What revision number is your ASUS P5A? Some Rev. 1.05 and all 1.06 use the later revision (G stepping) of the ALi Aladdin V which on this particular board won't play nicely with K6-2+/III+ CPUs. Yours seems to play nice though, so I guess that's not the issue.
Are you sure you set the board to 100MHz FSB? There is a 95MHz setting and 95 x 5.5 = 522,5.
100MHz is at FS0: 1-2 / FS1: 1-2 / FS2: 1-2 / FS3: 2-3
95MHz is at FS0: 2-3 / FS1: 1-2 / FS2: 1-2 / FS3: 2-3
It's only one jumper difference so they're easy to confuse.

Thanks for replying Doornkat..
My mobo is a Rev 1.03
and attached is a photo of the jumper settings I currently have..
It may well be that I have misunderstood something when reading the manual..
I will re-read and find that FSB jumper.
rgds
VS

You're welcome! 😀
The photo you posted shows the four VID and two VIO jumpers (which are configured correctly for the K6-2+ 😀) and the BF0-BF2 jumpers (which are also correctly configured to the 5.5x setting 😀).
There are some Jumpers laballed FS0-FS2 beneath the AGP slot/next to the first PCI slot. Those are your FSB jumpers and you'll have to configure them for 100MHz FSB to get the 550MHz core speed.
I have attatched a picture showing the jumpers at 100MHz FSB setting.

Hey, that K6-2+ is legit after all - good catch 😀

Vipersan wrote:

Thanks for replying Doornkat..
My mobo is a Rev 1.03

That has the E-stepping chipset and works fine with K6Plus. Only downside is the motherboard cache can't cache over 64MB, but a) more than that is overkill for a K6-2+ system anyway and b) performance impact of motherboard (L3) cache is minimal anyway, it's the 128kB of L2 on the CPU that makes the difference and that can cache anything you throw at it.

I have the 1.04 and have been able to run stable at 133MHz FSB 😉

My grateful thanks Doornkaat...
First job this evening ...
I'm not going to rush it just now as I have a hospital appointment (MS clinic) in the next hour..
cheers buddy.
Nice one Dion ..
yes indeed ..looks like a winner.
🤣

I believe the G-stepping of the chipset works fine with K6plus as well, it's only one particular motherboard (ASUS P5A rev. 1.05/06) which is screwed up.

dionb wrote:

Hey, that K6-2+ is legit after all - good catch :) […]
Show full quote

Hey, that K6-2+ is legit after all - good catch 😀

Vipersan wrote:

Thanks for replying Doornkat..
My mobo is a Rev 1.03

That has the E-stepping chipset and works fine with K6Plus. Only downside is the motherboard cache can't cache over 64MB, but a) more than that is overkill for a K6-2+ system anyway and b) performance impact of motherboard (L3) cache is minimal anyway, it's the 128kB of L2 on the CPU that makes the difference and that can cache anything you throw at it.

I have the 1.04 and have been able to run stable at 133MHz FSB 😉

Afaik with CPUs lacking on-die L2 cache motherboards using the ALi Aladdin V chipset Rev E and D can cache up to 128MB RAM. Motherboards using the chipset revision G can cache up to 512MB RAM with 512KB onboard L2 cache and up to 4096MB RAM are cacheable with 1024KB onboard L2 cache.
With the K6-2+ and its on-die L2 cache RAM caching limitations of the chipset no longer apply and all 4GB RAM adressable by the CPU can be cached. 😎

ShovelKnight wrote:

I believe the G-stepping of the chipset works fine with K6plus as well, it's only one particular motherboard (ASUS P5A rev. 1.05/06) which is screwed up.

From a quick read on the subject it seems you're right on this. I'll change my above post. Thanks! 😀
Also there seems to be a fix for the P5A.

Vipersan wrote:

First job this evening ...
I'm not going to rush it just now as I have a hospital appointment (MS clinic) in the next hour..

Good luck for both!

ShovelKnight wrote:

I believe the G-stepping of the chipset works fine with K6plus as well, it's only one particular motherboard (ASUS P5A rev. 1.05/06) which is screwed up.

Indeed, it's the combination. Apparently with a rework you can fix it and get K6Plus to work on the 1.05/1.06 properly as well.

Doornkaat wrote:

Afaik with CPUs lacking on-die L2 cache motherboards using the ALi Aladdin V chipset Rev E and D can cache up to 128MB RAM. Motherboards using the chipset revision G can cache up to 512MB RAM with 512KB onboard L2 cache and up to 4096MB RAM are cacheable with 1024KB onboard L2 cache.
With the K6-2+ and its on-die L2 cache RAM caching limitations of the chipset no longer apply and all 4GB RAM adressable by the CPU can be cached. 😎

What the motherboard can cache and what the on-die cache on the CPU can handle are completely independent of each other. The L2 on the CPU can always cache up to 4GB of RAM, but even with a K6Plus the chipset limitations still apply to the motherboard cache. To do 128MB with the earlier Aladdin V revisions you need to set L2 in WT mode. In WB mode it can only do 64MB. The 512MB limit with 512kB cache on motherboard and G revision is much better, but still applies with a K6Plus - it just hardly matters as motherboard cache is almost irrelevant to performance.

dionb wrote:

What the motherboard can cache and what the on-die cache on the CPU can handle are completely independent of each other. The L2 on the CPU can always cache up to 4GB of RAM, but even with a K6Plus the chipset limitations still apply to the motherboard cache.

I completely agree, so the limitations no longer apply (towards system performance).

dionb wrote:

To do 128MB with the earlier Aladdin V revisions you need to set L2 in WT mode. In WB mode it can only do 64MB.

Yes, but performance difference is minimal (single digit percentages) so I'll say 128MB is the caching limit.

dionb wrote:

The 512MB limit with 512kB cache on motherboard and G revision is much better, but still applies with a K6Plus - it just hardly matters as motherboard cache is almost irrelevant to performance.

See, that's the same thing I think about WT/WB caching: Both are almost irrelevant to real world system performance. 😀

Tomato - potatoe; I guess we're picking hairs on each others' wording here while we're basically in perfect agreement. 😎

EDIT: I re-read your previous post and now I understand what you're trying to say. 😁
Ok, yeah, the only downside of the onboard cache only caching 64/128MB of RAM still applies but it's not actually a downside with the K6-III(+) and K6-2+ CPUs because it doesn't matter anymore.

But this makes me want to check out what NB stepping My GA-5AX is using. I'd like to know just how miniscule that difference is.

moved over just one jumper rightmost in Doornkaats photo..
New results attached ..so looks like I was indeed using a 95mhz fsb.
cheers guys
😀

Not bad. Thing to do now is to test both CPU and motherboard for clock limits.

In general K6Plus CPUs could happily manage over 600MHz reliably for long periods of time, and most P5A boards could do 120MHz FSB. I'd recommend doing the motherboard first. Lower the multiplier to 4x, then increase FSB step by step and try to run 3DMark. See how high you can get. As I said, my P5A 1.04 can max out the PLL at 133MHz. Once you know max FSB, try the CPU, working up step by step. I forget my exact scores, but 600MHz needed 2.1V, and 2.1V was enough up to ~620MHz. Above that things moved fast, 2.2V worked for 630MHz, but 2.5V was necessary for 636MHz and I didn't dare overvolt above that.

However there are K6Plus CPUs out there that can do 720MHz (6x 120MHz) at 2.1V, so YMMV and you may get a pleasant surprise 😀

Thanks for the advice and encouragement Dion..
I will indeed test it thoroughly but obviously I want to settle for practical limits for long term usage.
cheers buddy.