The "133" in KT133 refers to the supported RAM clock, not the FSB. KT133 only supported a (double-pumped) 100MHz FSB.

KT133A was the first chipset for SoA that supported 133MHz/266MT/s FSB.

The reason this board has the silkscreen for bus speed jumpers is that it is a late KT133 board probably produced in parallel with KT133A boards. KT133A was pin-compatible with KT133, so they could use the same PCB.

Of course sometimes stuff can be clocked past what it officially supports (eg i440BX at 133MHz FSB or higher), but I've not heard of KT133 chipsets being pushed to 133MHz. This is a great board for the B-model Athlon Thunderbird CPUs and the early Durons, but not for CPUs wanting faster bus speed.

Speaking of CPUs, if you have an Athlon XP with 15x multiplier, it will be at least a Thoroughbred core. That adds an additional complication: the Tbreds use a lower core voltage than the older Thunderbird and Palomino cores. The motherboard needs to be able to deliver the 1.5 or 1.6V (vs 1.75 for the older ones) and I doubt this board can, although I can't find any documentation as all SoA Athlon/Duron CPUs at the time of its release ran at the same 1.75V voltage. The last BIOS update for the board was from February 2001, so before the release of AthlonXP CPUs. It would be safe to experiment with a Palomino, but a Tbred might not be such a good idea...

dionb wrote on 2023-06-23, 09:47:

The "133" in KT133 refers to the supported RAM clock, not the FSB. KT133 only supported a (double-pumped) 100MHz FSB. […]
Show full quote

The "133" in KT133 refers to the supported RAM clock, not the FSB. KT133 only supported a (double-pumped) 100MHz FSB.

KT133A was the first chipset for SoA that supported 133MHz/266MT/s FSB.

The reason this board has the silkscreen for bus speed jumpers is that it is a late KT133 board probably produced in parallel with KT133A boards. KT133A was pin-compatible with KT133, so they could use the same PCB.

Of course sometimes stuff can be clocked past what it officially supports (eg i440BX at 133MHz FSB or higher), but I've not heard of KT133 chipsets being pushed to 133MHz. This is a great board for the B-model Athlon Thunderbird CPUs and the early Durons, but not for CPUs wanting faster bus speed.

Speaking of CPUs, if you have an Athlon XP with 15x multiplier, it will be at least a Thoroughbred core. That adds an additional complication: the Tbreds use a lower core voltage than the older Thunderbird and Palomino cores. The motherboard needs to be able to deliver the 1.5 or 1.6V (vs 1.75 for the older ones) and I doubt this board can, although I can't find any documentation as all SoA Athlon/Duron CPUs at the time of its release ran at the same 1.75V voltage. The last BIOS update for the board was from February 2001, so before the release of AthlonXP CPUs. It would be safe to experiment with a Palomino, but a Tbred might not be such a good idea...

In the manual, it has a chart showing the jumper settings for the board. One of them is 133 and the chart even shows the PCI bus speed is a normal 33.3 at 133. Maybe it was a cut and paste mistake by biostar? That still doesn't explain why there are photos of this board with the header present unless that is the wrong board pictured on retroweb. I kind of think it might have been on there originally but was removed in favor of adjusting the fsb in the bios.

I have a thoroughbred sempron in it at the moment. It is working just fine though the bios has no clue what it is. It is a 1.6v cpu and the one I bought is 1.65v. You think I'll be alright with that? I don't have the cpu yet, or I'd just try it out and see. Originally it had a D750AUT1B which is a 1.6v duron. I put the sempron in because I had it lying around and was like "why not?"

I think I've picked just the right board to confuse myself pretty well. It was just so cheap though. They didn't even have it listed as a socket A board. They actually had it listed as an intel model.

EDIT I just looked at the photos again, and it is marked M7VKB, so its the same model. I guess they did drop the jumpers in favor of using the bios.

here's the retroweb page for the board, so you can see what I've been looking at.
https://theretroweb.com/motherboards/s/biosta … m7vkb#downloads

I have two of these boards. They didn't work with Thoroughbreds and I don't have any Palominos. There is jumper for 133FSB but this frequency doesn't work. I have bought two Thunderbird 1200B specifically for these boards. There is also Thunderbird 1400B, but I wasn't sure if it would run due to supposed maximum 12.5 multiplier supported by KT133 boards. These boards are a slightly better alternative to 440BX, but are very rare. CPUs above 1200Mhz start to benefit from DDR memory and there aren't Windows 98 games that can benefit from a faster system. 1400Mhz is Windows XP domain. There is little point in using a faster CPU in KT133. KT133A boards, released in early 2001 are even rarer than KT133. They became obsolete in less than a year with KT266A released in late 2001.

AlexZ wrote on 2023-06-23, 17:19:

I have two of these boards. They didn't work with Thoroughbreds and I don't have any Palominos. There is jumper for 133FSB but this frequency doesn't work. I have bought two Thunderbird 1200B specifically for these boards. There is also Thunderbird 1400B, but I wasn't sure if it would run due to supposed maximum 12.5 multiplier supported by KT133 boards. These boards are a slightly better alternative to 440BX, but are very rare. CPUs above 1200Mhz start to benefit from DDR memory and there aren't Windows 98 games that can benefit from a faster system. 1400Mhz is Windows XP domain. There is little point in using a faster CPU in KT133. KT133A boards, released in early 2001 are even rarer than KT133. They became obsolete in less than a year with KT266A released in late 2001.

Thank you so much! I knew I'd get some answer here. The board isn't quite what I'd thought, but its a keeper for sure! I'm really looking for something speedy that still has that AGP and ISA slot for testing bulk lots of expansion cards. The less time I spend waiting for reboots and installation wizards, the better. Got any suggestions what board to use? I was considering a MSI MS-6571 VER:2. It has AGP, PCI, ISA and they're readily available on aliexpress. They're socket 478, so it is a much speedier option. It seems ideal, but I bet there's a catch.

Itisnot1984 wrote on 2023-06-23, 14:02:

[...]

In the manual, it has a chart showing the jumper settings for the board. One of them is 133 and the chart even shows the PCI bus speed is a normal 33.3 at 133. Maybe it was a cut and paste mistake by biostar? That still doesn't explain why there are photos of this board with the header present unless that is the wrong board pictured on retroweb. I kind of think it might have been on there originally but was removed in favor of adjusting the fsb in the bios.

Having the settings does not mean that it has a chance of working. They do not recommend anything over 100MHz,

Biostar manufactured for various OEMs, an OEM board would have settings that users could mess up removed, like FSB speed jumpers.

I have a thoroughbred sempron in it at the moment. It is working just fine though the bios has no clue what it is. It is a 1.6v cpu and the one I bought is 1.65v. You think I'll be alright with that? I don't have the cpu yet, or I'd just try it out and see. Originally it had a D750AUT1B which is a 1.6v duron. I put the sempron in because I had it lying around and was like "why not?"

Well, those voltages might be a reason why not.

One thing though: which board revision is it? Sometimes later revisions might be able to do more than earlier ones. Failing that, can you look up the VRM chip on the board? If it can do 1.6V, you should be OK.

I think I've picked just the right board to confuse myself pretty well. It was just so cheap though. They didn't even have it listed as a socket A board. They actually had it listed as an intel model.

Your only confusion is wanting to run later CPUs on it than it was designed for. Nothing wrong with that in principle, but you need to know what you are doing and what is and isn't feasible.

dionb wrote on 2023-06-25, 00:21:

Having the settings does not mean that it has a chance of working. They do not recommend anything over 100MHz, […]
Show full quote

Itisnot1984 wrote on 2023-06-23, 14:02:

[...]

In the manual, it has a chart showing the jumper settings for the board. One of them is 133 and the chart even shows the PCI bus speed is a normal 33.3 at 133. Maybe it was a cut and paste mistake by biostar? That still doesn't explain why there are photos of this board with the header present unless that is the wrong board pictured on retroweb. I kind of think it might have been on there originally but was removed in favor of adjusting the fsb in the bios.

Having the settings does not mean that it has a chance of working. They do not recommend anything over 100MHz,

Biostar manufactured for various OEMs, an OEM board would have settings that users could mess up removed, like FSB speed jumpers.

I have a thoroughbred sempron in it at the moment. It is working just fine though the bios has no clue what it is. It is a 1.6v cpu and the one I bought is 1.65v. You think I'll be alright with that? I don't have the cpu yet, or I'd just try it out and see. Originally it had a D750AUT1B which is a 1.6v duron. I put the sempron in because I had it lying around and was like "why not?"

Well, those voltages might be a reason why not.

One thing though: which board revision is it? Sometimes later revisions might be able to do more than earlier ones. Failing that, can you look up the VRM chip on the board? If it can do 1.6V, you should be OK.

I think I've picked just the right board to confuse myself pretty well. It was just so cheap though. They didn't even have it listed as a socket A board. They actually had it listed as an intel model.

Your only confusion is wanting to run later CPUs on it than it was designed for. Nothing wrong with that in principle, but you need to know what you are doing and what is and isn't feasible.

The manual confused me because the chart in it shows different fsb settings and what the PCI bus will be for each FSB speed. According to the chart, the PCI bus goes up in value with the FSB but then goes back to normal when set to 133. I don't know a lot about this, but I remember the PCI bus being a limiting factor for fsb overclocking. I figured because the chart showed a normal pci bus at 133 that it wasn't actually an overclock and could work. Definitely not the case.

The voltages on the original duron match the sempron I already owned. I figured its 1.6v and the board can clearly do 1.6 since that is the voltage for the duron that was original to it. I am not sure about the 1.65v athlon xp I ordered though. Is 5 hundredths of a volt going to kill it, or could the board give a much higher than expected voltage? Also, it is a 15x multiplier so apparently it might not work anyway.

Its a later revision 2.0.

It's the KT133 northbridge chip that is the limiting factor, it does not go beyond 105-110 MHz FSB in most cases. The PCI bus hitting the speed limit happened one generation later with the KT133A which could often achieve relatively high overclocks of FSB 140-150 over the default 133, but the PCI and AGP buses were overclocked alongside it. The PCI speed is always set to 33 MHz for each "milestone" FSB speed that the chipset supports (100, 133, 166 and so on), and is always overclocked when you go beyond the FSB between these specs.

All Socket A boards should support voltages from 1.5-1.75v by design, these were the voltages required by the first generation of processors (Athlon and Duron). 1.65v is not much for an Athlon XP, there were some lower clocked ones that ran at 1.5v, but that was to save power. They would still run within specifications at 1.65v. The circuitry that is responsible for setting the core voltage is backwards compatible, so newer processors will still run at their correct voltage on older boards.

Multipliers do not need to be explicitly supported by a board, the processor does this internally. It only becomes relevant if the board offers to modify the multiplier, in these cases, the multiplier will be set to wrong values or cannot be changed at all if the multiplier range is not supported.

asdf53 wrote on 2023-06-25, 12:29:

The PCI speed is always set to 33 MHz for each "milestone" FSB speed that the chipset supports (100, 133, 166 and so on), and is always overclocked when you go beyond the FSB between these specs.

This is why I thought 133 could work. The chart showed there would be a pci bus speed of 33.3 when at 133 fsb. I don't know why they put this in the chart like that. Perhaps a copy/paste mistake or they were intentionally misleading.

As for the athlon xp I ordered, I'll just have to wait and see. It is international shipping, so it could be a while. Thank you for the reply.

Definitely misleading, but it's probably just laziness. My guess would be that this section was written for an upcoming KT133A board. Because the chipsets were so similar they could re-use the board design and 99% of the manual. Some manufacturers used the same manual for their KT133 and KT133A boards, the only difference being a section about the new supported 133 MHz FSB settings.

Itisnot1984 wrote on 2023-06-25, 21:27:

[...]

This is why I thought 133 could work. The chart showed there would be a pci bus speed of 33.3 when at 133 fsb. I don't know why they put this in the chart like that. Perhaps a copy/paste mistake or they were intentionally misleading.

No, the 1/4 PCI divider is there in the KT133, so there's nothing misleading about it. It's just not a lot of use given other limitations.

" I was considering a MSI MS-6571 VER:2. It has AGP, PCI, ISA and they're readily available on aliexpress. They're socket 478, so it is a much speedier option. It seems ideal, but I bet there's a catch."

Just got mine in the mail, can confirm SB Awe64 ISA works properly from a quick test in Doom, audio and Midi after disabling the onboard. No onboard vide, have a X800 XT in there 😀

I got my cpu in today. It is the AXDA2400DKV3C, and it is working. It is running at 1500 mhz with a 15x multiplier. This is a good result, so I'll continue on with the project.

I'm still thinking bout that ms 6571 motherboard and may switch to it, if there are issues. Thanks everyone who replied to help.

asdf53 wrote on 2023-06-26, 05:30:

Definitely misleading, but it's probably just laziness. My guess would be that this section was written for an upcoming KT133A board. Because the chipsets were so similar they could re-use the board design and 99% of the manual. Some manufacturers used the same manual for their KT133 and KT133A boards, the only difference being a section about the new supported 133 MHz FSB settings.

KT133 was presented as a 133MHz FSB chipset before launch, the decision to cap it at lower speed was basically a last minute one. IIRC Via came up with some silly excuse about wanting the mobo manufacturers to gain more experience, to avoid unstable products getting them a bad reputation. More likely, the problem was with the chipset itself.

If I were to guess, that's why there are 133MHz FSB references in the mobo manual.

From what I've seen on other socket-A boards, keep in mind that FSB jumpers and BIOS settings frequently don't do the same thing.
What I've seen with this era of boards is that if you set the FSB with the jumper then it sets all the PCI/AGP dividers correctly to match.
But if you leave the jumper alone and just set a faster FSB in the BIOS then, on many boards, that doesn't change the dividers.

In other words, typically the jumpers control the dividers, the BIOS does not.

So if the jumper is left on 100FSB and you rely on the BIOS to raise it to 133, the board will probably end up running at 133/44 because the PCI divider will be based on the jumper.
Not applicable to this board since it can't run 133FSB stable anyway, but it may come up if you get a different board later.
The way they mixed jumper+BIOS settings in this time period is unintuitive.

This is the last generation of motherboards that commonly had ISA slots, so it's a good one to hold on to. A few KT133A boards have ISA, but most of them don't.