Help with Pad/Vid mod for 775 Core 2

[Joe Sixpack]

Seems like these 775 systems just won't go away will they?

So my latest pickup was a Lenovo a58e Thinkcentre, I won't go too much into deal other than it comes with a 180w PSU which frankly I thought might just catch fire when I turned it on.. 180w? and a E430 celeron, The board uses a g31 chipset if im not mistaken.

I've done tape mods to 775's in the past (primarily 1066->1333) with great success (long live the q6600)

But that celery has to go, I just so happen to have recently picked up 2 dozen e7500's for $1/ea Thinking it would be a decent chip overclocked or not but why not try anyway?

I tape moded 3 of them.

1st one wouldn't make it past post without crash/freeze/reboot
2nd one was windows stable but reboot after 5mins of linx/linpack.. sky rocketing temps might not have helped.
3rd one was survived 15mins of linx with short 32mb problem sizes, needs more testing but likely stable, or at least stable enough.

I could have stopped there mission accomplished but I want to see if we can get that first one stable with some voltage increase.

the bios on this system does not give vid in bios, and obviously no control either.

so I studied some vid pad mods.. Honestly could not find a lot of detailed info on them.. looked at the e7xxx/e8xxx data sheet.

cpuz report says the vid max is 1.288 which I did not see on the chart as a valid vid.
I then downloaded the VRD11.0 design pdf and it shows more options, VID0 gives 0.00625v resolution (half of the normal 0.0125v stepping)

even so I can not find 1.288 as a valid vid value, the closest I could come up with would be 1.28750

I tried checking the vid pads against vcc/vss and also vss/vcc sense with a multi meter, my reading were not conclusive.
I was hoping to find a low resistance to vss to at least identify the "0"s.. but iirc only vid0 gave such a reading.

After reading the pdf's some more it sounds a lot like the processor actually transmits the vid rather than it being hardwired.


So here's my questions

I need to properly identify my VID so I know which pads to target.
how? CPUZ says 1.288 which is not valid

Once I do figure out what pads to mod I assume tape will set them to 1, but setting them low I need to ground out to vss.. but
do I need to ground the pad to vss.. or just the socket?
In other words.. do I connect VID0-7 to vss or do I insulate the pad first, so only the socket is connecting the 2 pads?

I've read vid mod guides saying both ways not sure which is the correct.. or if it even matters.

If I can't discover the correct vid to start with i may just do all of them to eliminate any doubt.

I've attached a text file with the vids sorted from the VRD 11.0 design pdf

 

[Joe Sixpack]

Ok well.. no one replied so I just went ahead with it.. I mean it's not like theres much $$$ in a system like this you can pick'em up off the curbside if you're lucky.

Ended up getting what I thought was a legit VID from Real Temp 3.0.. seems like they took out vid display after that version. and it agreed with my guesstimate of 1.2875v

So the way I did it was to just insulate any pad I was flipping to 0.. I used electrical tape to insulate the pad and then some HVAC foil tape (reflectix home depot $3.5 for a life time supply )

Alight so at stock voltage I could barely past POST.

First stop v1.32500 (HWmon reports 1.28vcore), random reboots unless 1 core disabled, but I could load up mini xp but unstable.. so we're getting some where.

So increased to VID 1.35 (1.312vcore HWmon), will boot mini xp but crash under load testing

I did'nt really wanna push it much past 1.35... but damnit we gotta try if only briefly.. I've seen some spec this cpu at 1.36 max vcore, intels data sheet say it's like 1.45.. not sure what to believe.

Increase VID to 1.4, Interestingly HWmon still only reports 1.312 vcore... it runs around 1.1v when idle (speed step) but on load it reports 1.312
I suspect it's unable to report a number higher than that.. whats the true vcore?
At this point I can enable both cores and get to visa (it's what was on it when I got it, and we're testing windows proper now)
Even at 1.4v the system is not stable.. it seems to be ok with 1 thread but 2 threads of linx it fails fairly quickly.

at this point.. im a bit stuck.. I don't really see any other way to proceed.

OEM board with no control options.. I've got ddr2 800 in it so ti's on 5:6 ratio
I do have some 533 I could stick in it, that would drop us down to a 1:1.. that might help... I suppose I could swap out for a better psu... but I actually don't think that's the problem here, if HWmon is to be beleived the regulation from psu is pretty tight (of course im sure it's not able to read the fine ripples)

I could yolo and go for 1.45v but im not sure how comfortable I am running a cpu 24/7 with maximum voltage even if it is within spec (or not)
On the other hand Im skeptical that it would even help.. we're not getting much return for our voltage bumps here, not a lot of road left.
some more juice to the NB might help but no options for that in the bios of course.

If the board had multiplier control I'd drop it down a couple and be perfectly happy with a 3.4 or 3.5 clock.


so any idea's on where to go from here other than throw this CPU in the "lottery loser" box and grab another one to test?