conroe fsb wall hacking. please help

[s e t h]

ok i discovered, with some help from around here, that conroes can and do have fsb walls.

this thread is started in an effort to gather information on how to break these fsb walls.

i proved my week28 e6600 has a wall at 470fsb on a p5b-dlx by swopping out a week24 e6600 that does 500fsb with a 6x multi.
using exactly the same setting and being nowhere near the week28's core speed limit the week28 would not go past 470fsb no matter how hard we tried.

so our week24 does 500fsb and the week28 does only 470fsb even though the week28 is faster volt for volt than the week24.

i remember back in the socket479 heyday that some super-clockers succeded in pushing their cpu's further by modding the cpu vcca voltage.

does it seem reasonable to presume the same vcca mods will help alleviate conroe fsb walls?

so rather than hunting for golden cpu's by testing tons of them and selling all the bad ones could we investigate potential vcca mods on the p5b-dlx.
this mobo seems like the only one that is likely to actually hit an fsb wall with any kind of regularity at this time.

i will look for some relevant pics and stuff in the near future.
i just wanted to get this out there in the hope that more talent involved in this project will lead to more rapid success..
thanks for participating

[freecableguy]

keep reading

http://www.xtremesystems.org/forums/...d.php?t=114998

[s e t h]

may i ask what you are getting at?
chipset speed has nothing to do with this fsb wall as both cpu's are 9x multi running at 6x multi with all other setting exactly the same.
that is unless there is something different about how the p5b-dlx treats a week 24 or 28 e6600?

[Absolute_0]

I don't think testing done with teh 6x multi is quite as conclusive as testing done with a 9x multiplier. Can you get some good cooling together and try the 9x on both?

From teh AT review, i've seen CPUs top out at different FSBs depending on the multiplier used, not CPU-limited.

[s e t h]

the week 28 also maxed at 470fsb on a 7x multi...
the week24 could hit 500 on the 7x though :/

unfortunately we haven't done any further testing under dri-ice or something.
but yes with -70degrees both cpu's should hit 500fsb.
i'm willing to bet this week28 will fail at 470 though.

has anyone else hit what appears to be an fsb limit even when knowing the cpu is nowhere near maxed?

[s e t h]

stole this pic from the p5b-dlx mod thread :P
i presume this is the puppy that will have the vcca output and sense pins (i could be very wrong though)


and here's the datasheet: http://www.analog.com/UploadedFiles/...20datasheet%22 ...< page #7 has the pin descriptions

if anyone has any ideas where to look for vcca i'll try give the week28 a bit extra and see where that leads.
any other ideas?

[kiwi]

Here is what I can say

Tested a few cpus so far

CPU #1, e6400
*Gigabyte ds3
Wall, regardless of multi:
air 430
SS 460

*p5b
current WR holder rofl, like 8x570fsb
chipset DI cooled

CPU #2, e6400
* DS3 no wall till 520 on air (mb/chipset limit)
* p5b dlx, max 8x520 under SS
or 7x550, probably unmodded board limit

CPU #3, e6300
* ds3, no wall till 505, then probably mb/chipset limit
6x530, no wall

CPU #4, e6700 ES
Haven't tested yet but up to 490 was fine with 8x under SS

[[cTx]Philosophy]

Subscribed to folow how this works out..

My DS3 has a wall @ 462

Trying the newest bios to see if possibly any way to pass it, but so far no cookies

[hipro5]

UNFORTUNATELY CPU's fsb wall exists........NOT only on Conroe BUT on EVERY CPU out there......

UNFORTUNATELY No2: The VCCA you're talking about would be a very good idea BUT nowadays mobo manufactures follows "by letter" Intel specs, so they have sort circuit CPU's VCCA pin to CPU's Vfsb pins.....Vfsb according to Intel should be 1.2Volts and VCCA 1.5Volts.......BUT mobo manufacturers by sort circuiting the Vfsb and VCCA pins, they feed those pins (via bios) from 1.2 to 1.55V max)......We NEED the VCCA to be OVER 1.85Volts.......

NOW: By feeding the Vfsb over 1.7Volts, we encode stability problems....VCCA is STILL lower than 1.85V min we want.......

FINAL WORDS: FORGET IT.......

[SynGroW]

UNFORTUNATELY CPU's fsb wall exists........NOT only on Conroe BUT on EVERY CPU out there......

UNFORTUNATELY No2: The VCCA you're talking about would be a very good idea BUT nowadays mobo manufactures follows "by letter" Intel specs, so they have sort circuit CPU's VCCA pin to CPU's Vfsb pins.....Vfsb according to Intel should be 1.2Volts and VCCA 1.5Volts.......BUT mobo manufacturers by sort circuiting the Vfsb and VCCA pins, they feed those pins (via bios) from 1.2 to 1.55V max)......We NEED the VCCA to be OVER 1.85Volts.......

NOW: By feeding the Vfsb over 1.7Volts, we encode stability problems....VCCA is STILL lower than 1.85V min we want.......

FINAL WORDS: FORGET IT.......

No possibilities to cut some traces and use a Maximizer for VCCA supply?

[hipro5]

No possibilities to cut some traces and use a Maximizer for VCCA supply?

Nope......it's inside the LGA 775 socket(underneath the pins).......

I've already fed the Vfsb/VCCA of an Intel Bad Axe from a DDR Maximizer and the max it could "hear" was 1.75Volts......

[hipro5]

Hipro,

It only needs ~100mA. Why not cut the pin for Vcca inside the LGA socket and solder Vcca to Vcc on the CPU substrate? More noise on Vcca possibly, but i think it would work

Tom

I could solder a "teflon surrunded" tiny cable on the CPU's VCCA "dot" and pass it through the holes of the socket so as to feed it seperate with voltage......Hmmm......

[hipro5]

Thats tricky - but best of luck Do you think the dedicated PLL Vcc needs attention also?

Tom

No I don't think so........It's all about VCCA......

I've already suggest this (about seperate VCCA) to a "known" motherboard manufacturing company....If they listen to me, we'll see a new mobo out there with extra VCCA .......

[s e t h]

hell adjustable vcca would be the trick

[Revv23]

That would be terrific Hipro, i cant get my e6300 over 520 FSB even though at 510 it only needs 1.4v....

Hit the wall hard and fast.

[cirthix]

I could solder a "teflon surrunded" tiny cable on the CPU's VCCA "dot" and pass it through the holes of the socket so as to feed it seperate with voltage......Hmmm......

that's very doable. test it and get back to us! it might be simpler to find an appropriate voltage point on the motherboard than to use the maximizer. maybe tie it to vdimm?

[Holst]

vdimm might be a bit much, but something like vbt (if such a thing exists on DDR2, ive never used them)

I think the difficulty is getting the voltage to the right pin.

I dont think I would fancy trying to solder onto the CPU itself... although you can do it it would be cool. Can you take these sockets to remove a contact to bits without breaking things (by you I mean hipro :P )

[hipro5]

It will be a bit difficult but I don't have anything to loose - well exept one pin from the mobo's socket - .....

[Holst]

Haha, good luck trying it.

[cirthix]

It will be a bit difficult but I don't have anything to loose - well exept one pin from the mobo's socket - .....

you don't even have to break it, just coat it with nailpolish or some other insulting adhesive that can be dissolved if it doesnt work. no permanent damage, except for a little solder left on a pad. and that can be fixed pretty nicely with some solder wick action . dammit, you're tempting me

[Holst]

you don't even have to break it, just coat it with nailpolish or some other insulting adhesive that can be dissolved if it doesnt work. no permanent damage, except for a little solder left on a pad. and that can be fixed pretty nicely with some solder wick action . dammit, you're tempting me

I think the wire will have to be threaded through the socket, otherwise no matter how thin a wire you use it will mess up the CPU mounting. (If ive understood corectly)

[cirthix]

I think the wire will have to be threaded through the socket, otherwise no matter how thin a wire you use it will mess up the CPU mounting. (If ive understood corectly)

i suppose the plastic surrounding the socket could be notched. that's probably the easiest way. notching the plastic a little won't interfere with normal use of the motherboard later.

[Holst]

So you think a wire will fit under the CPU (or is the pin at the edge) Ive not used one of these chips so I dont knw how much clearance there is in there. It looks like very little to me.

[cirthix]

So you think a wire will fit under the CPU (or is the pin at the edge) Ive not used one of these chips so I dont knw how much clearance there is in there. It looks like very little to me.

ther's very little space, but you can get really, really thin wires. i think there's space enough for a thin wire. call me unextreme, but i'm not going to do it untill it's proven to work heh. i can't afford to fry anything.

[cirthix]

any updates hippro?