tmash

Hello,

It seems that new Pascal gpus in mxm form factor no longer support Surround?

Any insight on why it has been disabled, or how to enable it?

980 desktop mxm supports mxm according to notebookcheck.

Thanks!

You should avoid Optimus as it adds noticeable latency. Use kyran 30awg wire or PATA/SCSI (preferred) solder the wires, then tape the entire wire (shield) with copper tape, and connect ground to the copper tape. Don't do twists, single flat 90 degrees is ok, helix like twists are bad. Keep no-dry flux over the soldered pins/contact with wire. Then cover the copper tape with kapton tape to avoid shorts. I was able to do pcie 3.0 gen x8 with my mxm laptop, should be easier with mpcie @gen2.0

Sent from my LG-D855 using Tapatalk

@Derlep Yeah liquid metal fixed my temp jumping, though not max temp as significant. Example: IC diamond stayed consistently at 90 for 50 seconds before it started to increase which is okay I guess. Took long to rise to 90 with gallium on the other hand with -5c difference max temp at 26c room temp. Thats 96c at [email protected], and 86c at 20c room temp which is significant due to airflow (active cooling effectiveness is not linear with room temp)

v0 and v1 only gives wrong voltage reading, just square the value to get the real one. Power delivery differs between brands and laptops, e.g my p150sm/p170sm supports i7 4940mx which is 65watts, my 4700mq running at 1.20v pulls 78watts at 4.2ghz (stock 48w) which is close. I don't think your model is designed to handle the extreme or high tdp CPUs so you might be putting too much stress on mobo since it's way far from the designed tdp, would explain why Dynamic voltage mode is unstable (my speculation tho).

You can also try cooling the VRMs if they're not being cooled, you can post your laptop model.

@Derlep Try MC v1, it works fine on a y510p which had the same issue

3 minutes ago, Derlep said:

 

It seems to be AMI Aptio 4. The "problem" / potential problem I see is that it has 5(x2) patches for different processors. Ubutool will happily patch it and leave only the Haswell (06C3) ones. I think it is probably fine, but who knows.

yes its totally fine, there is no risk at all playing with microcodes, the important part is bios size and padding insertion, UBUtool does that properly (checked aswell), you can go ahead and do it with confidence.

(Note: I don't have access to my PC for the next month)

1 hour ago, MiSJAH said:

Can anyone make this tutorial even more simple for a layman like me? I'm due an upgrade on my GT72 (4720HQ) soon so happy to take a punt with this. 

 

Cheers.

If your bios is AMI Aptio 4 or 5 ,use Ubutool: https://cloud.mail.ru/public/9SSs/YJbsWyC2V, extract, put the bios in the extract folder, run UBU.bat then change microcode to v1 or v3. I avoided this as I'm not aware of how it inserts padding. I'll update this reply soon once I have access to my laptop.

5 hours ago, Derlep said:

 

The microcode downgrade will have to wait as going to 0 is a more significant change and I'm not brave enough to flash that yet

 

I'm using TS 8.10b2; exe md5 95ABE57810F9C0DCFC549E41D83F2F9A

 

I did the MSR thing but I don't know if this is correct as there are no changes before/after, not sure what I should be looking for other than the PATCH ID which stays at 6.

 

I'm using legacy boot and booting from cold. Also attempted sleep-resume and hibernation-resume.

I don't remember what I did in 6, has to do with MSR199 198 and 150, I'll flash 6 tonight and check again 

16 hours ago, Derlep said:

Hello,

 

I am attempting this method on an 4710HQ. I downgraded the microcode to 06, but the repeatedly increasing max multiplier thing does not seem to be working. I can get x37 on all cores, but the max stays at 37.

 

Tested microcodes:

17: max 37-36-35-35

07: max 37-37-37-37

06: max 37-37-37-37

(stock cfg is 35-34-33-33 with +2 bins)

 

Is there any trick apart from re-opening the FIVR window after increasing the values? Is it possible that other components/BIOS are locking it somehow? Any ideas about ways to check what the issue might be?

 

Thank you

First, remove the microcode, or do 00h, then open R/w everything select CPU MSR register, do and post a screenshot of it before throttlestop and after throttlestop bin increase (in your case 37x all cores).

What throttlestop version are you using?

I had similar issue locking at 38x unless I shutdown completely. Also, make sure you aren't using Windows uefi mode, standard uefi is ok as it works here, bios works aswell. Disable fast boot when shutting down.

hey thanks to your quick guide, it gave me an idea where to start with memory OC/OV on my ancient 770m lol

I was able to go from ~2000 mhz memory to ~3200 with samsung memory @ 1.51v (stock 1.35v), used 4B pencil as i noticed its less grainy and thus dust and airflow doesnt wipe it out easily i guess, though still havent tested it >1mo.

I also had to raise vbios tdp limit since the core started to throttle after memory overvoltage, and now its perfectly stable.

http://i.imgur.com/3iRJwed.jpg

http://i.imgur.com/eA3MHcc.jpg

http://i.imgur.com/V7T8MoL.jpg

offtopic, did you actually remove the backside memory from your 980m? " MSI 980m 4GB (no pointless backside memory wasting power) " thats awesome for OCing...

Would be nice to see temps with CPU OC, I need more fins for increased "virtual" cooling capacity 

1 hour ago, aarpcard said:

This mod/hack looks fantastic.  Definitely want to try it - I'm doubtful I have the available cooling though to support over 4ghz.  3.9ghz on my 4800mq barely results in acceptable temps.

I don't mind throttling as long as long as it doesnt stay at tjmax temp, see this video i made for a fix (a.k.a adaptive voltage): 

switches 4ghz to 4.5 back and forth, 4800 can run at lower voltages, so yours should run much cooler, 180mv at 4ghz, and 270 at 4.5 ghz, insane 360 at 4.6 pulling 110 watts which worries me about mobo components and warping XD

Btw, you can always flash the latest uC after opening the FVIR (bin recursion bug) multiple times till u reach 80x with the old uC

Right now i'm on the latest, will add note...

On 4/4/2016 at 7:18 AM, Prema said:

...

This old code stuff is only good if you wanna fool around a bit and run a couple benches...

On 4/6/2016 at 8:33 PM, MC-Sammer said:

I flashed in DOS and the uC is no longer detected, though I'm not able to bump up the multiplier in throttlestop; I think it's because I have the ICC watchdog installed. I'm gonna try removing that some time soon and give it another try.

 

 

 

I'm also in the process of doing some cooling mods to my Laptop, and I have a bunch of mini sinks, sandpaper, thermal pads, etc, on order so I'm gonna have a busy weekend

 

My plan is to document it, along with temp differences, in another thread.

 

I also got 2 sets of CMSX8GX3M2B2133C11, which I'm hoping works well (and at 1.35V) because my current config is less than optimal.

 

Your uC is stock, not undetected, I'm on 06 right now since I no longer need to do overvolting over 250mv

Disable C1E for stability(as it introduces scaling issue), and BD PROCHOT for throttling (down to non turbo clock speeds instead of decreasing frequency by 1 step).

Ofc, P157 and P150 has the worst airflow design, try removing the back-plate and see the temp drop down, 10 in my case(punched holes), this is what i have done to both gpu/cpu and decreased around 5c: 

Check this awesome blog: http://null-bin.blogspot.com/2015/04/extensive-clevo-p150sm-cooling-and.html

Glad it worked for you!

6 hours ago, Prema said:

The OS microcode updates contain many bug fixes and improvements down the road and are generally a good thing.

We could simply stop the OS from updating by customizing the BIOS microcode, but I would't recommend using old microcodes as a daily driver.

This old code stuff is only good if you wanna fool around a bit and run a couple benches...

Thankfully, my system up time has been 3 days on OC, though I also think MC can improve performance, stock frequency bench on stock MC < latest, I also noticed that in newer MC , CPU voltage has been decreased, 45w mc17 cinebench vs 48 with 1% power throttle current (00) without modifying anything

40 minutes ago, MC-Sammer said:

@Prema while we're on the subject (slightly) how to you feel about Microcode? I know you have version 17 in your p15xsm bios, but the mcupdate_GenuineIntel.dll automatically puts in a newer version (I think mine was deploying something like a1). Would you put instructions on removing that file on your main blog, or would that be a legal hubbub too?

Done:

On 3/30/2016 at 9:03 PM, tmash said:

Windows Mod to remove auto update microcode on boot:

 

1. Click on Start
2. Type CMD in the Search box
3. Right-click on CMD and choose Run as Administrator
4. In the Command Prompt window and change to the directory where the file is located. To do this, use the CD command. You can follow the example below.
5. to change to the Windows\System32 directory you would enter the following command and Press Enter

cd \windows\system32

 

Now use the DEL command to delete the offending file. Type DEL mcupdate_GenuineIntel (and backup) Done!

EDIT: oops, thought you quoted me, sorry...

But why would you want to do it when stock (clean) bios has the update anyway? Would be useless without posting how to remove mc from bios first XD

2 minutes ago, MC-Sammer said:

Flashing with afuwingui (64 bit) initially didn't take (the first microcode revision was gone, according to MMTool, but there was another one present), but I came across https://downloadmirror.intel.com/11979/eng/Release_Notes_P09.10.TXT and discovered that microcode isn't stored in the main block

 

 

And I know, I should be using a USB to flashing, but when I'm in DOS my SO gets all paranoid lol

 

I'll report back on findings, if I have any.

Yeah, i think the microcode in volume 4 (in my case, see scr) gets uploaded when using UEFI

And btw, do you have any idea about the manufacturing date of a new battery? Having a model number P150HM sounds suspicious, that over 4 years old (-30% of capacity due to shelf life XD) 

Guess its going to cost me to get a new one, should fail soon in case it is over 4 years old :/ , properly disassemble , rip out the cells and replace, or end up getting unofficial batt: http://www.replacement-laptop-battery.co.uk/clevo-p150sm-battery-36615.html (still not cheap, yet cheapest i could find)

I was able to get ~6 hours on batt too with cpu undervolting and under Linux first month, and it degraded fast till mid 2015 (since Feb 2014)

I'm having the same problem with the battery case, except that i kinda bent the case, your post is also useful for knowing the boundaries to cut without making fire party XD

Why didn't you use Samsung cells instead of Panasonic? I think they have longer service life

Clevo P150SM

i7 4700MQ @ 4.6GHz, 1.2 vcore (stock microcode glitch), Cinebench ~830-840 > DT [email protected] 

RAM 16GB @ 1866mhz

GTX770M , core @1154MHz, Mem@2554MHz (samsung), core voltage offeset: +137.5mV (working on external PCIe MXM adapter for Titan X)

Collaboratory Liquid Ultra TP on CPU/GPU, arctic thermal pads on cpu vrms and gpu memory

85/120 hz Chemei modded display

128gb crucial m4 msata (SSD) + 1tb toshiba HDD

240w power supply

SED SSD Encryption (+password) BIOS mod

Optimus GPU mod to use jpeg compression instead of mux for latency (useful for demanding fps game)

Punched some d**n holes on cpu&gpu bottom cover vents, decreased -5c(-12c without cover) (highly blame Clevo on that)

Modded 8-cell battery pack with Samsung 18650 cells (old: panasonic), achieving ~6 hours of web browsing on battery and -100mv on cpu(@1.5ghz max with dGPU off)

5 hours ago, Prema said:

This modding method has already caused legal problems and a bunch of 'cease and desist' orders by Intel in the past involving major ODMs.

The original source of this 'mod' have been AsRock, Asus & MSI in 2014. Who renamed the v7 microcode to v99 to block it from being updated in Windows. 

I have removed the Link which is using the premamod as its base. Kindly do not alter and distribute the premamod version in anyway and instead use a clean BIOS base to do this modification. We can not legally be implicated and accidentally considered being the source of distribution. 

Sorry that I have to be 'that' guy...but there are lines we can not cross. One being this, another unlocking g-sync or converting GeForce into Quadros. 

 

Thank You & ENJOY! 

Hey thank you for mentioning this, sorry to be dumm about legal issues this might introduce and any wrong attempt I have done.

1 minute ago, MC-Sammer said:

Agreed, though now you've sent me down a rabbit hole of trying to find newer Intel iGPU GOP drivers and trying out different SATA driver Option ROMs

Haha, though I would be careful modding these if I dont have SPI flash programmer and soldered header :p, screwing up microcode part will change file size and UEFITool wont be able to rebuild which makes it kinda safe...

It would be really nice to try this on other Haswell CPUs, results would be interesting, guess 4900mq can go 4.8ghz stable due to having high quality ASIC compared to 4700mq, 200+mV to reach 4.6 should be less with better cpus

10 minutes ago, MC-Sammer said:

@tmash I was doing some more digging and found that (according to them anyway) the 07 Microcode update is the last to support overclocking

 

http://www.win-raid.com/t154f16-Tool-Guide-News-quot-UEFI-BIOS-Updater-quot-UBU.html

 

 

Perhaps something to investigate? I'd imagine all of the Microcode for haswell is universal (as an architecture), but I'm just getting into BIOS editing, so I'm just guessing

 

And who knows, maybe one of those microcode updates was actually important!

 

@Prema

 

Maybe this is something that could be rolled into your mods?

07 is for 36x on all cores, not +2 bin everytime you read ME settings, anything below 07 works, 07 just for locking all cores to 36x, and yes they are universal, what do you mean one of the microcode is actually important?

EDIT: forgot to mention that, 06 to 01 does not allow adaptive vcore increase over 250mV, 00 does

34 minutes ago, MC-Sammer said:

In

 

 

the block you have highlighted in your selection starts with a "01" not "00". Was this a typo, or do you start the selection at "00"?

 

Either way, after deletion it leaves a "01" there, I just was hoping for some clarification

yep its 01, sorry, will add note

15 hours ago, MC-Sammer said:

Oh I'm definitely gonna have to try this!

 

Thanks!

Feel free to ask any question