possible unlock of superlock procs

[bulk88]

closest yet to unlocking of superlock procs

This thread documents my attempt at changing the multiplier on a locked applebred by use the PowerNow feature of the Mobile Athlon. For latest please goto the end of this thread.

Summary

AMD disabled the ability to control multiplier on all the AMD k7 proceesors made after week 39. No wire mod, bridge mod, or BIOS mod can work now, and there is no way to unlock them really. But there is a way of changing the multiplier using a technology called PowerNow in Mobile Athlon Processors. Originally created to save battery life my decreasing the multiplier and voltageThis is done with a Windows Program, on the fly, with no data lose. It requires Chipset support and bridge mods.

More depth

Many people have gotten around the week 39 superlock by enabling powernow on locked desktop Athlon and Duron CPUs, using a powernow program, the multiplier can be changed on the fly.

To enable powernow on a locked or unlocked (it works on both) CPU, you must close 1 L5 bridge, and either 1 more L5 bridge or open/cut upto 5 more bridges (personal preferance).

Also to change the multiplier using powernow requires you to have a compatible chipset, and preferable BIOS. If there is chipset support but no BIOS there is a register change that can be done using WPCREDIT. Currently the only chipsets that DOESN'T work is the Nforce, Nforce 2 (hope), AMD 760MP/MPX (single only). THe AMD 750, and Kt133 and Kn133 are untested.

Original Post

Since the the multiplier traces are still conected into the die they have to go somewhere. Maybe they still go into the multiplier control unit, but AMD changed the amount of amps and/or volts to make a change by some internal die component change. In some other thread they did say their is more resistence going through the CPU, resistor in the line in the die? Has anyone tried using a sci lab regulated power supply to set the multipliers? If enough current is put through it could overcome the resistor and in. Also on the locked CPUs, why does AMD still set the bridges even though they are useless?

I am thinking of trying this but I have no clue what voltage or amps I should use, plus I would have to go buy one. This might kill the proc and remotely the mb (I'll be sure to use a $15 MB off of ebay and a applebred).

Also on the PowerNow/Mobile conversion attempt, did the guy's chipset have powernow on? You need chipset support to change the multiplier on the fly I assume http://www.cpuheat.wz.cz/html/AXP_multiplier/AXP_Multiplier.htm That article has alot information. I think the guy who did the mod did it wrong. Also the MAXIMUM multiplier ISN'T set by L3. I REPEAT the MAXIMUM multiplier ISN'T set by L3. So on a Barton 2500+ (11x) the CPU boots at 11x and that can never be changed because L3 can never be changed, but with powernow enabled you can goto the maxiumum set by L6 which is a differnt circut once your in windows. If it is done right and STILL doesn't work AMD must be fuse blasting or having 2 differnt cores or having a multiplier change enable wire coming from the core and if it is LIKE connected to VCC (+) change is enabled and if to VSS (-) change is disabled and all this time the PCB connected the wire to + and recently AMD changed it to -. I somehow doubt they did a die mod. They probably think they did a something differnt in the assemble and testing process.

EDIT, word changed from "It" to I, in red
EDIT, changed thread icon, added introduction to thread
EDIT, gave better introduction

 

[v8vega350]

i also have a stupid question but here it goes

what creates the multplier the die or the board

 

[The Coolest]

On the unlocked CPUs AMD used L3 bridges for Tbreds and Bartons to set the default multi. With the new locked chips there are two theories, one is stated in the first paragraph of bulk88's post, the second is that AMD has locked the multiplier internally, pretty much like Intel does with their chips.

I suspect this actually might work. I don't think AMD would lock the L6 bridges, if they didn't this trick should work as you are changing the multiplier Internally (inside the die) and not Externally (using '0's and '1's on the CPU pins)

*EDIT:* This is great for people with motherboards that either do not support multiplier adjustment in the BIOS or do not work with specific multipliers, like my ASUS mobo with the x13-x14 multipliers, I get a no boot with them, and as x14.5 gets interpreted as x21, my minimum multi on high range is x15. With this mod I could easily run my chip at low range, and have FSB as high as it will go and have my multi at any setting I want, without even bothering with no boots, or chipset incompatibilities with the chip itself.

 

[bulk88]

This still requires alot of bridge moding, connect the L5 set L6 to 24x (why anything lower this is the max, you can go lower than this but above startup multi ), set L8 to L11, do MB patch and check for on-the-fly/powernow support, dynamic voltage not needed. I am trying to find a MB in my collection that will allow powernow.

 

[The Coolest]

As I see it, L6s are just a reflection of L3s, then you should just cut/connect (which ever it is) all the L6s and you'll get x24 max multi. I think that most of the users here can connect a single bridge without much trouble (2nd from right L5), and on some mobos changing L8 *might* not even be necessary.

*EDIT:*

If I'm right, judging by this pic:

The L3s, state a x5 multi for boot and x12 multi as max multiplier. Seems to make sense to me. Just pop all the L6s, and get max multi of x24. About the L8s, they are just a mirror of the L11s, all you do is connect the cut bridges, for max vcore.



This pic from newegg proves my point, its a 2400+ mobile CPU, which has x13.5 multi and the L6s are set to L6.:|:|:, which represents the x13.5 exactly. Just mirrored L3.

 

[bulk88]

Thanks but I don't need that. I am using a Asus A7M266 AMD 760 chipset (non MP). I did the mobile mod but I am currently booting the system right now. The cache isn't responding either. Still posts as a duron. Be back with update soon. Also I could have killed the proc because I dug to a lower leval on on of the L6 bridges when I was clearing plastic.

 

[The Coolest]

bulk88 if you look at the table in your link in first post, the L9s are responsible for L2s working, just for the sake of it, how do the L9s on your duron look like? maybe if u mirror both L9 and L2 bridges the cache will turn back on? just a theory on that one tho.

I'm also wondering how could I check to be sure if this trick will work on my mobo, I really want to try this. if this works, this workaround for x13 multi might actually work for a change.

 

[bulk88]

L9s are all cut, and they only prevent the MB from messing with the cache but I can't think what MB would do that.

:
I took my apple 1.4 made it a mobile, it posted, but I forgot about a wire mod that was done and one of the wires got lose and jammed a bunch of pins together on the proc so the proc is dead now. But it did get to the Windows 2000 log on screen before it died. But I have its sister (was planning to use in SMP, same datecode and stepping, and voltage and everything), and now I need to mod that to mobile, ARGHH 2 hours. As much as I know the 760 can do that powernow thingie, looked at the datasheet, dunno if bios will allow me YET . Also I set L6 by scratching and 5v PS blasting it like on the old unplasticed PCB (I have 3 bartons like that, like 0305s or something). So I have confirmed it will get through POST, didn't remember what its tag was. I think it was unknown. Still my idea should work on paper.

 

[The Coolest]

Yep, on paper it should work great, now question is will it work off paper?
I'm closely following this thread, let me know what happens with the second chip. For testing purposes, don't use any wires this time, bummer to hear you killed a CPU tho, it's sad I know the feeling... At least you got a cool keychain.

One more thing could u point me to where u got the datasheet of ur board and where exactly you looked for the support? I got an ASUS board as well, so the datasheets should be similar in chapters

 

[bulk88]

Since I can't fins any documantation on PowerNow, I went into the K6-2+ datasheet http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/23446.pdf and found out how powernow works, I guess based on the xp to mobile site that it works the same. In the K6-2+ it says the CPU does a "Stop Grant" thingie around page 154. In the 760 datasheet http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/24088.pdf on page 39 it says that it can do a "Stop Grant". I belive Stop Grant and the FSB_Change to be the same thing.

 

[bulk88]

Just confirmed my test subject and possible future test subject superlocked. Also I am using DHD1400DLV1C MIXHB 0344TPMWs, anyone is stalking my posts the 1600 mhz applebred-B is another proc and I won't be mobilizing that, unless I kill my second 1400 Also I am using the pin mod. And the pin mod did work on a 0337 MIRGA (apple-A) that is unlocked in the same board, just to remove any possibilty of error. Now alot of scratching and filling and filling again. I am insulating the mobile bridge before I connect it. I know you don't have to do with the MP bridge but I am not taking any chances. I am doing it as if it were a L2 cache reenable.

 

[The Coolest]

Ok good luck, I hope u come back with some sweet results!!

 

[bulk88]

MINOR UPDATE

L5 is done, did it 3 times but the conductive paste spilled, so I reinvented using scotch tape to prevent it from going all over, I forgot to tape it. 2 of the L6s are scractch. One left, then its off to the blasting probes.. Also a 21 Gage (sp?) green blood colelction needle (Medicalease, its a Vacutainer) it much better than a micro flat screw driver. No, I am not a drug addict (if you think thats how I got one), my parents are doctors. Plus you can't inject using that kind of neddle.

 

[bulk88]

Done, now need to load into computer.

 

[bulk88]

OH MY SWEET ANGEL/SAINT OF OCING
I couldn't say God because this isn't ancient greece and I am christian so we have angels and saint instead of gods.

Anyways I went into WCPUID and in the PowerNow option degrayed, and it opened. I don't know what will happen yet. So I enabled Mobile mode sucessful, but I don't know if it works. YET. Wish me luck, say a prayer or whatever they call it in your religion.

 

[The Coolest]

Damn that rocks man, can you modify the multiplier using WCPUID?
What about getting that app, that the guy in the first post showed screenies of? You gotta update me SOON, coz I gotta go to bed soon as I got an early bus to catch tommorow, and I won't be home for a few days

 

[bulk88]

SOMETHING WENT WRONG WAY WAY WRONG

I check the maximum the CPU allows is 8x but I blasted L6 to 16x, I'll need to check my work. The L3 multi is superlocked at 10.5x BTW. WCPUID didn't want to change it, probably because the maxium allowed is less the the current (superlocked one). I'll try that CPUMSR program and maybe AMDs powernow program. OK I think I messed up the blasting of the L6 #4 because it didn't spark and I didn't scratch the holes enough making it not make connection or something and it is 8x which is okay according to the diagrams but I don't think powernow likes that L6 is lower than L3

 

[The Coolest]

dude give those apps a try.
even if u get it dooown to x8 thn itd be good, meaning u can change multies on locked CPUs, also why don't you just pop all the L6s so you'll be able to get the max x24 multi, although not practical, but always good to knows

 

[bulk88]

CPUMSR loaded and tried to change to 8x but the screeen froze when I hit the apply [set] button, I'll try correcting the bridges or playing with the prog. I think my MB is messed up and doesn't want to do the powernow change.

 

[Zerileous]

it would be sweet if a work around could be figured out. AMD might try a new locking method if this one becomes widespread but i dont thik the multi lock was just to kill overclocking, i think people must have been selling rebadged AMDs overclocked in a large quantity. This is all i can think of because, even if we [overclockers] make up a larger percent of AMDs market than we do Intels, we still make up a tiny percentage of the market.