[BIOS/vBIOS - MODS] Prema Mod™ & Stock

[Schumsta]

Only Problem i have is that my 980ms sometimes downclock from 1.02V = 1260.9MHz to ~0.970V = ~ 980Mhz at Full Load. So its not throttling like Mr. Fox has to 135Mhz. Just downclocking to ~980Mhz.

Here are 3DMark Firestrike Tests. Sry for bad res. But i think my problem can be seen.

Firestrike Graphics Test 1:

Firestrike Graphics Test 2:

[Guest]

Only Problem i have is that my 980ms sometimes downclock from 1.02V = 1260.9MHz to ~0.970V = ~ 980Mhz at Full Load. So its not throttling like Mr. Fox has to 135Mhz. Just downclocking to ~980Mhz.

Here are 3DMark Firestrike Tests. Sry for bad res. But i think my problem can be seen.

Firestrike Graphics Test 1:

Firestrike Graphics Test 2:

Sorry I have no time to work on a fix for Alienware...svl7 has a unit to work on and I am sure he will come up with a solution for Alienware, just give him time and be patient.

[joluke@inphtech]

Latest stock BIOS/EC are in the OP!

Latest Hotkey for P3-SMA: ftp://ftp.clevo.com.tw/P37xSM-A/OTHERS/Hotkey.zip

Whats the difference between those and these:

ftp://cftp.clevo.com.tw/P37xSM-A/OTHERS/Hotkey.zip

beside the size?

Edit: from Clevo's site, all hotkey zip's have different sizes:

Europe mirror:

ftp://cftp.clevo.com.tw/P37xSM-A/OTHERS/Hotkey.zip - 70,0MB

US mirror:

ftp://usftp.clevo.com.tw/P37xSM-A/OTHERS/Hotkey.zip - 88,8MB

Asia mirror:

ftp://ftp.clevo.com.tw/P37xSM-A/OTHERS/Hotkey.zip - 158MB

Any idea why the size difference?

Edit2:

Apparently the hotkey from this mirror:

ftp://cftp.clevo.com.tw/P37xSM-A/OTHERS/Hotkey.zip - Europe mirror

is still version: 2.31.42

The mirror from US the version of Hotkey is: 2.34.49

Will download from Prema's mirror instead and check out the version number.

Edit3:

Found a mirror at this site for the version: Hotkey 3.07.18a:

http://drivers.prostarserver.com//P377SMA/HK_3.07.18a.zip

[Guest]

Clevo always updates the main Taiwan server first: ftp.clevo...

The others maybe weeks older...

[BigKid]

First of all: Thank you very much Prema for your work ! I was able to OC my GTX980M in an MSI GT72 without much efforts:

+00.0mV / + 150 Mhz Core / +500Mhz Mem

+37.5mV / + 250 Mhz Core / +600Mhz Mem

+50.0mV / + 275 Mhz Core / +600Mhz Mem (whooping 10169 in 3DMark FireStrike http://www.3dmark.com/fs/3830495 )

So for whatever reasons your modded vBios seems to suit us GT72 owners better than the one done by slv7 (no pun intended).

However - something started to mildly "hum" within the notebook when I start a 3D app and I'm wondering if you modded the base power target and if yes - to which amount/extend.

I'd also like to know this to estimate how far I dare to go with the voltage adjust... +300MHz seems to be in reach but I do not dare to go beyond +50.0mV...

(Personally I'd consider 120Watt up from 100Watt safe, 140Watt daring but still something I could try, anything more would give me a headache :-) )

I also can't change the power target - the setting is still "locked".

And I can't "undervolt" which I would love to do too for a "silent" gaming mode but every time I change the voltage setting to a negative adjustment it just goes back to "0")

(I tried both using nvidia inspector - latest version - and nv driver 344.75 ).

I would also love to see a mod of the vBios used in the GT72 by MSI - I would donate to help to get this done...

[Guest]

The MSI base wouldn't make a difference, looked at it and it's the same thing. Power Target is not needed as it is completely disabled with the mod.

Remember 1.2v is stock voltage on this chip in desktops. So just use whatever you can keep cool and it'll be fine. Will see to add undervolting in next version.

[Kommando]

Will see to add undervolting in next version.

That would be awesome! *registered only to say this*

This and a modified bios for the P751ZM is all that keeps me from buying this otherwise awesome laptop.

[omer.sak]

Hi Prema, you're an absolute legend.

Got my P650SG today (first clevo and n00b) and straight away flashed it with your BIOS and vBIOS. While the BIOS flash went okay, I think something may have gone wrong when flashing the vBIOS (I did make a GPUZ backup of the original BIOS though).

I opened the 'update' file and it detected my 980m and I did as instructed - I uninstalled the NVidia drivers and opened, proceeded with 'y' twice and it beeped and the command window quickly closed. I restarted the computer, installed the drivers and the games appear to be working as they were, as too the GPU switching - but when opened the vBIOS 'update' file again it stated that it could not find an NVidia adapter - and when I forced a program to use the dGPU, the update command window gave me '

unconfigured display adapter found device not accessible

' and the window wouldn't close.

Do you think I have a bad flash? Gaming works fine but how do you suppose I should proceed?

thanks!

[Guest]

You probably just forgot to disable/uninstall the driver again before you ran the update the second time...

Just see if you can move the core slider more than 135Mhz in NVIDIA Inspector, which means the mod has flashed successfully.

On a semi-related- side-note: Almost all semi-bricks I have seen, folks created themselves by doing something unnecessary out of fear it may not work right.

It's like people create what they are trying to prevent!

[omer.sak]

You probably just forgot to disable/uninstall the driver again before you ran the update the second time...

Just see if you can move the core slider more than 135Mhz in NVIDIA Inspector, which means the mod has flashed successfully.

On a semi-related- side-note: Almost all semi-bricks I have seen, folks created themselves by doing something unnecessary out of fear it may not work right.

It's like people create what they are trying to prevent!

Haha, thanks Prema - I guess the experience creating ones own problems out of fear - and getting them solved is an early step in a noobs path to manhood.

I did as you suggested and the vBIOS flashed fine. I do have a problem with maintaining a steady GPU clock during gaming (before your vBIOS update even) - as the GPU/CPU utilization goes up, the GPU keeps dropping down to 400MHz at perfectly normal temperatures, causing stutters in games. When idle, or out of the game, it stays rock solid at 1197MHz as it's supposed to. My previous GS60 didn't exhibit this issue. Running 344.75 with the latest Hotkey (max fan setting) and Premamods.

Just want to know if this is normal? screenshot below:

[Guest]

That's normal as the CPU/GPU only use the clocks that they are actually required in order to do the task at hand (speedstep).

[Guest]

@Rhadamanthis No problem...

[BigKid]

The MSI base wouldn't make a difference, looked at it and it's the same thing. Power Target is not needed as it is completely disabled with the mod.

Remember 1.2v is stock voltage on this chip in desktops. So just use whatever you can keep cool and it'll be fine. Will see to add undervolting in next version.

Hi Prema. Thank you very much for your reply and your great mod ! If the MSI vBios is the same thing I'm absolutely fine with that so no need for a special version.

Undervolting would be a great thing. Sometimes one want's just to have a silent mode and not every game requires full OC (or even full power).

In respect of the Voltage - I'm not affraid about the GPU itself - I know that the GPU is running at much higher voltages in the desktop versions of the card.

I'm more affraid about the resulting power draw. The traces / conductor pathes have limits, the voltage regulators on the card have limits. The ac adapter has limits. OCing is always related to getting close to those limits and using the reserves the designer left for safety reasons.

GPU Boost is a game changers as it does not also protect the card from crossing the set power limit but it also tries to stay close to the set power limit - so to me it seems that every 3D app will go somewhat "furmark" if GPU boost is doing a good job.

I would feel much more safe if I could choose to increase the power limit instead of disabling it completely - or if I could get a version of the vBios that has an increased limit of 120W or 140W instead of removing it completely.

I've read posts of people who "grilled" their MXM module using a modded vBios and Furrmark / Kombustor. And while I agree that someone shouldn't run those tools at all on Laptops - I'm also thinking that a "reasonable" power limit could have saved them... The GPU will throttle to protect itself from overheat - but what about the other parts that do not have a thermal sensor - the voltage regulators, memory, etc too ?

I will trust you if tell me that you are considereing all this as you are doing this for quiet some time now but I'd like to be sure ...

I know you have much more experience in this so anything you can share is highly appreciated. I'm occing cards for 15 years now. In laptops I occed a GTX570M, GTX770M, GTX780M by using SVETS tool. So I never actually found out how to mod a bios myself but I used mod-tools. I increased the power limits by about 20-25% and nothing bad every happened - so let's call this my comfort zone. If there was any way to mod a GTX980M bios myself I would do it - but none of the tools is supporting the bios yet. I'm not asking anyone to do something for free - I'm more than willing to pay for a custom bios that gives me best of both worlds - my personal "airbag" (power limit) and yet some OCCability.

With your BIOS my laptop started to "hum" when a 3D app is starting - this COULD be indicating that something is operating at it's limits...

In addition a changeable power limit is a good way to easily set the laptop for a "silent gaming mode" - on my unlocked GTX780M I'm setting the power limit to 75% and the heat target to 70 degree celcius, reduced the voltage and also reduced the clocks a little and my system / the fans are much more silent... I know I sacrifice 3d power but for many games it is still sufficient...

[BigKid]

Moving this reply chain to this thread (where it belongs):

is that score 10169 overal firestrike score or graphical score ?

Overall score.

your gpu should be 1313mhz with +275 and your mem 6200mhz with +600 ...all i really want to know can you play games without crashing or freeze with those settings especially with +50.mV voltage which is really not enough to feed the gpu...i was using 1365mhz gpu + 6000mhz mem +100mV results an overall 10197 fire strike score and 13083 gpu score

I do not know yet if games run stable... I'm currently running games at stock settings for the reasons/worries I explained above...

But I found those values very easy and they might not even be the limits... I know that with +50,0mV +275Mhz is working and 300Mhz is not... But 290 might still work... so I'm quiet confident that this is about what will be possible stable in games.

OCCing is very dependant on the actual GPU/Chip quality and that is like a lottery.

Since I'm also able to OC +150Mhz Core, +500 MHz Mem without any voltage increase I might just have a good GPU.

But bottom line - results are never comparable but just a very rough indication of what is possible...

[Guest]

@BigKid

Volt + Temp ≙ TDP

With constant voltage but rising temperature the TDP rises.

With constant temperature, but rising voltage the TDP rises.

So being able to limit or regulate either one of them is enough to set safety limits.

Since the vBIOS allows us to lower the temp values, we can set a "silent limit" that way when overclocking.

(in addition to being able to choose our own clocks and voltage according to our systems capacity).

The FANs only care about the temperature, not the TDP.

The DTP values as we would see them in a software only represent a rough estimated power draw.

The actual power draw is defined by many factors like each of the systems components and even the efficiency of the PSU...

Also if the systems AC Adapter has breached its capacity it'll just shutdown to protect itself.

If we would give people slider to raise or disable all three values, then of course they can kill their cards with Kombustor, Furmark etc...

(Even then BIOS/EC safety measures would still kick in and shut the system down, but that maybe too late).

Also remember that even the highest possible overvoltage is just 0.138v over stock.

[BigKid]

@BigKid

Volt + Temp ≙ TDP

With constant voltage but rising temperature the TDP rises.

With constant temperature, but raising voltage the TDP rises.

So being able to limit or regulate either one of them is enough to set safety limits.

As always: Thank you very much for taking your time to reply.

I might have got you wrong: Maybe you there still IS a power limit - and I just got you wrong but I understood one of your posts in a way that the power limit is disabled with your vBios.

So I agree but for your last point (or maybe we agree already and this is a misunderstanding):

The reason is only the GPU has a thermal sensor. The heat of the GPU is depending on how much I OCC it and how much voltage I apply. So we agree here. Also agree that the hotter it gets, the more power it will need for the same tasks. So TDP goes up. Agreed.

However: The final GPU temp is also depending on how powerful my cooling system is. The GT72's one is quiet powerful and will be able to keep the card below 87 degree for quiet a long time by increasing the fan speed even when the power draw is exceeding all (default) limits by large. As far as I know only the GPU has a thermal sensor but not the traces, nor the voltage regulators. Those could be above "healthy" limits already for quiet some time. This is what I suspect to have happened to those that actually damaged their systems. Some of them have been able to repair by cooking their cards in the oven or by applying other means of extreme heat so maybe just some soldering points got too hot and disconnected - and baking the card fixed that again (this is a general observation and by no means I'm saying this happened with YOUR BIOS)

The power limit however (even when beeing a rough measurement) is much more indipendent of the cooling solution and in my opinion a better "airbag" - so I'm saying IF it is possible I'd prefer a version that allows to infuence the power target instead of the temp target but within healthy limits. Would it be possible to create a vBios that allows a certain range ? The GTX980M should be about 100Watt default so a range of 75 to 125 Watt should be quiet safe...

Since it is possible to change the fan tables in the firmware of the GT72 with some research someone could come up with a ec firmware that has even more agressive fan curve - or just use the "cooler boost button" to set the fan to full speed. This would keep the card cool while it could be even more off limits for everything but GPU temp already.

But again - we might already agree mostly here...

Since the vBIOS allows us to lower the temp values, you can set a "silent limit" that way when overclocking.

(in addition to being able to choose your own clocks and voltage according to your systems capacity).

The FANs only care about the temperature, not the TDP.

That is another thing. I can't set the temp limit - that slider seems to be locked on my GT72 with your vBios, nv driver 344.75 and nvidiainspector (lastest version - last checked about 3 days ago).

Could you please confirm which tool and driver version should work ? So I could retest this ?

The DTP values as you would see them in your software only represent a rough estimated power draw.

The actual power draw is defined by many factors like each of the systems components and even the efficiency of the PSU...

Also if the systems AC Adapter has breached its capacity it'll just shutdown to protect itself.

If we would give people slider to raise or disable all three values, then of course they can kill their cards with Kombustor, Furmark etc...

(Even then BIOS/EC safety measures would still kick in and shut the system down, but that maybe too late).

Also remember that even the highest possible overvoltage is just 0.138v over stock.

100% agreed again (with the exception that I feel like the DTP is still better because it is independent of cooling solution etc. and might still reflect better how far off the default the card is actually operating) - so maybe this is really mostly just a misunderstanding on my side.

[Guest]

As always: Thank you very much for taking your time to reply.

I might have got you wrong: Maybe you there still IS a power limit - and I just got you wrong but I understood one of your posts in a way that the power limit is disabled with your vBios.

So I agree but for your last point (or maybe we agree already and this is a misunderstanding):

The reason is only the GPU has a thermal sensor. The heat of the GPU is depending on how much I OCC it and how much voltage I apply. So we agree here. Also agree that the hotter it gets, the more power it will need for the same tasks. So TDP goes up. Agreed.

However: The final GPU temp is also depending on how powerful my cooling system is. The GT72's one is quiet powerful and will be able to keep the card below 87 degree for quiet a long time by increasing the fan speed even when the power draw is exceeding all (default) limits by large. As far as I know only the GPU has a thermal sensor but not the traces, nor the voltage regulators. Those could be above "healthy" limits already for quiet some time. This is what I suspect to have happened to those that actually damaged their systems. Some of them have been able to repair by cooking their cards in the oven or by applying other means of extreme heat so maybe just some soldering points got too hot and disconnected - and baking the card fixed that again.

The power limit however (even when beeing a rough measurement) is much more indipendent of the cooling solution and in my opinion a better "airbag" - so I'm saying IF it is possible I'd prefer a version that allows to infuence the power target instead of the temp target but within healthy limits. Would it be possible to create a vBios that allows a certain range ? The GTX980M should be about 100Watt default so a range of 75 to 125 Watt should be quiet safe...

Since it is possible to change the fan tables in the firmware of the GT72 with some research someone could come up with a ec firmware that has even more agressive fan curve - or just use the "cooler boost button" to set the fan to full speed. This would keep the card cool while it could be even more off limits for everything but GPU temp already.

But again - we might already agree mostly here...

That is another thing. I can't set the temp limit - that slider seems to be locked on my GT72 with your vBios, nv driver 344.75 and nvidiainspector (lastest version - last checked about 3 days ago).

Could you please confirm which tool and driver version should work ? So I could retest this ?

100% agreed again (with the exception that I feel like the DTP is still better because it is independent of cooling solution etc. and might still reflect better how far off the default the card is actually operating) - so maybe this is really mostly just a misunderstanding on my side.

Of course there is additional load with additional voltage to the MXM board and power circuits, but nothing will measure or adjust that, not a stock nor a mod vBIOS.

(Apart from that there has been no one with any damage with this vBIOS...that's other GPUs with other vBIOS)

You have to drop the believe that the TDP slider shows you any real world value.

It is an imaginary number (100% on stock vB is already "175%-ish" on the other mod vB, when you set it to 125% that is 25% on top of that "175%-ish").

Temp slider is tested and works just fine with latest Inspector and 344.75...that's all you need.

That way you can even make voltage and clocks throttle at 50c if you want...

[BigKid]

Again: I highly appreciate that you take your time to explain things to semi-noobs like me

Of course there is additional load with additional voltage to the MXM board and power circuits, but nothing will measure or adjust that, not a stock nor a mod vBIOS.

(Apart from that there has been no one with any damage with this vBIOS...that's other GPUs with other vBIOS)

Yes - no damage known with your vBios - I updated my OP even before your post because I didn't want anyone to get that impression ! I'm sorry...

It is an imaginary number (100% on stock vB is already "200%-ish" on the other mod vB, when you set it to 125% that is 25% on top of that "200%-ish").

Sure... I'm not disagreeing here... It's just that I modded my GTX780M bios to 120Watt. I've "seen" that value in the cards bios... And I'm fully aware that 100% are now 120 Watt instead of the 100 Watt it had previously.

All I'm saying is that I'd prefer to have a limit I know and could tweak a little - instead of having one I do not know

You have to drop the believe that the TDP slider shows you any real world value.

Temp slider is tested and works just fine on with latest Inspector and 344.75...that's all you need.

That way you can even make it throttle at 50c...

Ok... I will revisit my system before getting back to you on this - it didn't seem to work for me yesterday but this might either be a problem beeing located between keyboard and back of the seat (-> me) or something special with the GT72 (which I consider less probabale - I will just check and get back with my findings).

Maybe something changed in the approach nVidia has taken from the GTX780M to the GTX980M. For the GTX780M I was able to set the power limit in watts using SVETs bios tweaker. And I learned and experienced (also that might have been wrong information) that this seems to be really a power limit which the card seemed to actually be measuring and it's including the total power draw including memory. And actually I could see that increasing voltage and mem clock did actually result in lower core clock speeds until I increased that power limit also. That's why I'm harping on this power limit. It is something I "saw" and "experienced" whereas the GTX980M bios is a black box to me.

And the GTX780M is advertised to use GPU BOOST 2.0 as is the GTX980M - so officially no change there...

So please take my excuse for seeming so stubborn - I just want to be sure I understand: You are saying that the card is NOT measuring actual power draw but just estimating ?

[Guest]

...

Maybe something changed in the approach nVidia has taken from the GTX780M to the GTX980M. For the GTX780M I was able to set the power limit in watts using SVETs bios tweaker. And I learned and experienced (also that might have been wrong information) that this seems to be really a power limit which the card seemed to actually be measuring and it's including the total power draw including memory. And actually I could see that increasing voltage and mem clock did actually result in lower core clock speeds until I increased that power limit also. That's why I'm harping on this power limit. It is something I "saw" and "experienced" whereas the GTX980M bios is a black box to me.

And the GTX780M is advertised to use GPU BOOST 2.0 as is the GTX980M - so officially no change there...

So please take my excuse for seeming so stubborn - I just want to be sure I understand: You are saying that the card is NOT measuring actual power draw but just estimating ?

No-throttle can not be achieved by simply letting the power target do its usual work...we could raise it a bit with a software but it would then still throttle at higher clocks.

NVIDIA is not sitting idle and has more barriers in place for us mobile user. We are doing everything ourselves in hex...

Like I said the value for TDP base on the other vB is already on 200% (minus the undervolt), which according to your calculation would already be 216W (2x108W), yet we still have to raise the slider further in order for it to not throttle on higher OCs...you see it has no real world relation to actual Watts used by the GPU or entire MXM board...in a working non-throttling Mod vBIOS it is nothing more than a left over bogus value.

If we want to know any real world values then we can only measure the systems power draw on the DC side of the AC Adapter.

[BigKid]

No-throttle can not be achieved by simply letting the power target do its usual work...you could raise it a bit with a software but it would then still throttle at higher clocks.

NVIDIA is not sitting idle and has more barriers in place for us mobile user. We are doing everything ourselves in hex...

I'm quiet sure of that. OCCing a mobile card was always harder to archive (or not even possible in the past). I think that SVETs tool just autmates a lot of the magic you are doing using a hex editor...as it's capabilities slowly expanded over time and not every bios was supported right from the start - even within the same generation of cards.

So what I did with my GTX780M as I understand it is that I just expanded the OC limits but still had a throtteling vBios - it just throtteled later. Whereas you vBios will never ever throttle (unless the GPU starts to overheat) ?

Like I said the value for TDP base on the other vB is already on 200% (minus the undervolt), which according to your calculation would already be 216W (2x108W), yet you still have to raise the slider further in order for it to not throttle on higher OCs...you see it has no real world relation to actual Watts used by the GPU or entire MXM board...in a working non-throttling Mod vBIOS it is nothing more than a bogus value.

Before I answer - what are you refering to with "other vB" ? Your mod - slv7's mod - ... ?

[Guest]

I'm quiet sure of that. I think that SVETs tool just autmates a lot of that as it's capabilities slowly expanded over time and not every bios was supported right from the start - even within the same generation of cards.

So what I did with my GTX780M as I understand it is that I just expanded the OC limits but still had a throtteling vBios - it just throtteled later. Whereas you vBios will never ever throttle (unless the GPU starts to overheat) ?

Before I answer - what are you refering to with "other vB" ? Your mod - slv7's mod ?

I have no idea what svets tool does, never touched it...

Yes I took the only other GTX980M vB mod as example.

EDIT: Ah man, I just saw that I spent 3 hours on this...that's far more than this topic deserves.

[BigKid]

Like I said the value for TDP base on the other vB is already on 200% (minus the undervolt), which according to your calculation would already be 216W (2x108W), yet we still have to raise the slider further in order for it to not throttle on higher OCs...you see it has no real world relation to actual Watts used by the GPU or entire MXM board...in a working non-throttling Mod vBIOS it is nothing more than a left over bogus value.

If we want to know any real world values then we can only measure the systems power draw on the DC side of the AC Adapter.

Hmm... Interesting... So probably the systems total power draw is now able to exceed the limits of the EC controler even at stock settings of that other bios mod and thus the whole system starts to throttle - that would explain that some (if not to say all) GT72 owners are having issues especially in the power hungry combined test of 3d Mark Firestrike with that other vBios.

The EC limit of the GT72 is not known but the GT70 had a limit of 180Watt or 210Watt with a special EC firmware as far as I know. With the AC adapter of the GT72 beeing labeled as having 230Watt (might not be 100% accurate) I assume it's something about 200 Watt...

Even at stock bios I saw the GTX980M throtteling with the reason beeing "power limit" - GPUz seems to be able to show the throtteling reason and is giving different feedback for "power", "thermal" and "other"..

I assume that you are doing something different ? You have to be doing something different as your mod is working in our systems at stock settings and with OC.

I'm just Interested in all you can share without giving the recepie of your secret sauce :-)

[captn.ko]

after my vbios flash last week, yesterday was the first day where i had time to game a longer time on the P505.... but it wasnt a very happy time.

I have strange/hard Freezes/ FPS Drops while gaming. graphics freeze and sound stops for 1 second. This happens every 3-5 minutes.

Bios Update to 1.03.05 is flashed...same problem.

With and without OC....same problem

drifferent graphic driver...same problem

anyone has the same behavier with his book?

[Fidanza]

The MSI base wouldn't make a difference, looked at it and it's the same thing. Power Target is not needed as it is completely disabled with the mod.

Remember 1.2v is stock voltage on this chip in desktops. So just use whatever you can keep cool and it'll be fine. Will see to add undervolting in next version.

You have to be wrong dude.. and as i said before i get 9.2k with stock and 8.4k with your modded vb even using the same frequencies (+135gpu +200mem) from 3dmark which shows your vbios is not perfectly suits msi gt72..its better you should check again the vbios i sent

[Fidanza]

+50.0mV / + 275 Mhz Core / +600Mhz Mem (whooping 10169 in 3DMark FireStrike NVIDIA GeForce GTX 980M video card benchmark result - Intel Core i7-4720HQ,Micro-Star International Co., Ltd. MS-1781 )

So for whatever reasons your modded vBios seems to suit us GT72 owners better than the one done by slv7 (no pun intended).

However - something started to mildly "hum" within the notebook when I start a 3D app and I'm wondering if you modded the base power target and if yes - to which amount/extend.

I'd also like to know this to estimate how far I dare to go with the voltage adjust... +300MHz seems to be in reach but I do not dare to go beyond +50.0mV...

(Personally I'd consider 120Watt up from 100Watt safe, 140Watt daring but still something I could try, anything more would give me a headache :-) )

i looked at your 3dmark score and saw smthing interesting your overall score is very close to me mine is 10197 overall but your graphics score 12373 but mine 13083..?

your other score is better than me so you get 10169 even we have the same cpu i7 4720hq

my values are 1365mhz gpu 6000mhz mem 1.1volt(+100mV) causes to make max 76C, even in watch dogs i reached 83C then with cooler boost reduced back to 72C no problem..

and your +50mV is not enough to feed to gpu in real time test just like playing games and you get crash or freeze with your values, dont trust 3dmark