The asus p5a has something about it that makes it faster than other mobos with the chipset.

That note is why I recommend 1.04 😉

Also 1.05/6 has buggy voltage from my testing, 1.04 is just the best revision imho. I had 1.06 and 1.04 and hated the 1.06.

A user in the k6 3 aiming for the stars thread made a custom bios for the gigabyte equivalent board and managed to improve its performance by using some internal timing settings from the p5a bios. So the p5a indeed has some special sauce of some sort in it.

Most k6-3+ records in that thread are using it from my recall.

It would be quite problematic hunt for specefic MB version this days...
I looked at my HW list and I should have Asus P5A-VM rev. 1.05 with Ali M1541 + M1543C chipset, 512kB L2 somewhere... So as it is 1.05 it would also has issues, If I will dig it some day I could give it a chance...

K6-2/3+ Success on Asus P5A 1.06 - simple mod

Thanks for info, just for curiosity I measured the level on WB/WT# pin on my Gigabyte and it's already high (3.3V) so nothing to mod there... As I read the thread it seems to affect only K6+ CPUs? Someone with a non+ K6 reported that mod hadn't any effect.

Apparently only +. Correct.

RayeR have you made any revisions to your pcb design or do you plan any? Did you ever decide to incorporate any of the tillamook mods or are they already present.

Which pins do you recommend?

Thinking about ordering one

I updated PCB layout for known cache and VCC2DET mods (on my current PCB done by wires) but I don't plan to send it to JLC as I don't have any other CPU. Also the thermal via array below the core should be made a bit smaller to prevent solder leaking throug vias that are not lying directly under the core and some more blocking caps should be added on top side.

RayeR wrote on 2024-02-03, 00:19:

I updated PCB layout for known cache and VCC2DET mods (on my current PCB done by wires) but I don't plan to send it to JLC as I don't have any other CPU. Also the thermal via array below the core should be made a bit smaller to prevent solder leaking throug vias that are not lying directly under the core and some more blocking caps should be added on top side.

Okay. If you plan on doing this I will wait. If not, I may play with it if you uploaded source.

RayeR wrote on 2024-02-03, 00:19:

thermal via array below the core should be made a bit smaller to prevent solder leaking throug vias

JLC supports via in pad, they even have option of copper epoxy filled capped vias

rasz_pl wrote on 2024-02-03, 08:30:

RayeR wrote on 2024-02-03, 00:19:

thermal via array below the core should be made a bit smaller to prevent solder leaking throug vias

JLC supports via in pad, they even have option of copper epoxy filled capped vias

Copper epoxy? As in some sort of thermally conductive epoxy kinda deal?

More, smaller vias it seems like might be another solution as said. Let surface tension/capillary action whatever help it from oozing out. But epoxy would certainly stop the flow.

rasz_pl wrote on 2024-02-03, 08:30:

JLC supports via in pad, they even have option of copper epoxy filled capped vias

I know, I already used them for a prototype at work where I need via in pads of a fine-pitch BGA. But in that time I designed the interposer they probably didn't have it yet...

How’s it been working? Have you been using it?

I digged another modern-looking ATX S7 board Asus P5A-VM rev. 1.05 with latest beta BIOS with AMD K6-2 500.
https://theretroweb.com/motherboards/s/asus-p5a-vm
I tried my Tillamook in it jumpered at FSB 66MHz x 4, Vcore 2,1V, VIO 3,5 (minimum possible) but it even didn't posted (blank POST display). Strange, with other CPU and same jumper settings it booted.
But never mind, the real performance of this MB (despite nice benchmarks) is even more crappy than Gigabyte GA-5AA that I tested with K6-2 too and P5A-VM @500MHz is very close to GA-5AA @400MHz in Quake. Probably due to similar Ali Aladin chipset...
Table updated: http://rayer.g6.cz/hardware/tillamok.htm#CPU_TEST_TABLE

Try the tillamook mods

What else mods? I already have ADS# - ADSC#, BRDY# - BRDYC#, VCC2DET# - GND...
The PC Chips M590 didn't POST without VCC2DET# - GND but it's already done.

RayeR wrote on 2024-09-15, 00:35:

What else mods? I already have ADS# - ADSC#, BRDY# - BRDYC#, VCC2DET# - GND...
The PC Chips M590 didn't POST without VCC2DET# - GND but it's already done.

Sometimes removing mods is the answer.

But it sounds like you know of all of them.

You can also try other multiplier settings and verifying core voltage with DVM

Yes I tried measure Vcore and 2,1V was there, CPU temp seems to be like in normal operation. I tried also different multi...

RayeR wrote on 2024-09-15, 01:04:

Yes I tried measure Vcore and 2,1V was there, CPU temp seems to be like in normal operation. I tried also different multi...

Maybe:
DIY Bios Modding guide Jan Steunebrink k6-2+/3+ 128gb

This doesn't worth, similar crappy chipset...

But I finally found a good S7 MB. It's quite cryptic signature: Elpina BAE-VO but I guess it looks like a PC-Chips. Chipset has big letters PC100 AGPPro but it's identified as VIA Apollo VP3 82C598AT, better than ALi 😀 It has AT/ATX power, AGP, 3*SDRAM, 1MB L2... It detected Tillamook correctly, allows to set Vcore for 2.1V and VIO hardwired to 3.5V. It runs fine at 83*4MHz and provide much higher memory bandwidth to AGP or PCI VGA card. Due to this I got significant speed up in Quake at 640x480 to 21.7FPS. Next possible FSB step is just 100MHz without side steps. The CPU POST at leas with 2.7V but it's not stable even at 2.9V. It boots in DOS but hangs/crash in a secocd when I try to type a command there. On other MBs I hadn't luck at 100MHz too so it seems to be the limit of my piece. Maybe with peltier cooling it could do it but i don't want stress it with high voltage much more... I think it's good result compared to K6-2 500 I got before...

The description points to PCChips M577, which is about the only board I know from them with the MVP3 chipset.