Flyview

On 6/2/2016 at 0:06 AM, Khenglish said:

On that laptop the vbios will be included in the system bios. You will need to decompress your bios so that you can replace the vbios module within it. Phoenixtool is a good program for that.

 

As for the bios mod, the 620m is a fermi core. There's a Fermi Bios Editor, but that refuses to work with mobile gpus. I'm not sure how you would drop the voltage once the vbios is extracted.

On 6/2/2016 at 2:15 AM, Flyview said:

 

Thanks. I'll work on extracting the vbios for now. I know I decompressed the entire bios before trying to unlock the CPU. So has no one has been able to edit a vbios for a Fermi mobile card? That's not good news!

Alright so I decompressed my BIOS with PhoenixTool from MyDigitalLife forum. Now what, how can I tell which of these files is the vBIOS? Thanks for the guidance!

5 hours ago, Klem said:

To allow some undervolting for CPU, just use the nice program Intel XTU or ThrottleStop_8.1.

ME Region designed for another things.

The voltage options are locked on this Ivy Bridge laptop. I was hoping to unlock them through modifying the BIOS?

Hey guys,

So about a year ago I tried this and even though I was able to extract the bios ME region and modify it, it didn't unlock anything on my UX32VD with Intel HM76 board. Is there anything I can modify in the bios image to allow some undervolting? The CPU just gets way too hot when also running the dedicated GPU on this laptop. Attached is the ME region I extracted a year ago.

ME Region.zip

2 hours ago, Khenglish said:

On that laptop the vbios will be included in the system bios. You will need to decompress your bios so that you can replace the vbios module within it. Phoenixtool is a good program for that.

 

As for the bios mod, the 620m is a fermi core. There's a Fermi Bios Editor, but that refuses to work with mobile gpus. I'm not sure how you would drop the voltage once the vbios is extracted.

Thanks. I'll work on extracting the vbios for now. I know I decompressed the entire bios before trying to unlock the CPU. So has no one has been able to edit a vbios for a Fermi mobile card? That's not good news!

On 2/10/2014 at 0:02 PM, Astelith said:

The i7 620M it't not that bad, I can play almost everything on ultra (2048 x 1152 on Dell 23") with 40fps or more... at 1000 core 1v / 2900 mem the power draw is around 110/120W ...and for that I love the 620M !

But now I've found a good 920xm so let's see

- - - Updated - - -

I recover the card soldering a new vbios chip.

Hey I know I'm digging this up from 2 years ago, but I'm looking for a modded nVidia 620m vbios. I'm hoping to undervolt it as this card just gets too damn hot in my Asus UX32VD. Did you ever have a custom vbios for it?

The only "trick" I've managed so far is to disable nVidia PowerMizer with PowerMizer Manager app, set it in to low performance mode, and then up the clocks in that mode back to stock clocks. This allow me to run at 0.925v instead of 0.975v but it's STILL really hot. It'll reach 85c+ under load at 24c ambient. This is in combination with an i7 3517U with eliminated TDP for turbo power boost with Intel XTU. The problem is they share a heat pipe and heat eac hother up.

Would anyone be so kind as to build me a modded vbios or point me to some documentation on doing this myself? I can post the original. Thanks!

If you are interested, I think I know of a way to get BCLK control on HM76, PM or reply if interested.

To anyone who is having trouble getting this to work, it might be be required to change ICC Boot Profile Selected By Soft Strap from true to false; this is located under PCH Straps->PCH Straps 10. I don't know why this works, but when I was helping someone with HM77 chipset and the instructions from OP didn't work, I tried that change and voila, it worked.

So should I try this on my ASUS UX32VD with HM76 chipset? Could it cause something bad?

Yes. BCLK will also increase RAM/PCI/etc clocks as well; that being said, unless you can change ram multipliers, you probably wont be able to increase BCLK by a whole lot without being able to maintain stability. You'd most likely get more gain by changing turbo multis, with a lot less effort and risk. If you make a mistake changing a certain nvar you can just reset its value by changing or removing the part (corresponding to the nvar) for one boot.

I will take a look at your zip later as i do t have access to a pc atm

Thanks! Yea then I'm not even interested in changing BCLK, I know in the past when I've tried increasing FSB on other computers with PCI-E SATA harddrives, it would lock up very very fast. If I could unlock the multipliers, that'd be great. Were you able to do that?

Just try what i said but allow me to save you the headache and dont try too hard (unless youre confident), because you will end up with a brick like i did. IMO perf gain in actual world applications just. By increasing BCLK without ability to increase volts isnt worth the effort. And since bottlenecking isnt an issue for gpu/cpu relationship, you're better off trying to buff your gpu

Interesting, I opened up the ME region only .bin I posted earlier and you are correct, the changed bits aren't there (I guess the "Int" folder is no good). However, they are if I open up the full image, which I flashed. I also re-dumped the current FW and opened the image in fitc and the bits are changed. Here is my newly dumped .bin after the changes: http://s000.tinyupload.com/?file_id=61252391357967251527

I'm really just doing this to see if it's possible, not so much out of need, this is an Ultrabook after all. I would love to unlock the multipliers. The BCLK would also overclock other components right?

Make sure you set "number of flash devices" to zero so thay only the ME portion is created, do all flashing and dumping in DOS. If it still doesnt work you can try extracting an IFR structure.txt file and then looking for any vars that you might want to change. Probably wont get BCLK control, but you might be able to change turbo multis

Well I built and flashed the entire FW (8mB). I figured it wasn't necessary to change the number of flash devices if I did that? I did indeed do the flashing in Windows.

Is the modified ME Region an ME dump after flashing or an exact copy of one created in FITC? Because neither image has the necessary bits set.

Seriously!? The modified one I pulled from the "Int" (Intermediate) build folder. It creates these when building the full .bin image which I flashed.

If the changes that you actually made was in fact:

Clock Source Select: from 0x00011A33 to 0x00011A34

SRC Source Select: from 0x00033733 to 0x00133744

PLL Reference Clock Select: from 0x00088CBF to 0x000A8CBE

Divider Enable: from 0x000005EB to 0x000005FF

SSC Control: from 0x00010000 to 0x00000000

then you are probably SOL.

I am pretty sure that enabling BCLK OCing is either downright impossible on HM76 chipset or it requires much deeper modification (which I attempted, failed, and gave up on). Good luck on whatever you choose to do. If by chance you are worried about any possible bottlenecking then rest assured that a 3610QM did not bottleneck my GTX 970 eGPU setup.

Yea I guess so. Here's my original and modified ME regions (in .zip). I just realized the modified one is much larger. I took these from the files FITC first created for the original, and from the Int folder for the new one.

ME Region - modified.zip

ME Region.zip

Those look identical to the changes I had to make when I first stumbled upon this thread. You are good to go with those values. Just make sure you set number of flash devices to 0 at the top so that when you build your image, it will be simplified, and equal sized output image (this is how you "build just this portion").

@Tech Inferno Fan or @Khenglish

I got my flash descriptor unlocked, and I dumped my ME image in DOS using FPT, but when I was trying to copy the settings needed from the ocme.bin in the OP..... I can't tell which bits correspond to each other between ME10 and ME8. Could one of you please point me in the right direction or if you have time, modify it for me? Here is my ME dump. Thanks!

So I flashed my modified firmware (just built and flashed the full thing), nothing additional unlocked in XTU or ThrottleStop that I can tell. In fact, I could have sworn that in XTU, I previously could see "Reference clock - 100Mhz" greyed out but now I can't see it at all.

I tried attaching my original and modified ME regions here but it says invalid file. Tried both .bin and .rar.

Must be something else then. If you were clockblocked, you wouldn't even be able to set the clock. The core slider would be completely grayed out.

You are correct. I tried drivers 347.88 just to check and the same thing happens. I can choose the clocks to apply, but they don't actually apply. What else should I check? This is on a Asus UX32VD running Windows 8.1 Pro.

What is your actual core clock under load via sensor reading?

It always shows 625MHz in Afterburner when memory has also ramped up to 900MHz under load, no matter what I set it to. Memory changes on the other hand do take effect.

Sent from my Nexus 5

I was the one who made that thread. There is no clockblock on 600M cards. On 350.12 my 650M SLI overclocks same as it ever has. Maybe you should DDU and reinstall driver.

I just did that. Completely wiped with DDU, re-installed drivers 350.12, core clock selection still doesn't have an effect. Tried both with MSI Afterburner and Nvidia Inspector. I.e. I can set the clock, but when checking the actual clocks in Nvidia control panel "system information", the GPU clock is at 625. The memory clock overclocks fine and is reported as such.

I initially realized there's a problem when I noticed 3DMark11 was reporting my GPU core clock as stock, and the memory clock as overclocked.

What are you talking about? Overclocking has never been blocked on 600M, only on some 800M and 900M.

This: Nvidia clockblock: vBIOS (unblocked in 347.88, reblocked in 350.12) | NotebookReview

And it seems to be true, core clock changes don't take effect.

I've searched and searched but can't find an unlocked vBios for the NVIDIA 620m, which came in my Asus UX32VD. Is anyone willing to unlock this for me? I can post the original vbios if needed. I mostly need this because on the latest drivers, 350.12, the core clock overclocking has been blocked (again). Thanks!

blowntaha said:

Brick happened because of something stupid I did. I was trying to see if I could un-greyout some of the settings under Advanced->Extended ICC in my BIOS; all in effort to do what I couldn't do with ME FW. Basically I ungreyed some bits that were supposedly undefined, followed by a bad flash. Could have been recoverable but thunderstorm caused a temporary blackout and I forgot to put battery in. GG WP cyal8r mobo

Wouldn't hurt to try. Just do exactly was the OP says and change only things under ICC subsection along with the other one or two settings outside of that. Pretty sure chances of bricking are slim to none, assuming you only change what is stated in the post. Don't take my word for it though, I am only saying that based on what I have experienced with one motherboard, along with reading a lot of different threads on a lot of different forums (techinferno, etc).

Wow that's pretty unlucky! Well I'll give it a shot and see if it really is our chipset that's the problem. The OP hasn't said exactly which lines to change. I opened his modded ME in fitc and the lines I changed (posted above) seem to be the only ones he changed, but I'm not sure. Could you verify?

I have (or had, bricked but not beyond recovery) a y580 which is also HM76. I have spent hours and hours trying to get it to work and later learned that it is impossible to enable any sort of extended ICC control on chipsets (that Intel has labeled) which do not support overclocking. There is a chart in one of the PDFs from the IMEI system tools .zip file. I had a 3920XM in my HM76 board (it's an unlocked cpu !!) and had no more OCing capabilities than I did with my 3610QM, which is a locked cpu. Both before, and after flashing a modified ME FW, while also checking that the ICC registers were the correct values via cct.exe (from IMEI tools, and they were correct), ressulted in the same limited control in XTU/throttlestop/setFSB/etc.

I even went as far as to contact Intel and was told that it is a hardware limitation, despite the fact that firmware is supposed to control hardware, but I can't fully understand that concept so I might be wrong. You can look at my previous posts to view my progress and things I tried.

Just my two cents.

Damn, yea I noticed that you were having problems. Would it hurt for me to try? Are the changes I have made sufficient? Were you trying something else when you bricked it?

Yeah if you can get the firmware dump and are are to flash it will work.

Alright this is where I'm at with my Asus UX32VD, please help me proceed.

I've made the FW dump with fptw64.exe.

I opened it up in FITC. It's detected as "Intel 7 Series Chipset / Intel HM76 - Mobile". Someone on the previous page was having issues with a HM76 chipset?

I changed the following bits as I think this is all you changed, and this is what I saw recommended in the Intel PDF regarding overclocking in the ME section of the firmware.

ICC Data\ICC Profile 0\FCIM/BTM Specific ICC Registers:

Clock Source Select: from 0x00011A33 to 0x00011A34

SRC Source Select: from 0x00033733 to 0x00133744

PLL Reference Clock Select: from 0x00088CBF to 0x000A8CBE

Divider Enable: from 0x000005EB to 0x000005FF

SSC Control: from 0x00010000 to 0x00000000

ICC Data\ICC Profile 0\ICC Registers:

PI12BiasParms: from 0x08880888 to 0x00000888

Can you please verify this is correct? I really don't wanna brick this laptop. Also, how do I build just this portion?

I'm thinking: Build, build settings, check "build compact image". Any other settings I need to change?

setPLL was written originally because I wanted to overclock a 2530P's FSB to run faster CPU+MEM and more importantly, overclock the pci-e bus for faster eGPU performance. It uses look up table files (.LUT) to program your PLL, an idea that I got from perusing a setFSB clone for Linux.

I also wanted it to be *free* so other 2530P users could do the same. It was important too that users could add their own PLL to be supported. Anybody who has a programmable (not TME-locked) PLL, has the PLL datasheet and can decipher it would be able to add support for their machine. setPLL is also considerably faster at programming the PLL than setFSB is.

To answer your question, it will depend on how programmable you PLL is to answer whether the FSB and pci-e clocks can be programmed separately. The example LUT files included with setPLL have both the FSB and PCI-E clocks for several PLLs where both can have both programmed.

I found that pci-e overclocks over 12% usually result in the wifi card failing.

Hi, thanks for this program! I am trying to overclock my old Inspiron 9300 just for fun. I have found my PLL on the motherboard, it's ICS954201. Using SetFSB, the PCI slider is disabled for this PLL. This results in the overclock quickly becoming unstable after only 1-2MHZ FSB increase. It looks like the HDD locks up. If I create my own LUT file, do you think I'll be able to increase FSB and specify the stock PCI-E clock?

Just for kicks, I'd like to try this on an old old Dell Inspiron 9300. It has a Pentium-M 1.86Ghz 533MHz FSB. I tried installing XTU but it says hardware not supported. What would I use to actually change the CPU BCLK once I unlock the bios? I don't want to change the FSB only as that quickly becomes unstable due to other components.

Yeah if you can get the firmware dump and are are to flash it will work.

Sweet! I'll have to do this very very carefully. If anyone has done this on a Asus Zenbook (UX31 or UX32) please chime in!

Hi guys. I came across this thread referenced from ThrottleStop thread. I carefully read the first post, but before I even begin, would this work at all for an ultra low voltage CPU? I have a 3517U in my UX32VD. So far I've been able to increase the TDP limits with ThrottleStop which allows full boost indefinitely. That's nice, but a small overclock would be nice Cheers.