper setting for this board. :) […]
Show full quote

I found an undocumented jumper setting for this board. 😀

For Cyrix CPUs, jumpers JP1 and JP2 have only documented settings for multipliers of 2x and 3x (1-2, 2-3 and 1-2, 1-2 respectively) The other 2 combinations for JP1 and JP2 (2-3, 2-3 and 2-3, 1-2) are undefined for Cyrix CPUs.

So of course I had to try them. 😀

Bad news first: with JP1=2-3 and JP2=1-2, the motherboard does not POST at all at any FSB with a 6x86.

And good news: JP1=2-3, JP2=2-3 is read by the Cyrix 6x86 as 1x. 😀 So the CPU speed is the same as the FSB. This allowed me to underclock the 6x86 down to 50-83MHz. Yes, it is stable at 83FSB x1.

That’s wild, I had no idea that any socket5/7 cpus supported 1x

I had some fun recently with a 400gp figuring out it’s jumper settings
(Solved) Cyrix 400gp/366gp multiplier settings

stamasd wrote on 2022-01-20, 19:20:

Garrett W wrote on 2022-01-20, 19:06:

The 83MHz FSB bus however is another story, most devices won't be able to handle it. The PCI bus runs way too high at that speed courtesy of lacking a proper divider.

Like I mentioned above, this board's clock generator has the ability to run the PCI clock at a constant 32MHz independent of the FSB. It needs a slight hardware mod to enable that. I will get around to that, eventually. It's in the datasheet I posted a link to in the first post.

Frankly, what makes this motherboard interesting above others is that particular clock chip. With the option to run PCI async.

I have read the first pages of this thread with interest, in particular I find asynchronous PCI very interesting, I have no idea if you managed to get it to work, but in case it would be great, also because at high frequencies of FSB it would not cause problems for PCI cards, it would remain to be understood whether the controller also goes asynchronous or not.
Still on the subject of stability, I think it is normal that there may be some problems using the FSB 83, could it be due to the cache? Could the integrated chips in the MB be not fast enough? Have you tried disabling the L2 cache, to see if it works smoothly, obviously it will go a bit slower, but it would be interesting to understand exactly what causes the problem.
I created a page where I am collecting info on i430TX with FSB over 83MHz, a single motherboard starts with FSB 90, while an old ASUS MB with intel chipset has a non-working FSB over 100MHz, this however without modifying and replacing the quartz , it's here:
Re: Intel430 FSB settings beyond 83MHz (including overclocking?)

FWIW, the Cyrix 200 was stable at 83FSB, multiplier 1x. The only card in the system was a Matrox G200 PCI.

So how low does it go? Down to 50MHz?

Yes it goes down to 50 MHz. It's weird to see on the POST screen "6x86L at 50MHz" but it works.

stamasd wrote on 2022-01-22, 14:48:

I found an undocumented jumper setting for this board. :) […]
Show full quote

I found an undocumented jumper setting for this board. 😀

For Cyrix CPUs, jumpers JP1 and JP2 have only documented settings for multipliers of 2x and 3x (1-2, 2-3 and 1-2, 1-2 respectively) The other 2 combinations for JP1 and JP2 (2-3, 2-3 and 2-3, 1-2) are undefined for Cyrix CPUs.

So of course I had to try them. 😀

Bad news first: with JP1=2-3 and JP2=1-2, the motherboard does not POST at all at any FSB with a 6x86.

And good news: JP1=2-3, JP2=2-3 is read by the Cyrix 6x86 as 1x. 😀 So the CPU speed is the same as the FSB. This allowed me to underclock the 6x86 down to 50-83MHz. Yes, it is stable at 83FSB x1.

This is very interesting! I wonder which Cyrix CPUs can be set to 1x this way.
And nice to see this as a potential real purpose for these otherwise somewhat unremarkable CPU-curiosities 😜

So far I tested two 6x86 (one L and one non-L), specifically a 6x86-P166+GP and a 6x86L-P200+GP. It works for both. I also have an IBM 6x86L-166 which I haven't tested yet (because it needs a lot of work unbending pins) but I expect it to work also.

TBH I don't understand how it can possibly work. I'm going through the 6x86 datasheet at http://datasheets.chipdb.org/IBM/x86/6x86/6x86_ALL.pdf and this CPU only has one multiplier control pin: CLKMUL (page 3-7 in that sheet). It can only have 2 states: low or unconnected = multiplier 2x, and high = multiplier 3x. I have no idea how this board gets those CPUs to work at 1x.

Ah I think it starts making a little more sense. I was pointed to this thread from many years ago Subtle differences between Cyrix and IBM 6x86s where 1x for 6x86 was documented. The table in the first post refers to pins BF0, BF1 and BF2, which are not named so on the 6x86 according to the datasheet. But comparing the pinouts on the 6x86 datasheet ( http://datasheets.chipdb.org/IBM/x86/6x86/6x86_ALL.pdf page 5-1) and the Pentium MMX ( https://www.arl.wustl.edu/projects/archive/gi … X_Datasheet.pdf page 9) it appears that pins BF0 (location Y-33 in the latter) and BF1 (X-34) correspond in the 6x86 respectively to the pin CLKMUL and one "reserved" pin.

So if that "reserved" pin is indeed connected internally and influences the multiplier value, the findings above make more sense. And if the other combination of the 2 pin signals causes a 4x multiplier to be selected, it makes sense for the board not to POST in that case, because the minimum CPU speed would then be 50x4=200MHz, which I already know the CPUs I tested don't support (it didn't POST at 66x3 either).

Hmm yes and there is indeed a difference between the Cyrix-branded and the IBM-branded 6x86 chips. According to that table, the jumper combination that results in a 1x multiplier on Cyrix chips should give a 2x multiplier on IBM chips.
And practically, after unbending the pins on the IBM 6x86L-PR166+ and installing it with FSB=50 and JP1=JP2=2-3, it POSTS as a "6x86L-PR120+" which means it's running at 100MHz (not 50 as the Cyrix chip did). Still cool, but that means 50MHz is not achievable on this board with IBM CPUs, because there is no jumper to control BF2 and that would be the only method to get 1x for IBM.

More fun. I was able to set multiplier 1x on the IBM 6x86L. 😀

A little experimenting showed that jumper JP6, which is "reserved" actually controls BF2. With the FSB set at 50, JP1=JP2=JP6 in position 2-3, the multiplier was set to 1x on the IBM CPU and it POSTed at 50MHz.
So now we know what JP6 is for.

JP1=BF0; 1-2="1", 2-3="0"
Jp2=BF1; 1-2="1", 2-3="0"
JP6=BF2; 1-2="1", 2-3="0"

This will allow setting any supported multiplier on any socket7 CPU.

...for instance on K6-2 CPUs according to this table taken from the datasheet https://www.amd-k6.com/wp-content/uploads/201 … p_Datasheet.pdf page 92

stamasd wrote on 2022-01-23, 14:12:

Hmm yes and there is indeed a difference between the Cyrix-branded and the IBM-branded 6x86 chips. According to that table, the jumper combination that results in a 1x multiplier on Cyrix chips should give a 2x multiplier on IBM chips.
And practically, after unbending the pins on the IBM 6x86L-PR166+ and installing it with FSB=50 and JP1=JP2=2-3, it POSTS as a "6x86L-PR120+" which means it's running at 100MHz (not 50 as the Cyrix chip did). Still cool, but that means 50MHz is not achievable on this board with IBM CPUs, because there is no jumper to control BF2 and that would be the only method to get 1x for IBM.

There may be a multiplier lock implemented. I noticed that on pentium chips that I recently tested that some would remap lower multipliers if the upper multipliers had been messed with

Some would do 1.5x, some would do 2x on 2x setting etc.

stamasd wrote on 2022-01-23, 15:23:

More fun. I was able to set multiplier 1x on the IBM 6x86L. :) […]
Show full quote

More fun. I was able to set multiplier 1x on the IBM 6x86L. 😀

A little experimenting showed that jumper JP6, which is "reserved" actually controls BF2. With the FSB set at 50, JP1=JP2=JP6 in position 2-3, the multiplier was set to 1x on the IBM CPU and it POSTed at 50MHz.
So now we know what JP6 is for.

JP1=BF0; 1-2="1", 2-3="0"
Jp2=BF1; 1-2="1", 2-3="0"
JP6=BF2; 1-2="1", 2-3="0"

This will allow setting any supported multiplier on any socket7 CPU.

Excellent! 😁

Hi stamasd, I just got the same motherboard like you and really like it for SDRAM and MMX suppot, but I have some difficulties with MB so was trying google and found your topic. I'll be glad if you or anybody else can help...

The problem is that all time I restart the computer and doesn't matter if with reset button, CTRL+ALT+DEL or just by exit from BIOS the PC won't boot up again and finish with black screen. I have to power off and power on again... I can see I have same bios version so it is not the case, any idea what could be wrong?

Thanks!

koroner wrote on 2022-03-16, 15:34:

Hi stamasd, I just got the same motherboard like you and really like it for SDRAM and MMX suppot, but I have some difficulties with MB so was trying google and found your topic. I'll be glad if you or anybody else can help...

The problem is that all time I restart the computer and doesn't matter if with reset button, CTRL+ALT+DEL or just by exit from BIOS the PC won't boot up again and finish with black screen. I have to power off and power on again... I can see I have same bios version so it is not the case, any idea what could be wrong?

Thanks!

I’ve had some add in cards cause this before.

Sphere478 wrote on 2022-03-16, 21:41:

I’ve had some add in cards cause this before.

Unfortunately not the issue, only graphic card is installed so I tried different one in different slot but still same, also tried boot without FDD and HDD, changed simm to SDRAM and seems like MB issue..

Strange is that if I reset before "Award Plug and Play BIOS Extensions" shows up the computer restarts normaly.... any idea?

snufkin wrote on 2022-01-20, 21:08:

I know some Intel chipsets don't support async, so it'll be interesting to know if that works when you can try it. […]
Show full quote

stamasd wrote on 2022-01-20, 19:20:

Like I mentioned above, this board's clock generator has the ability to run the PCI clock at a constant 32MHz independent of the FSB.

I know some Intel chipsets don't support async, so it'll be interesting to know if that works when you can try it.

[edit: I've marked on the photos what I think are the pins you've found (obviously waiting for when you can check the PS/2). Do they look right?]

[edit: I've updated the COM port and removed the previous picture, hopefully that's now correct]

Belated reply. I finally got around to working on this motherboard again.
I made a custom adapter for the PS/2 port. Tried with both pin 1 = DATA and pin 3 = CLK, and the other way around. None of the 2 configurations work, a PS/2 mouse is not detected/usable with either of them. I used pin 4 = GND and pin 5=Vcc in both cases.

This is an old topic, but in case someone else struggles to find the correct mouse pinout, then my PCPartner
TXB820DS (35-8333-01) has the following pinout:

1 - Data
2 - Ground
3 - CLK
4 - NC
5 - +5V

Numbers correspond to the op markings.

I know the OP has tested this variation, but maybe his was broken, because I don't believe 35-8333-01 and 35-8333-03 have different pinouts as the layout is otherwise the same.

Zida Tomato C-Series Mainboards Cable Set has a mouse cable that has 5 pins in a row, but just need to be rearranged. If you have one of those, then the colors would be:

1 - Data (Blue)
2 - Ground (Yellow)
3 - CLK (Red)
4 - NC (empty)
5 - +5V (Green)