Asus TXP4-X Socket 7

[trodas]

Prepare for trip to the past :) Asus TXP4-X is really a Socket 7 mainboard and for these who did not remember - yes, this is a Pentium 1. The date is 1996. Intel 430 TX chipset.

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X1.jpg


So, a FSB frequency can be set from 50, 55, 60 a 66Mhz. By jumpers. Multipliers and Vcore as well. Vcore can go up to 3.5V, however this is a normal Vcore in that age for many CPU's. Pentium MMX 233 has a Vcore 2.8V anyway. Maximal documented speed is 233Mhz - 66 x 3,5 :D And on board are bulging I.Q. caps:

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X2.jpg


Near CPU socket I noticed side to side with bulging I.Q. caps also (s)Hitano caps. So, the mainboard was already attempted to be repaired to extend it's lifetime. Hitano caps are, however, also pretty bad caps. On most pictures TXP4-X on the net (or pictures opf the TX97-XE that looks very similar, just added audio and stuff) are showing only two bottom caps there. Obviously Asus cheated on customers already in 1996...

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X3.jpg


(s)Hitana are in the places where I would expect highest load, so there probably the I.Q. caps blow up/leaked too much, so replacements are necessary. Sadly the recap was not performed as well, as it should. Some soldering places ale slightly damaged. Probably the repair guy used too weak soldering iron and little or no resin, eh. Or had little experience in boards soldering. Never less the board survived.

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X4.jpg


Interesting is the use of small tantalus Kemet caps. At the time it surely way attempt to use quality caps. However what I know today's (thanks to Davmax) about noise in tantalus, irreversible changes in them and their very poor ESR/ripple parameters - well, I did not think anymore that this was a good idea.

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X5.jpg


Media buss 2 - what in the hell are supposed to be this? Anyone got a idea?

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X6.jpg


The mainboard got a USB and USB voltage filtering is done by 100uF caps. Only. As time will go, there will be used 470uF and later even on MSI boards 1000uF caps ;) And ceramics place caps? Wow. And what is this, Mexico?! Maybe there are rather that caps - (F1) - fuses against shortcut...

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X7.jpg


And this next "PCI set" chip will be in time called South Bridge :D Well, many years are going to pass... :)

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X8.jpg


Now a Pentium 233Mhz with MMX rule there! :) 66 x 3,5!

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X9.jpg


So, to rule well, it get a nice good quality caps - Samxon GC! :D

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X10.jpg


I added cap even where it was not before - this place is connected with the in background pictures tantalus cap and this get connected to the North Bridge chip, so we better add it. What if someone figure out, how to increase the FSB over 66Mhz?

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X11.jpg


Instead of questionable 100uF 16V caps I added a 120uF Panny FM caps :D

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X12.jpg


In socket there someone left a place for two pieces of bigger tantalus SMD caps. That is not going to be left like this. After scratching off parts of the PCB off the protective paint and some twiddling with tin I managed to add two pieces of 10uF 16V ceramic caps and literally pull the tin to them for better Vcore filtering right under the CPU. Overclock!

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X13.jpg


Overal look at the upper part of mainboard showing it's relatively modern design. ATX PSU, usable Vcore design and also dimms. That is not that bad, it is? Dimms can be from 8 to 128MB per bank (all 256MB total, wow!) and Simms take 4 to 64MB modules also.

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X14.jpg


Look at ISA slots are, on the other hand, gloomy. Their control IO (or this is keyboard IO?) is so big, it looks almost like Motorola 68000 CPU as big it is! Between PCI slots I added missing two pieces of filtering caps. They filter 5V, so a 1000uF GC Samxons for 6.3V are again ideal ones to use.

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X15.jpg


From the Vcore design someone stole one mosfet - probably it shall run parallel with existing one and maybe we can hope they at least used higher specs mosfet, when now there it is only one. Why I did not exactly like the writings "BUZ102SL" on it anyway? :)

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X16.jpg


On the other hand the Vcore regulator looks very stable and serious. Notice how thick wire and how many turns the first, only input voltage filtering coil, has. Beautiful, is not it?

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X17.jpg


Okay, enough talk. For recap of this mobo is need:

Asus TXP4-X
-----------
14x 1000uF Samxon GC 6.3V d8
2x 120uF Panny FM 16V d6.3 ( P12922-ND )
2x 10uF SMD 1210 keramic ( PCC216CT-ND ) - better get something bigger and for lower voltage, if possible - 1210 is roo small, 1808/1812 is much better choice, even 2211 size is applicable and well!

Some settings and so on you can find in manual there:
http://www.nodevice.com/driver/TXP4-X/get31584.html


How I run it with 233Mhz Pentium MMX you can see here:

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X18.jpg


I tried even the "WinChip" 200Mhz CPU and it run well too:

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X19.jpg


Done and further testing and possibly overclocking (I hope?) show the mobo owner - mech13 ;)
Because in Anandtech they say the mobo can run at FSB 75 and 83Mhz:
http://www.anandtech.com/showdoc.aspx?i=70&p=2

50Mhz
FS0: 2-3
FS1: 2-3
FS2: 2-3

55Mhz
FS0: 2-3
FS1: 2-3
FS2: 1-2

60Mhz
FS0: 1-2
FS1: 2-3
FS2: 2-3

66Mhz
FS0: 2-3
FS1: 1-2
FS2: 2-3

75Mhz
FS0: 1-2
FS1: 2-3
FS2: 1-2

83Mhz
FS0: 1-2
FS1: 1-2
FS2: 2-3

[Error 404]

That IS a blast from the past! I've seen some old boards, but never one with equal amounts of ISA and PCI slots. Good luck overclocking it: try reducing the CAS latency on the RAM :D, if you can.
Good work soldering on those extra/replacememt caps; some of them looked near-bursting.:shadedshu

[Jizzler]

Just like the board from my first build. Had a P200 MMX at 250Mhz, w00t!

With lapping, better cooling options, and your cap job maybe you can get a lot more out of that 233 :)

[Silverel]

The things I missed out on... subscribed anyways. Looks really interesting :)

[xmountainxlionx]

this is older than the stuff i work on at work! lol

[Error 404]

I've got a crappy Celeron 400 system, i might post some pics on here. FSB can be put up to 83 MHz, which means the 400 MHz goes up to 500 MHz! (Unstable, though).

[trodas]

Error 404 - latency reduced to 2-2-2-5 and ram is still holding up well ;) :D The board is surprisingly fast in booting Windows 2000 - and I overclocked to 75 x 3,5 - amazing 262Mhz, lol :D http://valid.x86-secret.com/show_oc.php?id=347811


Jizzler - I was hoping for the 83Mhz FSB to get it working, but when I switched it, something very different happend :laugh:


xmountainxlionx - ehm? Where do you work? :twitch:


Error 404 - todays it is unstable probably because of caps... nevermind. By little accident with the 83Mhz I managed something entierly different! :D


First take a look at official CPU-Z world records there: http://valid.x86-secret.com/records.php

As you can well see down there, the lowest submited valid CPU-Z scores are 15.02Mhz, 15.4Mhz and 15.8Mhz respectively.
Now when I set the "83Mhz FSB" setting, the machine took awfully long to post and even post screen took like minute... Post screen was showing 50Mhz clock, but memtest show something different:

http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X20.jpg

So I was thinking - well, it is not a WR in lowest clock, but what the hell. After two days figuring and badly cursing how to get some input device to work (no PS/2 stuff working at such low clock!) I submit that score just to see, where I end up with it. So I did: 25.08Mhz http://valid.x86-secret.com/show_oc.php?id=347812

But then I asked myself - how the 25.08Mhz originate? The multiplier is set to 3.5x - what if I set it (the Pentium MMX 233 - P55 - is unlocked) to multi 2x...?

So I did:
http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X21.jpg
14.32Mhz http://valid.x86-secret.com/show_oc.php?id=347812

And it was now all only too clear. The resulting FSB is 7.14Mhz and hence the board support only 2x, 2.5x, 3x and 3.5x multipliers (at least officially) for P233 MMX - what about P90 with multi hardcoded to - 1.5x...?

So here we go:
http://i4.photobucket.com/albums/y134/trodas/AsusTXP4-X22.jpg
10.74Mhz http://valid.x86-secret.com/show_oc.php?id=348684

This had to be done on Windows 98 se, since Windows 2000 refused to boot with the P90 CPU, once I set it to 10.74Mhz - on the boot screen never appear the progress indicator.


Guys, it is all just a big pile of random accidents, to be perfectly honest with ya. It was an accitent that after passing thru so many people, his current owner wanted me to recap the board. It was a accident someone flashed the board with TX97-XE bios - that is why on the CPU-Z screens we see TX97-XE instead of TXP4-X what the board really is. It was a accident that I decided to install the missing caps by design and hence probably made the 7.14Mhz even working. It was a accident that I stumbled upon the Anandtech review with the overclocking settings 75 and 83Mhz. And again it was a accident that I decided to try them. It was a lucky accident that I got such nice unlocked Pentium 233 MMX to play with. And the discovery how to save CPU-Z dump when almost nothing work at the superlow clock was another accident. VERY lucky one, mind you. And on the top of that, it was again accident that I even discover that the board is working with the FSB 7.14Mhz - because just to make the board post at this clock, it took like 1 min... that means, before you see the graphic card bios...! Most people would declare the board dead after like 30 sec of no post, right? Will you wait after like 40 sec? After 50 sec? ;)

So, that is the story of new lowest CPU-Z WR score. Just a big pile of random accidents, lol.

Now how to get a multiplier 1x CPU that will work in the mobo? Pentium 60 or 66 sounds a good way to go, however this suxxka need 5V (!!!) and max. the mobo support is 3.5V... ( http://en.wikipedia.org/wiki/Pentium )
What are the chances that Pentium 60 or 66 will post at 3.5V - even at 50Mhz or 7.14...?
Will the board even post with it - or some bios problem prevent it?

Another idea is based on fact, that there is 9 CPU multiplier settings (3 switches). Only 4 of them are documented. Maybe - if I'm superlucky - maybe I can get a multi 1x from the mobo somehow. If the unlocked Pentium MMX 233 are even physically able to use 1x multi. Anyone know that? :o

[mullered07]

wow validate that shit and set a wr :rockout:

and how in hell do you get 15.8mhz with a pentium d, thats impossible Link (http://valid.x86-secret.com/show_oc.php?id=194410)

[xmountainxlionx]

i work at a IT company. we have about 10 systems that have pII's and III's

[Error 404]

Haha, good work! Does CPU-Z run on windows 95? If so, I think I can break that record. :p
My friend has an old 80286 lying around, so I reckon I could bring it down to about 8 MHz or so. :D
Would 95 run on 640 kb of RAM, though??

[strick94u]

This thread robbed me of about 5 minutes of my life:laugh:

[trodas]

mullered07 - already validated, new WR - 10.74Mhz - is set - just the lazy hall-of-fame code is not updated yet, lol :laugh: :rolleyes:


xmountainxlionx - interesting. I thought more about computer museum anyway :o How come you run such old machines?


Error 404 - yes, it does. The 10.74 was made even on Win98se anyway. Win2k refuse to boot, lol.
I think I can break that record. :p
Good luck :D
old 80286 lying around, so I reckon I could bring it down to about 8 MHz or so
That is the catch, man. To get a CPU-Z score the CPU-Z must run on the system. And that require a CPU that support cpuid instruction. That means some latest 486 DX4 and possibly maybe even some latest 486 DX2... but don't count on that. I tried:

Cyrix 486 DX2 66Mhz:
http://i4.photobucket.com/albums/y134/trodas/Cyrix_DX2_66.gif

As you can see, no go.

Would 95 run on 640 kb of RAM, though?
Impossible. At least 8MB, IIRC.


strick94u - be happy we did not rob you more, man! :D

[Error 404]

Hmm, i'm pretty sure i can easily lay hands on a 486 processor, and Underclocking it could be done by changing the clock crystal and BIOS settings. :D
And i didn't think 98 would run on 640 k or RAM...

[trodas]

Yea, but - what good it can be, when you can't have a CPU-Z score with it? ;)

Besides, the chipset sync problems make you impossible to reach anywhere near as low, as I did ;) And beware about nonfunctional input devices, lol :))) Just like I say - try it. It is unbelievably hard to do.