It means you should be able to plug it into your UM8810PAIO socket 3 CPU ZIF and run it at 5 V without the proprietary ECS voltage regulator module.

mmmm I see. the plan is to run a cyrix 5x86 120 so if the amd is soldered onto the voltage regulator board Im unable to use it. I will keep looking but thats a good find. thanks 😀

checking the um8810 manual it mentions the cyrix 5x86 cpu quite a lot however the best AMD chip that gets a mention is the amd 486dx2/4.

Indeed, the Cyrix/IBM 5x86 chips have a lot more fun features to play with compared to the AMD counterparts. This eBay item is only for that particular AMD chip. You'd need to find a CPU on a socket 3 - to - socket 3 regulator board, like in the Gainbery photo I posted. Oddly, that voltage regulator was set to 3.45 V and not 3.6 V for an IBM 5x86 for use at 120 MHz. It probably only needs a simple resistor swap to make it a variable voltage regulator module though. Same concept as in,
Modifying your motherboard's voltage regulator for overclocking

http://www.ebay.com/itm/AMD-Am5x86-P75-AMD-X5 … =item5d3765dcf8

I think this is what you want

haha I really appreciate you guys keeping your eyes peeled.

we are getting closer 😀

"Hi,
the processor stuck in a socket and is not soldered.

The voltageboard has jumpers to change the voltage to:
3.45V or 3.30V

The board is namend PGA-DX4/2
Regards Rolf
"

No 3.6.......

Im suprised no one has the official UM8810 regulator or a PCB to regulate the voltage that plugs into the CPU for 3.6 volts. They are not common?

I'm pretty sure all you have to do is change a SMD resistor on that interposer board to alter the relatively useless 3.3 V setting to make it output at 3.6 or 3.7 V.

de-solder the resistor and replace it with a different one? ha ha 😀

are you near Brisbane feipoa? 😀

That is pretty much the only way to do it. I haven't seen any interposer boards which natively had a 3.6 or 3.7 V option.

It is very simple really. Each SMD resistor near the voltage regulator sets a different output voltage. Remove the one which corresponds to the 3.3 V setting, you can trace the jumper block to find it. Once you desolder the SMD resistor of interest, solder two small wires to it and use alligator clips to attach these wire leads to a variable resistor (trimmer). The resistance of the variable resistor chosen should be in the ballpark of the resistance of the resistor removed. If the resistor removed is 2 K, pick a 5 K trimmer (0 - 5Kohm range). Then hook up a volt meter to the bottom of the interposer to the VCC and GND pins to read the voltage at the CPU. Once the system is powered up, turn the knob on the 5K trimmer to see what voltage you get on the multi-meter. Once your desired voltage, say, 3.7 V, is discovered, remove the trimmer and measure the resistance with the multi-meter. Then source an SMD resistor of this resistance to solder onto the interposer board. Alternately, if you have space, you could solder the trimmer right onto the interposer and have a variable voltage socket 3 interposer board. In fact, someone should have designed this 15 years ago. I think it would have been a hot product. If designed today, I think I'd die before getting rich off this.

Even the gainberry adapters with the Cyrix chips don't do 3.6V (from what I've seen)

I measured my Gainberry some time ago and found it to be 3.45 V, even for the units which are advertised as Cyrix 5x86-120 chips. I immediately removed the cpus from the adapters after discovering this.

I don't think there are any adapters that give the 3.6V needed for the 120. Hell, 3.6V isn't even all that common on 486 motherboards that supposedly support the 5x86. Perhaps there was a reason why the Cyrix officially supported motherboards list was so short.

I got such a board recently without VRM. I plugged 128 MB EDO and a 486DX2ODPR66 on it. Until now it works without problems.
The L2 Cache covers 32 MB.

3dbench 1.0 52.6
3dbench 1.1c 50.7

Some Vidspeed benches with different cards show that the limiting factor is the PCI bus transfer speed. So reasonable fast PCi graphics cards show exactly the same speed.

After much testing I have found the UM8810-PAIO rev2.1 board to be very stable but it does have quirks.

The board is definitely not as refined as say a Biostar MB-8433

4/5 sound cards seem to have issues with the board and it took me some time to find the right card.

The board is reasonably quick and seems very stable.

I have not checked the slow down feature this board has.

Cyrix 5x86 120 CPU
64MB of Ram
Matrox Millenium

3dbench2: 86.1

FP_FAST and BTB_EN = on

[img]hhttp://imagizer.imageshack.us/v2/1024x768q90/20/lgq4.jpg[/img]

rgart wrote:

After much testing I have found the UM8810-PAIO rev2.1 board... Cyrix 5x86 120 CPU 64MB of Ram Matrox Millenium […]
Show full quote

After much testing I have found the UM8810-PAIO rev2.1 board...
Cyrix 5x86 120 CPU
64MB of Ram
Matrox Millenium

3dbench2: 86.1

FP_FAST and BTB_EN = on

You are really cutting your benchmarks short if you don't set LSSER = 0.

Yeah sorry. The batch files I use for my cyrix PCs are set as follows:

5x86 /btb_en=on /rstk_en=on fp_fast=on /loop_en=off /lsser=off

I find windows does not crash with the options set this way but I have not bench marked in windows yet. I have been experimenting a little with cyrix enhancements.

dos works flawlessly with the above enhancements.

Im still waiting for a few pieces of hardware to arrive.

When you set /lsser=off, is the Lsser bit at 0 or 1? We want LSSER at 0. I recommend using the evergreen bit enabling utility.

I will test the evergreen bit utility now.

When I set /lsser=off I get a better bench result than /lsser=on.

So wouldn't that indicate effectively /lsser off = /lsser 0

Probably, but best to look at ET586.exe /s to be 100% sure. LSSER = 1 by motherboard default, which is not optimal.

After months I have finally found myself a very affordable VR-100/VR-102 with an ECS UM8810 Rev 2.1.

Its on its way to me now.

What a pain in the butt though to design a motherboard with an external plug-in voltage regulator 😜

Did they all have a Phoenix BIOS?
I'd say this board wouldn't bee too bad if it had an AWARD BIOS.