Alienware 18 BIOS A07

If the locking of multipliers is not just an honest mistake, then it seems that AW would rather neuter the CPU in order to work within the power handling limits, instead of fixing the root issue.

 

I am going to give benefit of doubt that it was a mistake. I have to believe it is, but it certainly seems that they are struggling to get a good BIOS released. We are getting close to a year since the product launch and little progress has been made. With the manual fan controls finally made available to everyone in A07, but the Extreme CPU functionality broken it is very disheartening.

But, I'm back in business again with A05 and broken fan controls. I'll take the CPU that runs right over manual fan controls. Hopefully, they will fix whatever they broke in A07 so we can have the best of both worlds in A08.





Quoting from another thread, since it's relevant here also...

 

Mr Fox, I agree that it showing ES should not matter, but I can't think of anything else that might explain why your 4930MX is stuck at 2GHZ on A07. Mine seems to work pretty well with the OCLVL . I am going to bench with it and see what happens. I also agree with you that this is unacceptable and something has to be done. It is not right to neuter a CPU because of design flaws that limit power. I wonder if anyone else with a 4930mx is experiencing the same 2GHZ you are.

I also have tried re-flashing with stock settings, and then reloading default settings two times, with no luck. The fact that lower end cpu's do not have these features grey'd out leads me to believe there is more to the story than we know. Maybe an oversight, or maybe an intentional lock out. Who knows, but I think for 4930MX owners A05 is the bios to run right now.

 

It was running 2.0GHz by default. I was able to change settings to force it to run faster, but it was still throttling under load like it used to with early BIOS versions. Before flashing back to A05 I noticed that if I changed a setting just one click with XTU or ThrottleStop the clock speeds would jump way up. So some kind of BIOS setting seems to have been getting stuck that making a minor change we get unstuck. For example, I could use XTU to change the multiplier on one core, or change base clock one tick mark, click apply and then the CPU ran at correct speed. It would throttle about 300-400MHz below what it was set to run when placed under load. I could change it back and the settings would stick until rebooting. After rebooting, then it was back to 2.0GHz again but all the XTU settings were being displayed the same as what I set before rebooting, make a minor change, click apply and change it back and click apply, LOL. Very strange, indeed. It appears a range/register is locked that should not be, and it appears an Extreme CPU is triggering the malfunction.

I didn't try uninstalling and reinstalling XTU to see if that would fix it, but I don't really care at this point what was causing it. Since A07 is not functioning properly for me the way it is, I will be sticking with A05 until they release something that works better.

 

Ok so my objective observations:

Do I get manual fan control over the GPUs in A07?
No. As for why Mr. Fox did and I don't, I am willing to bet a dime that it's due to him running Legacy boot, while I am on UEFI. I always have this hunch that manual fan control over the GPUs is messed up due to UEFI.

Is the multiplier locked in A07?
No. It's greyed out in the menu, with only OC LVL 1 available.

What is something that's locked in A07?
Short duration and long duration power, and I believe, the boost duration. We went from big nitro to small nitro. If "Overclocking feature" is left as disabled, you get 71W long TDP, 75W short TDP. If "Overclocking feature" is enabled, and you choose "OC LVL 1", you get 71W long TDP, 75W short TDP. And most of all, if you enable "Overclocking feature" and leave "Extreme core ratio bla bla bla" as disabled, you can 57W long TDP, and 72W short TDP.
I didn't bother to time the duration where the short TDP kicks in, it's less than 15s for sure.

What is something that's still unlocked in A07?
Multiplier is unlocked, core voltage, core voltage offset, core current, processor current, and presumably cache freq are still unlocked. Stuff are greyed out in BIOS, but like I said, I really see no point in messing with stuff in the BIOS now, since there's ThrottleStop. In hindsight, I could say I have some seen this coming, with Dell messing up more stuff when they introduce the 880M SLI boosting up to 993MHz. Short of a process shrink, you can't magically reduce the power consumption.
Something has gotta give. The base clock speed of the 880M SLI is suspiciously close to the max overclock that we can get on our 780M SLI, without overvolting, around 950MHz. That's about the max one can push before the power consumption spirals out of control.

Since ThrottleStop is working impeccably for me, I just stick with it. It only applies whatever overclock/undervolt after Windows has started up, and that also reduces the chance of getting into no POST.

Why would Dell do so, you ask?
Like I have mentioned before, the AW18 is running close to its power limit, esp with how aggressive Dell was at setting the multiplier and voltage on the CPU during the launch of the 18. They didn't do it the right way though, they gave the CPU too much voltage, too little current. It seems like the engineers still think that the Haswell CPU would function like the older CPU without integrated voltage regulator. They have learned a lot since then.

Then Nvidia released the 880M SLI, which because of it running at higher core clock than the 780M SLI, if Dell don't lock down the power limits, they will get a lot of complains about machine rebooting for no reason. It doesn't have to be black screen or blue screen, it's a hard reboot/reset. There'll be no log, the only logs you can see in the Event Viewer would be after the machine has booted up. So unless one's keen and rather observant, it'll be tough to diagnose.

 

As I was saying, no manual fan control for me in A07, so some unknown reason.



I went back to A05, because I want unlocked power limits. Below are some benchmarks of my CPU running at 4.2GHz, running at what seems to be 1.19V. Avg power dissipated through wPrime 1024M seems to be around 85W.

 

Great observations kh90123! With A07 I lost nearly 2K points in 3D mark11. I haven't tested the fan control, but I suspect I will not have it as I don't run legacy boot. I have a feeling I will be going back to A05 later tonight as well.

 

So is this something the rest of us with lesser CPU's should flash? I just have the base 4700mq in mine, should I bother with this or just stay with what I have which is currently A05? On a side note, despite the problems some of you are experiencing with the higher end systems/CPU's, I'm still jealous of the machines you are running. Best of luck to you and thanks in advance for any guidance.

 

NVIDIA GeForce GTX 780M video card benchmark result - Intel Core i7-4930MX,Alienware 01W2J2

My 3DMark11 Physics score is back to the same pathetic Haswell performance level as it was before. This is about 700 3DMark points lower than it is capable of since I ran it under Windows 8 instead of Windows 7. Something is still getting stuck after rebooting that did not before I flashed A07. Using ThrottleStop or XTU will break loose whatever it is, but it certainly seems like they locked something down that was not locked before. I'm just glad that ThrottleStop and/or XTU will correct it.



I think kh90123 could be right about the extra power draw from 880M driving some decisions, but I still do not accept that as a valid excuse. If that is the basis for it, they should not be selling a product with a flawed platform that does not function correctly. They need to fix it the right way before selling it.

 

So if I am running a 4930mx with 880m SLI should I use this bios version?

 

I believe the general consensus is that anyone with a 4930MX should not be running this bios. It appears to cripple the CPU to a point that greatly affects performance in a negative way. At this point I don't think there have been enough people with other CPU's to posts results to decide. The choice of course is yours to flash it or not, but those of us with a 4930mx who have tried this bios have experienced the same issues with the bios.

 

Well it doesn't really make sense for them to refresh the design again. It's a pigeonhole they have dug themselves into. Dell locked down the turbo boost power limits with the A07. Flashing back to A05 should revert everything, so nothing should be stuck.

I have mentioned time and time again, the first priority for them to do now, is to push Delta and Flextronic to make a higher capacity PSU. All these problems related to the AW18 underperforming stem from the fact that the 330W PSU can no longer adequately power today's system. Even during the time of M18x R2, one has to have dual PSU to push the system further.

 

Seems like the power handling limiation of the AW18 is finally starting to show its true ugliness. PSU limitation wouldn't be as much of a problem if one could utilize dual PSUs. There's no reason that a newer AW18 can't handle dual PSUs; the refresh would've been a perfect opportunity for AW to address this issue, but it seems they have taken the easy route of simply gimping CPU performance.

Anyway, let's just hope that this is truly an honest mistake, and they'll make things right with A08.

 

It's big nitro to small nitro now.

I really doubt the 880M will be pulling that much more power. And I think that the 4940MX should be a tad more efficient than the 4930MX, due to later stepping.

Too bad, I don't have a system with 4940MX and 880M SLI running A05 Bios. I'm pretty sure there's a way to push it to be as fast as the 4930MX + 780M SLI, if not faster, without tripping the internal circuit breaker or the PSU circuit breaker.

I was hoping that AW wouldn't be gimped all the time. So in the future I will build a microATX desktop. Can't wait to put the Haswell E 6/8 cores CPU and the AMD's 295x2 into a tiny case, like the Bitfenix Prodigy M. I need the system to be very powerful, but small enough to fit into a luggage.

Edit:
Time and time again Dell has shown us that the changes in BIOS are not mistakes, they always seem to be changes that are intended. One can feel that the engineers in Dell are always playing on the very safe side, they want to get their hands clean.

A05 might be a result of direct push from users like us, but it was followed up very quickly with the A06 and A07.

 

So I guess I should stay on A05 even with the base 4700? I wish I would have read more that the new AW laptops sucked compared to last years model I would have held off on buying on.

 

confirmed that MX cpu's are greyed out. my system has 4940MX with 880M and just flashed a minute ago from A05 TO A07...............

 

Out of curiosity, would removing one video card free up enough power to overclock the CPU as it should be? I am wondering if these overclock options are removed because the "system" knows how much available power is present. Not that I think we should have to run a single card to get the most out of our laptops, but it would be a neat to perform and see the results to.

 

Yes it should, I have mentioned this somewhere in the AW 17 thread a long while ago. I found out that the default (stock) turbo power limits are much higher on the 17 than on the 18.

The system doesn't know how much available power it has, this is not a Porsche 918 or McLaren P1. It's far more less sophisticated than that. There are analog chips, most likely current sense chips that were put onboard, and if you hit the cap, then the system will shut off. The system can tell you how much power it's using at one time, you can see all the info in HWiNFO64, you can estimate the total power (rough estimation) by summing up the CPU and GPU power, and by adding the RAM power, the display (typically 10-30W), the disk drives power (depending on whether you have SSD or HDD), the mainboard power (should be no more than 20W).

The 17 uses 240W PSU. Say the 780M/880M has a max TDP of 120W. That leaves 120W for everything else. You can have the CPU pulling 100W and it should be fine with it.

The 18 uses 330W PSU. Say the 780M/880M SLI has a total max TDP of 240W. That leaves 90W for everything else. Simple estimation shows that the 18 will run out of power far sooner than the 17.

Mr. Fox has shown that, by disabling the 2nd GPU, he can push his 780M to almost or above 1.2GHz on the core clock. The 780M GPU core is basically a GTX 680, except for the fact that the VRM on a mobile GPU, being 2 phase for the core (2+1, 2 phases for core, 1 phase for vRAM), is simply not robust enough for long term overclocking. Even if it's, the heat will get to it first.

 

They will still need to fix whatever defect or engineered limitation is present with the 18 motherboard. Simply adding a higher capacity PSU won't help because the system powers off before it maxes out the dual PSU capacity. It shuts down at the same point as it does using a single 330W PSU, in some cases without ever tripping the breaker on the 330W AC adapter. So, whether it makes sense to do a refresh is open to debate. If not a refresh, they need to at least release a revised motherboard service replacement part for the current model that can handle the power that it takes to operate correctly with a 4930MX and 780M SLI or 880M SLI without emasculating those components. They would not even need their vendor partners to produce a higher capacity AC adapter... they could offer an adapter/junction box like Clevo does and sell a dual AC adapter setup as an added-value option for those willing to pay extra for it. But, they do need to address whatever is wrong with the current motherboard before it could take advantage of that option. It may be something that can be addressed through firmware if they have implemented a software power "governor" of some sort. Dufus already showed us where the firmware in A03/A04 had a "feature" that detected the presence of GTX 780M SLI and dropped the Processor Current Limit below the correct default minimum in the Alienware 18. Shortly following that revelation, although likely unrelated to it, Alienware fixed that problem in A05.

You don't even need to remove it. Disable SLI and uncheck the box for EVGA Precision X to sync the GPU clocks and you can overclock the snot out of the 18 CPU and one GPU without it shutting down. With SLI disabled the second GPU will sit there in a low power state and do nothing. As soon as you enable SLI and sync the cards the show is over. I can overclock one of the 780M GPUs in the Alienware 18 (or the M18xR2) well beyond what I can overclock two of them in SLI, even in the M18xR2 when using the dual AC adapter.

If my understanding is correct, the 330W AC adapter works perfectly with the Alienware 17 and the Alienware 14, so owners of those models need only to buy an M18xR1/R2/18 330W AC adapter to have more power than their systems are capable of drawing.

Thank you very much for confirming this for us, bro. :thumbsup: Much appreciated.

 

Well,it makes me feel bothered about the overclockers using this machine who are having to deal with this bitter (and expensive) letdown,and for me too,thinking of the shrunken top perfomances in the long run. Maybe with the double adapter I would have delighted my hunger for the best performance on new titles for some more time, supporting future cpu and gpu that still has to come and use em to their fullest potential, a bit like someone already does with the M18

However,I don't get the decision of Dell engineers over this matter. I wonder what they thinks about who buys a power machine like this,and invests so much in it,instead of getting a budget laptop and a lesser expensive assembled desktop.

I think a gaming computer costumer isn't much different from a supercar costumer. Any computer can run a game as any car can brings you from point A to point B,who buys a more expensive car do it to enjoy what lies between those points,as well as those who get an expensive laptop wants to get the best experience out of it. Both probably do it for passion. So why do limit instead of meeting the customers needs?

Having reached this power limitation might be the best time to argue about it and about an optional dual PSU with Dell on their forum,and make things clear. It's not something that really touches me as not being an hardcore overclocker and benchmarks enjoyer,but if you guys will choose to do so I'll give you all my support

 

Well my question is more or less wondering if with A07 and a single 780 will our overclocking options show back up. Much like what you mentioned brother Dufus found with A03 and A04, I am curious if there is something in the bios that limits the options when the system seems the Xm CPU and dual 780's

 

Well I have Dual GeForce GTX 770M and the options aren't disabled on A07, maybe coz I have the MQ4800 CPU

 

Yeah the current sense chip built onto the mainboard. Well it could be that they found out about the power problem (due to the limitation of the PSU) then they decided to set a clamp.

It does I think, somewhere sometime ago, I read that the BIOS has a bit in the hex code, that shows whether a extreme CPU is installed. I think it's svl7 (either him or Dufus) that found out about it. Since Mr. Double Helix said the options are greyed out for him, it's likely what you said is true.

 

Is it something you can change or turn around with just a firmware update of the mobo? a motherboard replacement service looks too odd to see Dell let it happen

 

Ok, I just updated to A07 and my multipliers are greyed out (using 4900MQ with 880M SLI)

Just reflashed A05

Sent from my SM-N9005 using Tapatalk

 

Just a quick question, back on A05, I set the following multipliers on my 4900MQ (42, 41, 40, 40) and it seems to be holding so far. Will that be ok, considering that the TDP is only 47W ? Wouldn't it decrease the SLI GPU performances as well considering there is only a 330W AC input ? If so it might actually better for me to get back to the OC LV1 values.

What do you think ?

 

You should be fine with 47W draw. With only 4.0GHz on 4 cores, your 330W AC adapter can handle that just fine.

The 4930MX draws as much as 130W with a heavy overclock, which is not OK with a 330W AC adapter and overclocked 780M/880M SLI. If you are not overclocking the video cards that's still OK. I can run the 4930MX at 4.5-4.8GHz with the video cards at stock clocks and no overvoltage and it still works fine. It is not until you try getting fancy with both CPU and GPU overclocking when the system exhausts the available power resources. For this reason I do not expect it to be fixed because Alienware only guarantees stock performance. It also raises questions about the theory that Alienware is crippling the BIOS for the system to handle these high power components because it can handle them fine with all of the components running at stock clock speeds.

I think they are crippling the BIOS because it makes their "partners" happy (think compliance with the edicts of the Micro$haft empire or doing whatever Intel tells them to do), the BIOS engineering model is the same as what they do with ordinary mainstream systems and requires no unique process for Alienware systems. If people keep buying the product without these enthusiast-centric concerns being addressed, there is no incentive for them to ratchet things up and design something robust enough to handle extreme overclocking. And, the deal with the fans is probably because they have listened too intently to whiners that complained about fan noise or they have a "spec" for a certain decibel threshold for all Dell systems that cannot be met if they don't slow them down. That doesn't explain why they locked everyone out from manual fan controls to the exclusion of those that don't care about the fans being loud, but all of this is just speculation and logical assumptions. Unless they provide us with a detailed explanation of why things are the way they are, all we can do is assume.

All of this seems like doom and gloom, but there are similar scenarios that play out with their competitors. How many "gaming laptops" made by Alienware competitors are outstanding? None... nada... zero. It's conceptually no different than NVIDIA GPUs not performing at their maximum potential because they don't have to in order to be equal to or better than AMD. So, they do the bare minimum required to get by instead of releasing something insanely phenomenal and leave it up to enthusiasts to mod and make things excellent.

 

Thanks for such a great reply, that makes sense.

About CPU overclocking it seems the 4700MQ keeps going between 3.7Ghz idle and 4.1-4.2Ghz under load, is there any way to increase the idle frequency with XTU ? Is it safe ?

I've got another (unrelated) question. Sometimes (like 1/3 of the time) when I cold boot my Alienware 18 it gets to the Alienware + Klipsch logo and then turns off on its own and I need to press the power button again for it to power on up to Windows. I assume that's not standard behaviour but I could be wrong. Then again it's not enough of a hassle for me to return it so I can live with it. Did this ever happen to you or did you hear about that issue before ?

Thanks again for your replies !

 

You're welcome.

That erratic behavior may be attributable to the CPU power settings not being optimal. I don't have access to a 4700MQ to test it, but your ability to change power limits on a non-Extreme CPU are very limited. TDP is locked. Haswell behaves goofy when it is working correctly, and even more goofy when something needs to be optimized. You will need to experiment with whatever power settings are available in XTU. Make small changes up or down to see how it affects the behavior. When you get it where you want it to be, use those settings.

Not sure about the turning off during boot issue. I have not seen that and it sounds like a hardware malfunction of some kind. Unless you have already been messing with the CPU power settings, which might cause that if they are way out of whack, you should not be seeing this behavior. I would not recommend living with it that way. It might get worse later. If you have a 3 or 4 year extended warranty you'll be OK, but if you have a 1 or 2 year warranty I would recommend getting that problem diagnosed and corrected right away, before the warranty runs out.

 

Hard reboot is typically a sign of the CPU core voltage being too low. Recently I have been getting hard reboot, nope, it doesn't give me 124 BSOD like it used to be.

I'd suggest sticking to the same multiplier across all cores, like 41x 41x 41x 41x, if the CPU allows it. If not go to 40x 40x 40x 40x, if not go lower to 39x 39x 39x 39x. Setting single core at too high of a freq will require a much higher core voltage, and that in terms generate more heat when all 4 cores are stressed. The integrated VRM in Haswell CPU is smart, it adjust the core voltage in a dynamic and adaptive way, but it's not smart enough yet when it comes to the offset voltage.

There's a lot of info over at the TS thread: http://forum.notebookreview.com/har...31329-throttlestop-guide-196.html#post9602771

Read from abt page 196 onwards to the last page, and you will get a better idea.

Idle freq will go up and down, due to Intel EIST and C states. If you want the CPU to be running at max freq all the time, set the Power Option to Performance. Or you can go in there and adjust the Profile settings. Set the minimum processor state to 100% and your CPU will be running at max freq all the time. Latest Intel CPUs are quite smart, even though the CPU could be seen running at high freq, the transistors are gated, so the power consumption shifts based on load.

In short, you should run ThrottleStop, and start giving the CPU a bit of offset core voltage. Or you can do it in XTU. I prefer TS because XTU is wonky. Alternatively, you can lower it the CPU to something like 39x 39x 39x 39x, I am pretty sure once you do that the problem will go away. If it persists, then it could be something else.

 

The issue I have isn't hard reboot, it's power off at coldboot (right before the UEFI starts booting off the harddrive) and it does not always happen.

Another issue I have is with the USB, some devices will not detect properly (an hard drive and the xbox wireless pad receiver) and I have to reconnect them multiple times before they get recognized, which is rather annoying (I might contact dell about this)

 

I agree... ThrottleStop is awesome. I use XTU mainly for when I want settings committed to the BIOS. With the 18's locked down BIOS, XTU is the only way to do that for some settings.

For all other tuning I use ThrottleStop pretty much all the time. It runs at Windows startup on both of my beasts, so it is 24/7. XTU is launched only when I need to.

I think a few people have reported issues with the USB ports malfunctioning. I have seen examples of where the USB ports shut down completely and become non-functional as soon as you switch to battery power even with the option to turn off to save power disabled in Device Manager. Some of the ports output less power than USB 3.0 maximum spec, so some of those USB peripherals might be glitchy depending on which ports they are connected to. I have also seen Bluetooth radios that immediately stop functioning on battery power, even with that option disabled in Device Manager. I don't know what some do this and others do not, but it has nothing to do with software. Reinstalling Windows changes nothing.

 

Try out what I suggested and see if it fixes the problem.

 

I have set all the multipliers to 4.1Ghz, I'll see if anything changes with the coldboot issues. I hope the cpu won't get too hot from running all 4 cores at this speed.

 

You really don't want to leave too many things to chance, test out everything. Run Cinebench R11.5, run wPrime 1024M, run the 3Dmark11. The most brutal stress test of all, is the Intel Linpack. Under normal circumstances though, it's very hard to stress the CPU to the extend Linpack does. I won't really run Linpack on a laptop, but on a desktop the AVX2 Linpack is the hallmark of stability test for desktop Intel Haswell CPU.

What core voltage do you have for 4.1GHz?

Default Dell OC settings sucks.

 

Anyone notice that yesterday A07 read 10MB on Dell's support site, and today it says it is 11MB? I wonder if something changed?

EDIT: Nevermind, I just downloaded it again and it reports the exact same size... Well there goes wishful thinking

 

The file is 10.7 MB.

 

yea I noticed the file is the exact same size, there site must of just changed the listed size.. oh well it seemed like a hope and a glimmer

edit: In other news I just heard from a dell tech that the removal of the overclock options are intentional, and they are working on a new release that should have them back. I would of course that that with a grain of salt, but I figured I would share.

 

I actually kept the default voltage settings, not too sure what to set yet considering it's my first time doing CPU overclocking on a laptop.

I already ran 3D Mark Advanced, worked fine so far, I haven't tried the other benchmarks yet, according to HWiNFO sensors the maximum temperature I get off a CPU core is 84°C, isn't that a bit high ?

Ok, I just tested cinebench, cores went up to 90°C and the multipliers hit back to x37 so when the CPU is too hot it'll just set it back to a (much) lower clockspeed. Either that or the power isn't high enough.

 

The default voltage settings are generally too high. Even the 4930MX (at least the one I have) can run 43x4 with only 1.175V and the default is more than that.

Your CPU may also be load throttling, or as I like to call it "power throttling" (as in lack of power). Haswell has a nasty habit of doing that even when it is not too hot. You might need to increase Core Current Limit and Processor Current Limit (if those are not locked with your CPU) to try to circumvent that silly behavior. Increasing these values generally does not cause overheating like excessive voltage does. Running at a higher sustained clock speed can increase temps, but the increased power limits do not cause it apart from allowing the CPU to sustain the overclock.

 

For 4.1GHz, to get the max score you can get at that clock speed from Cinebench, you will need about 80-85W of power for the CPU. You have to check the short and long duration turbo boost power.

Unfortunately, overclocking isn't really a plug and play experience, it's more like tuning. It's like strapping turbochargers to an engine, can't really just strap turbos on and expect everything to be peachy. You have to fiddle with the ECU, the fuel injection, intake manifold, etc etc.

 

Hehe mine can't do 43x with only 1.175V. I think it's related to the ambient temp, or the intake air temperature for the CPU fan. I actually need 1.19V to be perfectly stable at 42x. It could pass Cinebench or wPrime fine, but the system could have a hard reboot when I open up 150 tabs in Chrome, or it could crash after hours of playing D3.

Since there's another gentleman (Mr. Scracy if I am not mistaken) here that beat everyone's score in XTU benchmark with a 4900MQ, I think it's safe to say that running on A05 & A04, the 4900MQ has everything needed unlocked, like the 4930MX. The only thing that are locked are the maximum multiplier, for the cores and I presume, for the ring cache.

 

Yes, everything is unlocked with the exception of the max multipliers and those are already set too high anyway. I just don't know what values to set, I am afraid of getting into a NO POST situation since there is no way (to my knowledge) to reset these values if anything goes wrong (which is insanely stupid for a gaming notebook), as long as I can't recover from a wrong CPU setting, I won't be pushing my luck too much with these.

 

It's good to qualify suggestions with statements such as "if those are not locked with your CPU" merely for the benefit of readers that do have a locked CPU. If I am not mistaken, the 4700MQ and 4800MQ have some settings locked down that the mostly unlocked 4900MQ does not. As those running A07 just discovered, a fully unlocked Extreme CPU may even have functional impairments introduced by the BIOS. No telling how flaked out the 4900MQ might have become with A07.

kh90123 figured this out a long time ago. You can recover. I used his method yesterday. The process is outlined earlier in this thread. I had to do it because A07 bricked my motherboard when I tried to apply a known good XTU profile with A07. I am running that same profile right now with A05, LOL.

 

The 4900MQ has the same limitations as the Xtreme CPUs on A07 while 4800MQ and 4700MQ do not seem to have their multiplier locked (although 4700MQ's max multiplier are rather low to begin with)

Are the Xtreme non turbo boost max multipliers unlocked ? That's one I'd love to set higher since x28 on the 4900MQ is still a bit low to my tastes.
I guess we can somewhat set it so it sticks to turbo boost at all times anyway (though I haven't quite figured out how yet, so far my cpu won't go any lower than x37 unless under extreme duress (like when I try to run linpack on all cores set to x41/x42, then it goes back to x28 for the most part.

What's the Processor cache ratio (set at x38 here) ? Any use changing that ?

 

Too bad they arent like car manufacturers and have to recall faulty products that do not function as they should be.

 

I have read the whole process, it seems like too much of a hassle to me xD I rather avoid to "brick" it if it's possible.
If only they added a reset process in their damn bios...

 

It's funny I was thinking the exact same thing last night

 

So one has to insert the RAM before the machine reboots again to fix the no POST. Hmm I have not had no POST for some time now, hopefully I don't get to test it out soon. Maybe that's the trick to get it to work everytime.

I am not sure why, but in my case Samsung Rapid seems to be related, a lot of times when I messed with the settings with Rapid enabled, no POST happened. Another time it was related to the conflict between BIOS, TS and XTU. A long long while ago I recommended to Unclewebb (the developer of TS) to include Haswell core voltage and offset voltage control into TS, to have separate voltages for each profile, so I can run each profile which has different clock speed, with varying amount of undervolt. For eg, 1.19 V for 4.2GHz, 1.10V for 3.9GHz, 0.95V for 3.4GHz, and 0.85V for 3.0GHz. One can push the voltage real low at lower clock speed, that's where Haswell shows its efficiency. Since Unclewebb incorporated these changed, I have stopped dealing with the buggy XTU altogether. Unclewebb has told me, he observed that there were times where XTU wouldn't write the correct stuff into the registers, and that's why there are some who complained that settings don't stick after changing them in XTU.

Non turbo boost multiplier is unlocked on the 4930MX. Leave the processor cache ratio alone, it's too much hassle for almost no improvement. It's what used to be the L3 cache freq. Since Haswell, Intel has decoupled the L3 cache, it's like a ring cache now, running its own clock.

My CPU runs at about 3.5GHz in Linpack, as is Dufus' 4700MQ.

Intel recommends the same max processor current limit for both 4900MQ and 4930MX.

 

...

Wow, maybe I should cancel my order. I had the DPC latency and stuttering issues with my M17x and those were eventually (mostly) fixed but it sounds like Dell is up to their crap again.

I ordered the 4810MQ/R9 M290X... I'm assuming I'll have issues as well? I don't know about the R9 M290X but in general AMD cards have a habit of huge power draw and high heat...

I really want a new machine but I don't want a repeat of the M17x. I just ordered it today. Is this something that just affects the 4930/4940MX and the 4900MQ?

Either way... Typical Dell bs.