Ethrem

Temp readings are probably not adjusted in this vBIOS and display wrong values. Sounds like it'll throttle when reaching 87c just as the stock vB.

Thanks Prema. It throttles worse than the stock when it hits 87C. Instead of throttling down in increments it drops straight to 405MHz. I didn't want to bug you for a 1v version of your vbios but this one drops my temps with max fans down to 73 and 71C in Heaven and it takes quite awhile to hit 87C without max fans so my cards definitely like the lower voltage.

Is there any reason why this vbios drops to 405MHz when it hits 86C? I love the fact it runs at 1v (although it isn't stable out of the box at 1202 it works perfectly at 1126) because I have cards that get insanely hot. The problem is when they hit 86C the clocks drop to 405MHz until they drop to around 80C and then the cycle starts over. It isn't a temperature target issue as that appears to be set at 99C out of the box unless it's being read wrong?

Nope, BIOS voltage control is only in SM-A & ZM systems...I'll have to play with BET one day and see what is possible to add...

That's crazy. I guess we have to get desktop CPUs now if we want an enthusiast machine... Well as long as overclocking the video card is still possible... Otherwise enthusiast laptops are gone after the SM-A and ZM. Sad day...

@Ethrem your system comes with an XTU BIOS interface which works a s a proxy for the NVRAM. P6 and W-series don't have that and XTU seems unable to reach NVRAM unless Legacy mode is used.

I would be happy to be proven wrong but I can only test on W230SS, where XTU settings didn't stick either unless XTU was started again.

That's crappy of them to take that away. Does your BIOS have voltage offset settings on those machines at least? It would get annoying having to load a profile every time I booted the machine.

I read that the IC Diamond is good, but could have some scratching effect on the chip. Do you know if it's true?

It does scratch if you aren't careful removing it. Get the Arcticlean kit and let the Arcticlean 1 sit for a minute or two (you'll see the paste break down and start to mix in), take a qtip and gently roll it around to pick up all the mess, repeat until it's done then put number 2 on there to get the surface ready for new paste. Same process with the heatsink.

It can leave some pretty gnarly scratches if you try to scrub it off... I have a big long scratch that goes from the left side of my die to about 3/4 across on my 4940MX.

Try to use an older XTU version...

EDIT: Actually if you are using Windows in UEFI mode it prevents these settings from being permanently written to BIOS.

You need to install Windows in Legacy mode in order for XTU changes to be 'permanently' in BIOS.

What? My XTU settings have always been permanent unless the system goes unstable and crashes and I have used UEFI the whole time. Never had an issue with it sticking, even without changing 8.1's default boot and shutdown behavior. My understanding is that XTU interfaces directly with the Intel management interface. Unless that's a secure boot issue? I don't use secure boot for obvious reasons.

Added the Premo BIOS to my new Eurocom X8 based on the P377SM-A. I have to admit, my bum was twitching but the instructions were superbly clear. Thanks Prema. If I build up enough courage then I'll flash my Gcards

Is there a way to remove speedstepping to force a constant frequency on the 4940MX? Is it simply to disable speedstepping?

In a word. No.

More detailed answer - if you can get the heat under control, you can raise the TDP and current limits until they don't throttle anymore. Realistically, that's not possible without serious hard mods of the machine and liquid ultra. If you set your TDP to 90W and give it 150A, most 4940MX chips with do between 4.1GHz and 4.2GHz on stock voltage, however, they will run into the 90s and be at risk of throttling due to thermals at 95C. Liquid ultra can help but only so much and some loads will still blow past that 90W limit which will cause it to throttle and if you're going to raise the TDP that high, you might want to consider a second PSU and the converter box to run 660W to it, especially true if you plan to overclock the graphics card at all.

What I found is that setting my chip to 80W TDP and 4.1GHz with 150A current limit will result in me averaging 3.9GHz under heavy load (the stock behavior of my particular CPU is to fall down to 3.4Ghz with a heavy load) and between 4 and 4.1GHz with gaming. This required liquid ultra and resulted in average load temps of 89C.

Right now I'm just running it stock with an 85mv undervolt for 3.6GHz under load... Liquid ultra is a pain in the ass and not worth the risk to my machine. Basically, the 4940MX is a huge waste of cash in the SM-A. The only machine that seems to be able to handle them properly was the Alienware 18.

Well they're still making mobile quads, but just terrible HQ ones.

And MSI's GT60/GT70 is still running too. They just cannot make board revisions or anything to them, as that'd require them to make BGA sockets.

But yeah, Intel wants integrated only. Only Clevo feels like bothering to use the desktop chips. It's the end of a long era. I can hope it'll rise again after being killed like a phoenix, but I doubt it. I just do.

I've always associated MQ with mobile quad, sorry for not explaining myself lol.

Is Nvidia afraid that their next graphics card (Gtx1080m?) perform just a little more or likewise as a overclocked gtx980m? If Nvidia manages to help laptop manufacturers to stop overclocking of their mobile chip, maybe more people will buy a new laptop with next generation mobile chip...

Nvidia already nerfed the power circuitry on the 980M along with using low voltage memory to offset overclocking gains a bit. But if they can get manufacturers to block it entirely, no doubt they'll go that route. Clevo locking down with BIOS does not surprise me at all and it wouldn't surprise me to see it activated out of the box on Sager models. Sager really hates replacing video cards... Or any parts really. The hassle I went through with my machine and their warranty process... Yeah.......

Even further with AMD the ultimate mobile fail, I'm guessing we will see two Maxwell rebrands and won't see Pascal until the end of 2016 or even early 2017. They have no motivation to do so. My prediction is overclocked 980M and then full GM204 before we even hear of mobile Pascal unless AMD comes out with something in 2016.

This is the whole point. Clevo really doesn't care. I mean they make powerful systems, and regardless of overclocking the GPUs are powerful. But their general locked BIOSes and other such things just show they're more caring of their corporate customers I'd say.

*sigh* I called it. I called it last year when this was gaining traction and it's shown up now, in full force. I can't even say to go MSI because MSI is still using HQ chips, even though they promised they'd never block OCing with vBIOS etc.

Can't really say anything about the GPU. Intel is not making a mobile quad anymore. I half suspect that Clevo is using chips they bought in bulk and that when those are gone, it will be desktop or HQ as well.

I saw it coming too but it's still sad to see the end.

Yeah. This is my last gaming laptop. I'll keep it probably for as long as relevant, but the market just doesn't seem to care as a whole, and the makers are dead set on breaking us.

Yeah Clevo adding a video card overclock bit block to their system BIOS, regardless of whether Prema can get rid of it or not, is the final nail in the coffin for laptop gaming for me.

Next announcement prediction: encrypted system BIOS.

Not a good sign... *sigh*

no issue for mine although it runs quite hot under BF4, roughly 85deg after more than 10 minutes... but perf is here and no throttling.

About your temperature topic dear prema, I will not help you but I can bring some concerns other than overclocking one.

I have been working in smartphone SOC since 8 years now, especially in power and was into the deep when going 1Ghz, 45 to 28 nm, and we have alsways struggle between speed and thermal question.

Generated power are dynamic and static type, dynamic means transistor switching power (so increasing with F*V^2), and static means leakage, increasing with V^4 and temperature.

At the time I was working on early 28nm, around 90 degree C, the generated static power could be more than dynamic depending on chip quality and size, and this must be dissipated otherwise it could by itself lead to thermal runaway. I bet that this was a concern at the beginning of 28nm and now the process has been improved. I was also working on different package and size than our preferred video accelerator. Then the thermal policy inside vbios is supposed to detect quick increase of temperature and should be able to shut voltage down to prevent thermal runaway, otherwise it leads to destruction.

On the 980m, I saw a very thin package (we can barely see the transistor over it and we just place a not-so-massive radiator and heat pipes...

At the end, it is just a matter of how deep the user will use its game card... what kind of game it will use.. how much juice it will get from it.

-benchmarker will always get the maximum, whatever the temperature. For them, a Fn+1 exists, thanks!

-BF4 and other very demanding game will tolerate noise, but for portability reason, they could play in rooms where 45 to 50db is too much for others, so for them it would be better to have a high speed (may be 2/3 - this is arbitrary) as soon as possible so that to reach the lowest end temperature and avoid reaching loud fan speed. Let's admit end temperature 80 degree +-2. For them we should trigger high speed fan very soon (50 degree?)...

-less demanding game will have a lower end temperature (65 to 75 deg), so will make less noise. For them, a more progressive and regulating temperature could be sufficient, such as letting the chip getting warm, because even at max freq and max voltage, it will not use 100 performance and will not generate too much power. Then the warmer the chip is, the more power it will dissipate (at 90degree, a radiator dissipate more power than at 40)... For them you can set a moderately high fan speed until it reaches 90deg.. and go higher until end temperature is found.

-again less demanding game (pacman, tetris) could be run on integrated video .

I just let my fingers type, this could be no sense at all for you, but in case it could help... From my point of view, it should be a per-game thermal design - which is probably the case of standard driver...

The issue with the mod being adjusted is because Clevo changed the fan tables in the system BIOS. Fans used to kick in at 87C, they now kick in at 90C so people were throttling at the 87C mark when the fans didn't kick up. Clevo's fan management leaves a lot to be desired. Max fans is obnoxious. I read 67db coming from my P377SM-A with max fans. Best behavior for me is auto fans which top at 52db (still loud by any standard) but the chip throttles before auto fans can kick up. No idea why Clevo has done this...

QA problems with Eurocom...? gulp!

Advice noted. Thanks

Not Eurocom specifically, there have been issues with the new nVidia cards. Happens with any new video card but you want to test it before you push them with a vbios mod. The biggest issue seems to be coil whine but there are at least three cases I know of where the cards are running much hotter than they should. I happen to have a pair myself.

Just ordered a Eurocom X8. Would you all recommend me to flash Perma's VBIOS straight away?

After you have verified that there is nothing wrong, yes. There have been QA issues popping up. Make sure the system is performing without issues before you flash mods and of course make sure you back up your stock vbios.

I'm planning on trying the mx-4 myself and have ordered some. I'm kinda curious on how well this will work. Mostly cause I'm coming from using artic silver 5.

Mx-4 is the worst thermal paste I've used thus far. IC Diamond and GC Extreme are neck in neck with the latter not scratching the die (although it tends to be harder to find and more expensive, IC Diamond is great if you don't want to repaste for a long time and if you get Arcticlean kit to remove it and follow the directions it won't scratch). MX-4 starts off strong but pumps out within a week or two because it isn't designed for the low pressure mounts in laptops.

Stock Clevo fan uptick with the new cards isn't until 90C for whatever reason (acoustics I guess) where it was 87C previously which is why your fans never hit max.

I am reworking the temperature table of my GTX9xxM Mods...too many systems run the fan at automatic max (80%) only at 89c, so the current 87c throttle has to be raised a bit.

I am thinking of 90c or 92c, what do you guys want?

I say 92C. Some of these cards are running pretty hot compared to the rest (mine and I know of two others with hotter cards) and it would be nice to be able to set the throttle point higher. We can always change the temp target down if we don't like the high temps but my master likes to run right on the border between 86-88 and then it's stutter time.

i checked some sites says the undervolted msi and lenovo gtx 860 bios how is that if nvidia blocked undervolting?is it vendor specific or can we flash old vbios to make it work again ?

MSI made their own voltage table that is undervolted and my guess on the 860M is that nVidia saved that for Maxwell 2.0

Hi my w230ss is on perma bios. Today I tried to undervolt my gpu but can't find any way to change values . I've tried nvidia inspector but still no success. Anyone can help pls?

Nvidia disabled undervolting on Maxwell vBIOS :/

It doesn't matter if the BIOS doesn't recognize the card if Windows does. Sager sent my 9377 back to me with 980Ms and no BIOS support, still worked.

I just checked though and that's an EM isn't it? I thought the EM only could support 6 and 9 series cards?

damn ethrem, do u EVER get lucky with ur hardware? i keep reading those epic fails of yours @ components running super hot, throttling, failing *lol* [emoji14]

Tell me about it........ I checked and the surface contact is perfectly fine so I have no idea what the issue is. I'm not going to mess with it for a few days but when I do I'm going to see if there's any difference using Shin Etsu.

At least you don't have my machine... I have core throttling on the new drivers despite repasting twice now with GC Extreme. I suppose it's possible my GC Extreme needs to be replaced with a new tube? It's actually running hotter now than the stock cake of IC Diamond they used (and I mean *cake*...)

However, based upon a recommendation from another 4790k owner in another forum, I disabled virtualization in the BIOS. Presto! My bus speed is now pegged at 99.76 MHz! I have virtualization enabled on my other laptops, and it seems to have little impact, but on this processor (and perhaps all Haswells), the effect is very noticeable.

Its not a Haswell thing it's a Clevo thing. My desktop 4790k is set at 100 BCLK and stays at 4700MHz with virtualization enabled. I can't speak about other laptops but spread spectrum is not enabled by default on my Z87 Maximus VI Formula and doesn't cause any issues. Enabling it will actually make my overclock unstable. Haswell is much more sensitive to core frequency instability than previous chips - my 4.7GHz overclock is totally stable - at 4.7GHz. If I turn on power management instead of high performance, it will BSOD at some point when the clock frequency is going up and down.

@Prema Any word on a vbios that has undervolting on 980M? My cards run on the hot side and undervolting would probably help a lot, especially with my master that does 1321 at 1.062v. Thanks

Thanks for that update. it is moving so fast!

Sounds like good news for new buyers. I wonder if there will be official vbios updates?

Highly unlikely. It would undoubtedly lead to a lot of bricked cards.

The people who care about overclocking would know how to find a mod or would figure it out.

I try ic diamond it suck then ACS 5 it was ok. Mx 4 (1c under ACS5) is what i use now.

MX4 is horrible paste for low pressure mounts. It starts out great but goes south fast as it pumps out. You're better off with Gelid GC Extreme. If you didn't get decent temps with ICD btw you most likely didn't do it right, the only reason a lot of us don't use it is because of the serious scratches it leaves.