Nice range of results; good work. If the board had a working Turbo Switch, you'd have an extra parameter to further fine tune speed settings. BTW, does the Powerleap adapter utilise boards with Turbo Switches?

j^aws wrote:

Nice range of results; good work. If the board had a working Turbo Switch, you'd have an extra parameter to further fine tune speed settings. BTW, does the Powerleap adapter utilise boards with Turbo Switches?

I'll add that parameter on the next boards with working turbo switch.

Good question about the Powerleap. It certainly works with boards with turbo switch, but I've never tried to disable the turbo with the Powerleap inplace. In my opinion the board should behave the same regardless of the Powerleap adapter as it just increases waitstates and does not actually lower the clock on the CPU.

Okay, cool. See if you can get chkcpu speed results with Turbo enabled/ disabled. I'm not sure if different boards use this switch for waitstates and/or clock speeds. E.g. I've used chkcpu to display 15MHz on a socket 7 board.

j^aws wrote:

Okay, cool. See if you can get chkcpu speed results with Turbo enabled/ disabled. I'm not sure if different boards use this switch for waitstates and/or clock speeds. E.g. I've used chkcpu to display 15MHz on a socket 7 board.

I'll check with those boards that supports it 😀

Very thourough work!

Could you bench the Cyrix 6x86 at 40x2?

Would you be able to bench the Cyrix 6x86 at 33x4 if you manually set the appropriate CPU pins to HI/LO? Same question applies for the MMX Tillamock at 4x.

What was the non-working symptom for the AMD K6-3+?

Is the system stable with the K6-III at 480 MHz? Could you, for example, install WinXP or W2K without any issue? I find the Final Reality benchmark to serve as a pretty good stress test as well.

Which multiplier options does your Powerleap have? Does the multiplier setting feature of the Powerleap only work with AMD K6 II/III CPUs? What about K6, K5, Cyrix 6x86, Cyrix MII, Winchip, Winchip2, and Rise?

Great thread! But man, this mobo needs some love. A good wash with soap and flat paintbrush would do miracles.

Edit:

Do you add your results to Phil's database?

The one thing I love about ASUS is that they still have support for most hardware they ever manufactured. You can even register most of your old products at their website if the S/N label has not perished!

feipoa wrote:

Very thourough work!

Thanks 😀

feipoa wrote:

Could you bench the Cyrix 6x86 at 40x2?

Yes, that would be possible when I receive it (I have it on order as I mentioned in another post to you) 😀 I could also bench the newer versions (6x86 PR166 and 6x86L PR200 at the same clockspeed against the 6x86-80GP to see if there are any performance difference between the revisions)

feipoa wrote:

Would you be able to bench the Cyrix 6x86 at 33x4 if you manually set the appropriate CPU pins to HI/LO? Same question applies for the MMX Tillamock at 4x.

As I wrote in the post, even with the CPU's in the Powerleap I could not increase the multiplier.

feipoa wrote:

What was the non-working symptom for the AMD K6-3+?

Board would not post at all. Black screen.

feipoa wrote:

Is the system stable with the K6-III at 480 MHz? Could you, for example, install WinXP or W2K without any issue? I find the Final Reality benchmark to serve as a pretty good stress test as well.

This is a more time consuming test you're asking about 😜 If I find the time and opportunity I'll do it. The system was stable for several benches with Quake and Doom as long as I used a good cooler (CPU was overclocked from 400 to 480).

feipoa wrote:

Which multiplier options does your Powerleap have? Does the multiplier setting feature of the Powerleap only work with AMD K6 II/III CPUs? What about K6, K5, Cyrix 6x86, Cyrix MII, Winchip, Winchip2, and Rise?

My Powerleap PL-K6-III is just a newer version of the PL-ProMMX. It supports all the CPU*s you mentioned, but I don't own the Winchip's and Rise. Possible multipliers are from 2.0x to 6.0x and voltage from 2.2V to 3.5V. The Powerleap multiplier only worked on the K6 CPU's on this board. I ran a AMD K6 at 300mhz (4.5x multiplier) with no issues.

Have you tried manually writing to the CPU registers, either by direct wiring of hi and lo pins, or using software? I know, for example, that I am able to use software to adjust the multiplier of the Cyrix 5x86 between 1x, 2x, 3x, and 4x, although a certain order of operation for going between multipliers was required.

Good idea to cross-compare the original Cyrix 6x86 with later models at 2x40.

feipoa wrote:

Have you tried manually writing to the CPU registers, either by direct wiring of hi and lo pins, or using software? I know, for example, that I am able to use software to adjust the multiplier of the Cyrix 5x86 between 1x, 2x, 3x, and 4x, although a certain order of operation for going between multipliers was required.

Good idea to cross-compare the original Cyrix 6x86 with later models at 2x40.

I don't have the expertise or equipment for any soldering or small electrical wiring between CPU pins. So unless someone wants to do this work for me on some CPUs it is out of the question.

Software is a possibility, where can I find this for changing multiplier on the cyrix?

I've sent you a compilation of all my Cyrix 6x86-based programs. One of these programs may be able to change the multiplier. I have never had to use software to do this before, so I cannot be certain.

Cyrix 6x86 benchmarks as promised earlier in this thread available here:
Cyrix 6x86 80GP (2x40mhz) Benchmarking

I own one of these boards.

I did only play around with it for an hour or two before puting it in a box.
This thread will be a great reference for the next time I use the board.
I do own a Power Leap K6-3 adapter 😀

At first, good review, thanks.

vetz wrote:

I could not get the board to run with a higher multiplier for the Pentiums and other CPU's (except AMD K6) even when using the Powerleap. I tried all multipliers possible. 2x multiplier is the maximum for this board

Have you tried to measure "multiplier pins" to ground/Vpp? Are they connected to something at all?
Because some time ago I've modified another Asus MB (P/I-AP55TV): there were only two jumpers for the multiplier but I wanted to install K6-2 and I needed third one to get CPU clock up.
So I found pins in socket and soldered external switch to control "third jumper".
If I remember correctly K6-2-300 was running at 83 MHz * 3.5 = 290 MHz.

I just picked up one of these Asus PCI/I-P54TP4 motherboards, for the sole purpose of the original Cyrix 6x86-80. Took a long time to find. I hope it works.

Vetz, what PCB rev do you have? The official Asus manual doesn't note the 40 MHz jumper settings. Did you jot it down?
EDIT: Found it. Interestingly enough, the 40 MHz setting is silk screened on the MB, but is not mentioned in the manual at all. 40 MHz: JP20/21/22/23 = 1-2/2-3/2-3/1-2

Trhat was my first Pentium board with a P100 and 512kb cache. Don’t ask me why - that number sounded good to me.

Only when having sold this board to a friend I played around in detail and I discovered the 80mhz setting which run fine at 1.5 with the P100.

What’s most important for me:
I also discovered that turning off l2 gave quite a large performance boost. I simply didn’t and do not have an explanation for that. If I remember correctly it was also not a bios error since when turned off in bios also HW checks showed no L2 - IF I remember correctly.
I first thought it might be a sync issue between my EDO and that async cache. But apparently not. Again no idea what’s the reason. I also remember that at that time fake cache modules were “modern”.

Anyway I was not very happy with the board at all since the P55t2p4 with lot faster pb cache came out quite close after my purchase.

Small correction of the article: the successor boards p55tp4 did have later chipsets. They provided sockel7 by 430hx chipset instead of this sockel5 430fx chipset
Update2. My fault. Only the p55tp2p4 offers HX chip. So you are right

BR

Can you renew original pictures, please?

And one question. How the board compares to Intel Zappa?
https://theretroweb.com/motherboards/s/intel- … vanced-zp-zappa
Would you say, it's better or worse, than Intel Zappa? For I am considering to keep only one socket 5 board in collection, rest only socket 7 versions.

W.x. wrote on 2024-09-19, 06:57:

Can you renew original pictures, please? […]
Show full quote

Can you renew original pictures, please?

And one question. How the board compares to Intel Zappa?
https://theretroweb.com/motherboards/s/intel- … vanced-zp-zappa
Would you say, it's better or worse, than Intel Zappa? For I am considering to keep only one socket 5 board in collection, rest only socket 7 versions.

Fixed the images!

Both of them use the same chipset (Intel 430FX). Based on previous testing I'd say they are very similar to each, at least in terms of performance.

About using the 430xx series and asynch cache. Slower write cache speed. This post explains it:
https://dependency-injection.com/intel-430fx- … riton-l2-cache/

Great review but I think it slightly undersells this board (a lot of nostalgia for me with my P54TP4). It was almost too easy to get the board working with a P150/166/200 (this works on any FX board).
As the multiplier is on the CPU on all Pentiums, you just need to connect pin BF1 to it's adjoining ground (VSS) pin. I just used a very thin U shaped wire in the socket pinholes with the CPU removed. No soldering or modding the CPU itself, this info was all over the web and Usenet at the time. When I bought a non-mmx P200 for this board in Feb 1997 it took about 2 minutes to do. Using the jumper for clock multiplier on the board will then change from 2.5 or 3x multiplier instead of 1.5x or 2x.

About under clocking for old games, I flashed a Mr-Bios for this board around 1998 (there are around 8 versions available), and it has a neat software turbo mode. You hold down CTRL + ALT and + or - to turn turbo on or off when in Dos. Doing this with L1 / L2 off/on opens up more speed steps for older games.
This motherboard was in my main PC from Feb 1995 with a P75 then a P200 until I upgraded to a 440BX in '99, it even ran Win2k fine with 128MB ram. Always thought the P55tp version Asus brought out a few months later and the Intel overdrive chips were scams when you could connect two pins on the CPU with the same effect for free.
The Pipeline Burst Cache version of this board is as rare as hens teeth but I would like to see it benchmarked against the P55p4 (it should be exactly the same).

Very nice comment. I wish I would have known this hack back in the days. It was one of my most HW enthusiastic times. During pentium I switched from the 430fx to hx and even to Tx that time. It was mainly due to:
- switch to HX: PB cache and 83MHz Bus. I just learned about the 80MHz setting of the p54tp4 when installing my sold board at / for a friend again
- switch to TX: allowance for SDRAM mix mode on my special Tyan board even with secret 100MHz bus support. Which didn’t work reliable at all …

Ja these days … 😀