ASUS XP55T2P4?

Ah, the legendary T2P4. Perhaps the best Socket 7 motherboard ever.

She's famous for doing 83mhz FSB, stable, if the PCI cards are up to the task.

I noticed there's a later revision 3.00 board with the same name but a very different layout.

The rev3.00 manual says it supports split voltages but only down to 2.5v. I don't know if that manual was meant to apply to the rev1.4 board.

The older version (I think it says 1.4?) appears to have a linear regulator for the CPU, while the rev3.00 is a switching type.

From appearance I'm not sure the rev1.4 board would support split voltages. It's picture shows a missing 2nd regulator that it would probably need for that. I think the regulator with the large heatsink would be the 3.3v supply. I'm just speculating from the appearance though.

I have both revisions. Rev 3.0 is the one you want. Shamino is correct in his post.

Rev 3.0 has:
- coin battery
- better CPU support
- more voltage options
- 512kb cache built in
- ATX connector (rev 1.4 sometimes comes with only AT connector)

The ATX version of XP55T2P4 does not have 83mhz FSB option which the AT version of the board does have. Other than that it is a kick ass board with great performance and compatibility. They are very stable and troublefree. There should be AMD K6-III+ support and BIOS hack for 128GB harddrives on the IDE if you go here and download: http://web.inter.nl.net/hcc/J.Steunebrink/k6plus.htm. You can set the multiplier to 2x on the K6's which translate to 6x = 400mhz.

about the voltage thing which i am concerned about:
both t2p4 and xp55t2p4's manual only documented following options:
single voltage (3.3 and 3.5)
3.2
2.9
2.8
2.7
2.5
but then its known that t2p4 has undocumented jumper combinations which provides vcc all the way down to 2.0, allowing it to run k6-2 without anyout trouble. i am not sure if the xp55t2p4's voltage regulator design is the same as t2p4 and so the jumpers would work the same way?

btw: do gigabyte, abit, msi, dfi, iwill and others have atx 430hx boards? i think aopen has one named ax53, but havn't looked into the manual yet.
then again, all of them seems pretty rare, compared to t2p4 and other at 430hx boards still seems plentiful.

noshutdown wrote:

i am not sure if the xp55t2p4's voltage regulator design is the same as t2p4 and so the jumpers would work the same way?

for 2.0V you had to set jumpers 2.5 & 2.7 & 2.8 & 2.9
for 2.1V you had to set jumpers 2.5 & 2.7 & 2.8
for 2.2V you had to set jumpers 2.5 & 2.7
for 2.3V you had to set jumpers 2.5 & 2.8
for 2.4V you had to set jumpers 2.7 & 2.8
for 2.5V you had to set jumpers 2.5
I can't find any mention that it won't be the same across both boards, and from what I remember the jumper config is the same on my AT and ATX versions of the board (yes I have the AT as well as both ATX rev's)
https://web.archive.org/web/20070210052552/ht … oup/P55T2P4.htm

noshutdown wrote:

btw: do gigabyte, abit, msi, dfi, iwill and others have atx 430hx boards? i think aopen has one named ax53, but havn't looked into the manual yet.
then again, all of them seems pretty rare, compared to t2p4 and other at 430hx boards still seems plentiful.

I can't think of any other 430HX ATX board at the top of my mind, only AT boards.

Well, there is the Gigabyte GA-586UX and Gigabyte GA-586DX. But you have to keep in mind that using ATX was still very uncommon at this time.

Yes, HX is a little too old for ATX. There are plenty of ATX TX boards though.

If you do not mind being limited to Intel cpu's you can usually find some nice Intel HX motherboards from OEM machines.

RacoonRider wrote:

Yes, HX is a little too old for ATX. There are plenty of ATX TX boards though.

TX has the 64mb cache limit, but the boards normally feature UDMA support as well as SDRAM.

TX is surely preferable above VX, but VX can already use SDRAM.

Does the 64MB cacheable limit also affect K6-III(+) and K6-2+?

havli wrote:

Does the 64MB cacheable limit also affect K6-III(+) and K6-2+?

Yes and No.

Yes for the 512KB motherboard cache, it wont cache more than 64MB regardless CPU. No for the CPUs own L2 cache, it can cache up to 4GB.

havli wrote:

Does the 64MB cacheable limit also affect K6-III(+) and K6-2+?

Not at all, install as much ram as you want 😀

I remember seeing somewhere a speedsys picture of K6-3+ where L3 (motherboard) cache worked actually slower than RAM 😁 Does too complicated caching scheme render the point of L3 obsolete?

Thank you.
One of my S7 430TX boards seems to be perfect for K6-III... onboard cache is defective, so not very useful with anything else. 😀

RacoonRider wrote:

I remember seeing somewhere a speedsys picture of K6-3+ where L3 (motherboard) cache worked actually slower than RAM 😁 Does too complicated caching scheme render the point of L3 obsolete?

It depends on the chipset and probably alot of other factors. With VIA MVP3 you want to use the L3 cache and keep within its cacheable range (128MB with 512KB, 256MB with 1MB and 512MB with 2MB cache).

I posted this image in another thread to show the memory and cache performace of VIA MVP3 with a K6-3+, the motherboard is a PC Chips M577.

havli wrote:

Does the 64MB cacheable limit also affect K6-III(+) and K6-2+?

Why would you use one of those with anything other then a super socket 7 motherboard? Even you your board's bios supports the K6-3, it will run like crap compared to how it would run on a MVP3 or ALi Aladdin V. You're also limited to 83MHz (on a good day) x 3.5 on i430 chipsets (as far as I know).

I can understand building an i430 based machine for a Pentium or Cyrix CPU, but using a K6-2+ or K6-III si a waste in my point of view.

The mobile chips allow software dial-a-clock, so it's really the most flexible solution in terms of DOS and early Windows retro gaming. In a case like this, running 66 mhz fsb is advantageous as it allows the machine to reach a little deeper down into the 386 performance envelope. So, i430 based chipsets can work just fine, depending what you want to achieve.