According to page 6 of this: https://www.onsemi.com/pdf/datasheet/cs5165-d.pdf then the CS5165 does support the full range, including 1.65V.

Thanks for looking over the datasheet snufkin. I had pulled that document up, but I didn't see anywhere in the bullet points at the top summary of what voltage ranges it supported. Wasn't sure what I was looking at, but if it was listed in the DAC codes then it would be good to go, thanks for confirming.

I am going to see if I can get myself a 1.65v CPU since they are a bit more common in the higher speeds and see if this will work albeit at a 100MHz FSB since it looks like this board doesn't support overclocking the FSB.

Get Pentium III 1Ghz and run it at 7.5x100. There are also 100Mhz versions of Pentium III but those are very rare. Alternatively get a Coppermine Celeron 633 and run it at 950Mhz.

can you post a photo of the 28 pin chip near the CMOS battery and the 14.318 crystal, or just give the part number? It's possible it has some undocumented frequency outputs.

AlexZ wrote on 2022-04-01, 18:57:

There are also 100Mhz versions of Pentium III but those are very rare.

They can't be that rare as I have a system 2 x Intel Pentium IIIE (Coppermine) with an FSB of ~100MHz @ ~1.00GHz .

Here is a picture of the 28 pin chip in question, and another 28 pin chip that probably isn't related, but including it as well.

Looks like the clock gen will do other frequencies, but only if it's set via software (maybe setfsb or softfsb? Looks like they can be used on 440BX chipsets). But it might need a bit of hacking to make it work, since I think R225 and R226 will need to be fitted to connect the clock gen chip to the i2c bus. I guess the board designers allowed for either the simple switch between 66 and 100 via R170 (probably some setting in the BIOS?), or the more complicated i2c setting, and then choose which was used by choosing which resistors to fit.

Thanks Snufkin for looking into the FSB freq. I managed to find a 700/100 slot 1 cpu at 1.65v, so will report back when it gets here. I may attempt a fsb mod in the future, but will start out at 100.

If it's a clone of BCM QS440BX, looks the same as https://soggi.org/motherboards/bcm/QS440BX.htm then probably BIOS from that page will work as well. It declares support for Pentium III up to 1Ghz. These slot 1 boards tended to support only 100 FSB officially. I would not count on anything above 100 FSB being stable. On my ECS board 133 FSB boots into Windows 98 but 3d Mark crashes. 112 FSB passes prime95 and 3d Mark in a loop but fails fairly quickly when decompressing big rar archives (a few GB) in Windows 98. Maybe it was stable when the board was new but on these old boards I wouldn't count on anything else than what was supported officially.

AlexZ wrote on 2022-04-03, 15:40:

If it's a clone of BCM QS440BX, looks the same as https://soggi.org/motherboards/bcm/QS440BX.htm then probably BIOS from that page will work as well. It declares support for Pentium III up to 1Ghz. These slot 1 boards tended to support only 100 FSB officially. I would not count on anything above 100 FSB being stable. On my ECS board 133 FSB boots into Windows 98 but 3d Mark crashes. 112 FSB passes prime95 and 3d Mark in a loop but fails fairly quickly when decompressing big rar archives (a few GB) in Windows 98. Maybe it was stable when the board was new but on these old boards I wouldn't count on anything else than what was supported officially.

Does anyone know how the PCI bus divider works on 440BX chipsets. I remember ~2000, plenty of people were running 440BX boards with EB model Pentium 3s, but I'm not sure if the chipset itself sets the 1/4 divider automatically, or if you need a BIOS that's configured for it.

mattferris@hotmail.c wrote on 2022-04-04, 03:33:

Does anyone know how the PCI bus divider works on 440BX chipsets. I remember ~2000, plenty of people were running 440BX boards with EB model Pentium 3s, but I'm not sure if the chipset itself sets the 1/4 divider automatically, or if you need a BIOS that's configured for it.

Early 440BX slot1 boards had 3x ISA, those are 100Mhz FSB stable and support only Katmai PIII
Mid 440BX slot1 boards had 2x ISA, 100Mhz FSB stable and support Coppermine PIII - the best CPUs for them are PIII E versions
Late 440BX socket 370 boards had 1x ISA, 133Mhz FSB stable, Coppermine support - AGP will be overclocked but it should be fine with Nvidia Geforce

Divider setting should be automatic. Having 133 FSB in BIOS doesn't mean anything. Older revision of my ECS board has 133 option in BIOS but it doesn't boot with it at all. If you want 133Mhz FSB, pay extra for a name brand late 440BX board. I didn't, because it doesn't matter that much for games from 2001 onwards where Athlon XP/64 is a much better choice.

mattferris@hotmail.c wrote on 2022-04-04, 03:33:

Does anyone know how the PCI bus divider works on 440BX chipsets. I remember ~2000, plenty of people were running 440BX boards with EB model Pentium 3s, but I'm not sure if the chipset itself sets the 1/4 divider automatically, or if you need a BIOS that's configured for it.

There are possibly 440BX motherboards that allow choosing the right divider via BIOS but I don't know which ones that would be. More common is, I think, that the right combination can be set via jumpers or dipswitches. The options depend on the motherboard configuration and on the fsb frequency chip that is installed. Here an example of two ASUS boards:

It sounds like the 440BX chipset is not involved in setting the PCI bus speed then, based on what you're saying. It happens externally via some other motherboard mechanism (jumpers in the case you showed above). My memory was a little rusty (it's been 20 years). I was under the impression that the chipset would just pick either 1/3 or 1/4 divider based on which speed of FSB was picked. Yours makes more sense, since it's not clear how the chipset would choose a divider for a weird FSB setting like 112.

Besides that, when you look at the second part of the above overview you'll notice that there are two options for fsb 133, one with 1/3 and one with 1/4 divider. Therefore I do not think that the chipset is able to 'choose' at all - it is likely fully dependant on the signal from the frequency generator.
As an aside, I have a motherboard with the ICS9250-08 chip but the jumper configuration on the motherboard only allows, unfortunately, to address half of the options that the frequency generator offers. The options look like this:

So it is not only the chip that counts, you need some luck too.

That table explains why the 133 setting is not always stable - early and mid 440BX boards feature max FSB/3 PCICLK and PCI would run at 44Mhz. 100Mhz FSB selects /3 and 66Mhz FSB selects /2 PCICLK divider. In the table above FS2 also selects PCICLK divider. Due to this simplification it cannot support the other half of frequencies as separate jumper would be needed to select the divider.

You may be able to select those frequencies via SoftFSB or a similar tool as long as PCICLK divider remains the same.

Got the 1.65volt coppermine and happy to report it indeed works without any issues. Thanks for everyone that chimed in. Appreciate this community very much.