VOGONS


First post, by FullYes

User metadata
Rank Member
Rank
Member

Greetings

I’ve been doing some o/c experiments with an MS-5158 based 430TX board

https://theretroweb.com/motherboards/s/msi-ms-5158-tx5

It has the clockgen I mentioned in the heading. I found a datasheet for this model on TRW:

https://theretroweb.com/chips/6011

My actual chip

The attachment IMG_8595.jpeg is no longer available

Like a lot of other TX boards. It will run at 83MHz without much trouble. I am a bit hobbled by my CPU (a k6-3 400AHX, 2.4V) the CPU doesn’t seem to want to go over 458MHz (83x5.5)

I have discovered, moreorless by accident, that this clockgen will output 100MHz, and the system boots at 100x4.5 and I can run benchmarks in DOS, but, a 50MHz PCI bus is creating problems for the IDE controller(!)

I’m running a voodoo3 2000 PCI

My benchmark table

The attachment IMG_8615.jpeg is no longer available

My hope is that someone on here can suggest a way to enable a 5/2 divider. The clockgen appears to support it and I believe the setting at 83MHz in the dataset is the one that’s outputting the 100MHz setting on my board, but instead of running the PCI at 40 MHz, as the datasheet implies, it’s running it at 50MHz

I assume the datasheet is the wrong version for my clockgen. There doesn’t seem to be a lot of info about this chip online. I found only 1 other board with the same chip. An MS-6117 slot 1. No mention on that of any FSB support above 75Mhz

Anyone got any ideas?

Thanks in advance folks!

Reply 1 of 8, by rmay635703

User metadata
Rank Oldbie
Rank
Oldbie

If you could disable the ide controller and use one that was used with both pci and pci/66 you could then possibly live with 50mhz pci which boosts performance if you have parts that run that fast.

Worth noting sometimes it isn’t the controller itself causing all the issues.

Using a uide 66 hard drive with an 80 conductor cable made one of my systems with unstable ide on an overclocked bus work fine.

430tx from what I remember has problems with async pci bus clocks, so even if your clockgen creates a different divider I don’t think it solves the problem.

Reply 2 of 8, by FullYes

User metadata
Rank Member
Rank
Member

Thanks for the quick reply. I’m tempted to try that route. I have been looking and Found a few Promise Ultra 100 (and 133) TX2 cards which seem to support 66MHz PCI fairly cheap. Perhaps someone on here will have a recommendation for such a device?

I’m pretty sure it’s the controller but I’ll try an 80 conductor cable. Good suggestion. I’ll need one anyway if I do get one of these add-in cards

Reply 3 of 8, by bakemono

User metadata
Rank Oldbie
Rank
Oldbie

On my Shuttle HOT-433, the onboard IDE would stop working at 60MHz PCI, BUT it would work with XT-IDE BIOS installed. So installing an ISA card with an XT-IDE BIOS EPROM, or inserting it as an option ROM into the main BIOS, might be worth a try.

GBAJAM 2024 submission on itch: https://90soft90.itch.io/wreckage

Reply 4 of 8, by FullYes

User metadata
Rank Member
Rank
Member

That must have taken some detective work!

I would still need a suitable card to add XT-IDE functionality. I don’t think I have a suitable network card or bios chip I could use but thanks for the suggestion! Was the XT on the ISA bus?

Reply 5 of 8, by rmay635703

User metadata
Rank Oldbie
Rank
Oldbie
FullYes wrote on 2026-01-17, 16:49:

That must have taken some detective work!

I would still need a suitable card to add XT-IDE functionality. I don’t think I have a suitable network card or bios chip I could use but thanks for the suggestion! Was the XT on the ISA bus?

Get an isa Ethernet for XTIDE

If that doesn’t work moving vio up from 3.3 volt to 3.6ish may help if your just on the edge.

Reply 6 of 8, by FullYes

User metadata
Rank Member
Rank
Member

Thanks. I'm not sure I can set VIO independently of the CPU voltage. There are only two options in the manual for VIO, 3.3V and 3.52V and 3.52V only seems to be available when setting up the voltage for a single voltage CPU. I could do with perhaps getting a Pentium P54C CPU for this board so I can test undocumented voltage settings

Reply 7 of 8, by rmay635703

User metadata
Rank Oldbie
Rank
Oldbie
FullYes wrote on 2026-01-19, 13:27:

Thanks. I'm not sure I can set VIO independently of the CPU voltage. There are only two options in the manual for VIO, 3.3V and 3.52V and 3.52V only seems to be available when setting up the voltage for a single voltage CPU. I could do with perhaps getting a Pentium P54C CPU for this board so I can test undocumented voltage settings

What’s strange is that at least 1 cpu required a 3.52vio but was a split core chip.

Back in the day I had one board that allowed that but it always was perplexing how most folks could actually use said chip.

The Infamous PCCHIPS m590 used a 3.8 volt vio so they could overclock the chipset.

My guess is if you found no other option you could manually break out the vio so the regulator for that always sends a higher voltage, no guarantees that it’s even enough to get stable pc100 on the built in ide though

Reply 8 of 8, by FullYes

User metadata
Rank Member
Rank
Member

Looks like it’s about the same cost for an isa network card and one of those promise IDE controllers. Buying the IDE controller seems more straightforward, as I don’t have a chip programmer or a chip I can use. The only issue with the IDE controller option, is they apparently don’t support CD-ROM detection in DOS.

I’ll lose some Upper memory to bios shadowing, but I guess I will have that with XTIDE as well

My other issue is I don’t yet have a chip that can take full advantage of 100MHz FSB. Having tried to push a bit further - my k6-3 won’t do more than 450MHz at 2.5V. (There is nothing on my board between 2.5V and 2.8v according to the manual) I haven’t yet tried my K6-2 but I doubt I will get it to do more than 500. I need a k6-2+ or a 3+

So i may have to revisit this when ive found a suitable CPU. The prices are quite steep right now here in the UK for the + CPUs

Thanks for your help so far all those that replied!