• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

running Tualatin on CuMine MB w/o Powerleap

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
I first tried nailpolish, but it kept scraping off in the slotket, so I went with wire insulation. Instead of IDE cable I used a strand of phone cable (the 4 conductor kind).

I don't think jumping 4 to 26 will work on the slotket, since 4 is insulated from slotket by nailpolish/wire cover. It has to be done on the cpu.

My abit slotket wouldn't make it past safe mode at even default speeds, and even after a clean repart/format/install of win2k. Since I could run sandra and stability test at 133 in safe mode, I think Darth Via may be at work here.

For wire, go to Rad shack. They have 26G enamel insulated wire (called magnet coil wire). You scrape off the enamel at the ends to wrap it.
 
heres what i did for the AK4 to AK26 connection:


http://chosenfool1.homestead.com/files/ak4toak26a.jpg

its weird cuz its rock solid stable at 130MHz FSB, but any other sppeds higher than that, at any other voltages, it just craps out. The best, lowest possible voltage for this chip to be stable at is at 1.7V, and at a healthy 37C temp. I guess i hit the sweet spot for this specific chip.

1570mbm.jpg
 
Not sur eif it matters but all the traces on the chip are copper. You replaced that with the less conducting Aluminum. Not sure if it matters but it may.
 
I spent all morning trying to get my 1.0a to boot on my abit slotket/via mobo, using the same method that got me into safe mode before. I could not get past DMI verification on this attempt (and I rechecked my wire wraps and pin covers).

Worse, when I was putting my p3 back in the slotket, the ak4 pin came off. So now I have a p3 keychain and a mobo/slotket combo that won't play well with this mod.

I will dig up another slotket and give it one last shot. If that fails, I may either grab an ST6 and use this 1.0a or just go p4 1.6a/DDR. I suppose if I could pick up a cheap BX board that would work with the mod I might try that.
 
ol' man said:
Not sur eif it matters but all the traces on the chip are copper. You replaced that with the less conducting Aluminum. Not sure if it matters but it may.

it's the core that contains copper in place of aluminum...the traces in/on the organic substrate are probably made of the same stuff they were with the coppermines
 
My Update-
First, thanks Ol'man for the advice about not removing the slug - good reason. I started fiddling with doing it but mine looks well anchored down and didn't want to chance screwing up the traces. I used a flat file, yes a flat file!, to flatten all the edges on the slug. It was evident by the rubbing marks on the heatsink that the edges were high. I figure the effect of any scratches, or lower than flat around the edges are negligible compared to whatever amount of out of contact they were causing at the center.

I messed around with drilling, desoldering, etc, with a couple of generic slotkets - without volts jumpers - and decided there is no clean way to modify the three pins on the slotket, even though my first drilling on the Abit slotket 3 works great. After seeing that it works on my BE6-2 without these three pins in contact, I saw no reason not to remove the pins from the cpu. I don't plan on putting it back in my one Tualatin board since I can run it on BX. The two interior pins were more stubborn than I expected.

The generic slotket did not work - would not even boot.

I don't understand some of the problems being mentioned.
I started with BE6-2, slotket with a Cel2/1.0G running stable at 10x133, and win98 on this machine.
I replaced with the mod'd Abit slotket3 and with Cel2/1.0A
Boot was normal- the first bios volt was off a little,
but after setting it 1.85v, and FSB to 100, it worked fine and stable in SiSoft burn in.
Final settings were 1.75v and 131FSB
Then I tested it with an Asus S370DL-133 slotket, set for coppermine and default vots, and with the same retail heatsink and it works fine - stable in Sandra cpu benchmark burn-in for over 2 hours so far, at [email protected].
MBprobe and sisoft report the same volts as bios settings, ans sandra reports it as a celeron 10x133 and 256kb cache, bios reports it as a Pentium3@1332, so everything is really normal - no problems whatsoever.
These are my current Sisoft Standard v2001.5.8.11 benchmarks
CPU benchmark Dhrystone 3734 - Whetstone 1781
Memory bench marks - Integer ALU 496 - Float FPU 514
CPU Multi-media benchmarks Integer SSE 7238 - Floating Point SSE 8985

Does anyone see a reason to do any more mods than just removing the three pins? I don't know if there would be more to be gained.

I'm going to try an older version coppermine slotket, when I find it.
Thanks again to all - I'm just tickled pink with the results!
 
Last edited:
sammy:

youre right in saying that the generic slockets did not work - i cant make the two generic ones i have to boot at all, even with the 1.5V voltage short, the 3 pins covered, and the AK4 and the AK26 short.
i wonder what it is that we need to do to make the generic slockets work...if anyone has used ageneric slocket and made it work, can you verify what you did? im curious to know what i may have missed.

In summary, Its pretty straight forward with the slockets with voltage selectors(and this is the VERY FIRST POST of this thread too) :
1) connect AK4 to AK26 first on the CPU.
2) cover the 3 pins mentioned.
3) (optional) jumper the slockets with the appropriate voltage for the tualatin (if you dont have a voltage selector), to lower the default voltage.

im set, and im still psyched about the mod. Currently i already have offers to BUY the modded chip and adapter...i may yet sell it, but for now, i'll bask in the glory first. :)
 
this generic thing is really starting to bother me.
As i ONLY have genaric slotkets...
And i cant seem to get the 1st post
It has AUTO for voltgae (so CPU defualt controlls, and BIOS)
And 66/100/133/AUto for my FSB... So the BIOS controls it.

(have everything on AUTO so my BIOS will controll)

The slotket is very stable with coppermines, never had a problem.....
But if i have to buy a new one (damb)
what DO YOU have WORKING???
 
my first post in these forums... :)

Anyway, I do wonder where the information about VID4 came from initially. If you study the data sheets, true, VID4 needs to be pulled to get into the lower range of voltages. However, it has only ever been a slot1-pin. It has never existed in socket370.

In the PPGA data sheet, pin AK36 is specified as Vss. Same in the FCPGA spec.

so what's the point of connecting AK36 to anything at all?
 
martinus said:
my first post in these forums... :)

Anyway, I do wonder where the information about VID4 came from initially. If you study the data sheets, true, VID4 needs to be pulled to get into the lower range of voltages. However, it has only ever been a slot1-pin. It has never existed in socket370.

In the PPGA data sheet, pin AK36 is specified as Vss. Same in the FCPGA spec.

so what's the point of connecting AK36 to anything at all?

infact vid4 seems to be connected to vss yet (by the socket)
but i get more stable system sonnecting it also to vss pin ;)

this seems a paradox...:eek:
 
ok, i did my first few rounds of 3DMark2001SE.
at 1.70V it freezes.:(
at 1.75V, its stable and finishes thru the whole benchmark.:)

at any rate, 3dmark2k1SE is still videocard dependent - only when i upgraded the drivers did it have significant rise in the score. The CPU has little (or not as much) to do with the increase (as compared to a previous mark i had with a [email protected]). The benchmark at least stresses the CPU and the videocard, and it tests its stability. Next up, quake3 demo, and VulpineGL mark.:p
 
Good news guys!!
I fiddled with the bios, and got BOTH chips to work. I disabled the bios virus check and it booted! Does this mean I have a virus?

Well, anyway, I made it into regular windows mode with the 1.0a, so far as high as 135 fsb. I have tried higher, but windows locks up on startup, and I've tried Vcore as high as 1.75. I'm pretty sure it's either my slotket or mobo....I used to hit similar walls with my p3.

I've run a few sandra benchmarks. It seems only sandra reports cpu temp and fsb accurately. Both bios and cpuID get it wrong. Do I have to reinstall MBM??

Another strange thing, I got a BSOD when trying to boot at 66fsb (to do a burn in). I upped to 133 and it went right up.

I'll do a burn in at 1.75v, hopefully @66fsb (maybe safe mode), but I don't expect to get anything higher, at least not with this slotket and mobo. I may try to dig up an iwill slotket.

I'll take some screenies and find a place to host them. How do you take screenshots in cpuid and sandra?
 
Ruiner said:
Good news guys!!
I fiddled with the bios, and got BOTH chips to work. I disabled the bios virus check and it booted! Does this mean I have a virus?

Well, anyway, I made it into regular windows mode with the 1.0a, so far as high as 135 fsb. I have tried higher, but windows locks up on startup, and I've tried Vcore as high as 1.75. I'm pretty sure it's either my slotket or mobo....I used to hit similar walls with my p3.

I've run a few sandra benchmarks. It seems only sandra reports cpu temp and fsb accurately. Both bios and cpuID get it wrong. Do I have to reinstall MBM??

Another strange thing, I got a BSOD when trying to boot at 66fsb (to do a burn in). I upped to 133 and it went right up.

I'll do a burn in at 1.75v, hopefully @66fsb (maybe safe mode), but I don't expect to get anything higher, at least not with this slotket and mobo. I may try to dig up an iwill slotket.

I'll take some screenies and find a place to host them. How do you take screenshots in cpuid and sandra?

congrats!!!

you can try to sign up for homestead.com, for like 14 days they host for free. not sure after that. thats what i use for my pics. or if your ISP has webspace for you when you signed up for them, you can use that too.
 
well, I've done the mod today, twice actually. First I did my Asus slotkey. I opened the socket and took out AN3, AJ3, AK4. No connection between AK4 and AK26! Voltage selection is handled by jumpers on the slotkey (vid pins). Runs like a charm in my AOpen AX6BC Pro - whereas the powerlep adaptor seems to be incompatible. ;)

I got to 133 FSB at 1.5V, but my Matrox G400 starts crapping out at that speed, so 129 is the highest stable FSB at the moment. The cpu is obviosly capable of so much more!

Next I modded my trusty Asus CUBX system, which I know to do 150 FSB stable. Same method. Unfortunately I miscounted and took out a Vcc pin. Annoying, but there are enough left, so no worries. ;)

The celery 1000A did not want to run in the CUBX. I then proceeded to connect CMOS PWRGOOD with VTT PWRGOOD, using wrapping wire. I don't know, maybe it was too ghetto, or maybe it just doesn't want to play. No go.

Anyway, I'll have to get a new video card and try to get the AOpen board higher... :cool:
 
Back