Robbo

It's not worth upgrading for games, 3840QM is good. If you're aiming for 60fps then you'll be fine - if you're aiming for 120fps then get the better CPU.

That's a good upgrade, I'd probably be buying that 970M with some serious overclocking for my system, but alas it is not compatible!

Cool, that sounds really interesting, do you clock down the CPU when the GPU is clocked up because of the limitations of the 180W power brick, or is there some kind of block perhaps (like a limit in max Watts allowable) in the system regardless of which Power Brick is used (180W/240W/330W)? EDIT: just realised I misread your post, you're saying it needs BIOS modding to prevent the downclocking of the CPU when GPU is overclocked. Does it even need that when you're using a 240W or 330W adapter?

You've probably already considered this, but could you just run at 1080p on the TV in order to get 60Hz in the mean time?

I think it will be fine, will be a good upgrade!

Hi, I'd choose a 780M out of all of those. 980M isn't compatible, won't work at all. 880M may work, but don't really recall any users having done it on the R3. The 780M will work, and with a modified vBIOS from svl7 will outperform even an overclocked 880M - the 880M's were just buggy & throttled loads when overclocked. 780M is your best bet! Either that or buy a new laptop with 980M, there's no other card upgrade really worth it over a heavily overclocked 780M. (Before you buy the 780M just check this thread or others that there are no issues with 780M and 120Hz display in the R3, I don't remember there being an issue, but I'd check if I was you.).

It looks like there's only the 6GB version of the 970M to download. On the first page of the thread svl7 said that he would mod a 3GB version if he had time - looks like he didn't do it. Prema has been doing vBIOS mods of the new Maxwell cards, and my understanding is that they will work on Alienwares too - see if you can find his threads on Tech Inferno & confirm that they're compatible with your Alienware.

I think 92 degC would be more sensible, because temperature will continue to rise slightly for a short time after 80% fans come on at 89 degC, so by putting a throttle point at 92 degC would mean that you're giving the fans a chance to bring down the temperature before throttling occurs. If you set it at 90degC you might quite likely have a short period of throttling before the increased fan speed could take effect.

Oh, ok, thanks. Why did you edit my post though, I included a list of a number of different graphics cards to illustrate that this driver natively supports many different types of cards, but you just deleted the entire list (which wasn't a stonking great long list either) & left the 850M/860M - I don't really like people editing my posts, I like to think users should be the ones that edit their own​ post unless there is some kind of infarction - I don't like you editing my posts (for trivial matters).

This driver installs on my laptop without any modified inf required - there's loads of cards listed generically in the stock nv_whql.inf file - for some people there's no need to use j95's or other people's modified inf for installing this driver (even on non-standard systems like mine - the first driver I've ever seen that has such generic support). No performance differences seen for me with this driver vs the latest notebook drivers - same synthetic benchmark scores & game benchmark scores. I downloaded the unmodified version of this driver from laptopvideo2go (didn't even need to use their modded inf either): LaptopVideo2Go: Drivers Here's just an example of some of the strings that this driver has generic support for (built into the stock & unmodified version of this driver), pretty much all the new Maxwell cards too: %NVIDIA_DEV.11A1% = Section048, PCI\VEN_10DE&DEV_11A1 %NVIDIA_DEV.13C0% = Section154, PCI\VEN_10DE&DEV_13C0 %NVIDIA_DEV.13C2% = Section154, PCI\VEN_10DE&DEV_13C2 %NVIDIA_DEV.13D7% = Section160, PCI\VEN_10DE&DEV_13D7 %NVIDIA_DEV.13D8% = Section160, PCI\VEN_10DE&DEV_13D8 %NVIDIA_DEV.13D9% = Section160, PCI\VEN_10DE&DEV_13D9 %NVIDIA_DEV.17C2% = Section154, PCI\VEN_10DE&DEV_17C2 %NVIDIA_DEV.17F0% = Section166, PCI\VEN_10DE&DEV_17F0 NVIDIA_DEV.11A1 = "NVIDIA GeForce GTX 670MX" NVIDIA_DEV.13C0 = "NVIDIA GeForce GTX 980" NVIDIA_DEV.13C2 = "NVIDIA GeForce GTX 970" NVIDIA_DEV.13D7 = "NVIDIA GeForce GTX 980M" NVIDIA_DEV.13D8 = "NVIDIA GeForce GTX 970M" NVIDIA_DEV.13D9 = "NVIDIA GeForce GTX 965M" NVIDIA_DEV.17C2 = "NVIDIA GeForce GTX TITAN X" NVIDIA_DEV.17F0 = "NVIDIA Quadro M6000"

Sounds like it's probably a hardware problem if your white & blue screens are happening at different points of the boot process. I'd try to ensure that the card is padded up level (so it's not bending), take the card out and blow out the PCIe slot with a can of compressed air, clean the gold contacts of the card, replace the card into slot (mine snaps into place). Then make sure you're following j95's instructions on Page 23 Post#222.

Hi, so it sounds that you're just struggling to get the NVidia driver installed. I don't think you needed to uninstall sound drivers as part of your process, it just had to be disabled temporarily during your install process, as described by j95 on the page 23 that you quoted (post #222), but that's probably not causing your problems. Try using a program called DDU to remove all traces of NVidia drivers (you didn't need to uninstall the Intel Graphics Drivers either I don't think), then try reinstalling j95's driver or a driver using a modified inf from laptopvideo2go (LaptopVideo2Go: Drivers), or mod the inf file yourself to do the NVidia driver install (http://forum.techinferno.com/nvidia/5249-%5Bguide%5D-modding-nvidia-oem-inf-files.html).

Well, you need to make sure the pads are padded up properly to the layout of the chips on the new card - the Kepler chips cards differ from the Fermi cards, and the Maxwell cards then vary from the Kepler cards - you have to check pad placement. Also, if you're changing between AMD & NVidia you need to get the X-bracket that matches each manufacturer, and you should remove any black tape on the heatsink GPU core contact plate because die sizes are different & black plastic tape is not a very good heat conductor!

Yes, I think you can flash that one - the one you're referring to is for the Clevo 680M 4GB card, which seems to the be the card you have, (Dell didn't do a 4GB 680M so yours must be Clevo). I think you'll be fine flashing the one you've chosen.

Haha, yes, I guess this is one occasion where you can't believe everything you read in Wikipedia!

There aren't any ARM chips on any of the Maxwell cards, whether they be mobile or desktop cards. Initially, Nvidia was going to include ARM CPUs on the Maxwell GPUs, but that never came to pass. Maxwell GPUs aren't compatible with the M17xR3 according to forum member 'Reborn' from notebookreview forums - he tried, they don't get past the post screen.

You could request svl7 to produce a higher voltage version of the vBIOS at his vBIOS request thread, but I wouldn't hold out too much hope of it happening, but doesn't hurt to ask: http://forum.techinferno.com/general-notebook-discussions/4635-bios-vbios-modification-request-thread-svl7.html (I think I heard you saying that your overclocked card was nearly as fast as a stock 980M, it's way slower than a stock 980M, mine with a core clock of 1124Mhz is even way slower than a 980M. Those 980M's are very impressive cards. All benchmark scores for you to check at notebookcheck.net).

I assume you're using the K4000M vBIOS on the first page of this thread. I'm not familiar with that particular vBIOS, I don't know if it's voltage adjustable. If it's voltage adjustable you can increase the voltage of the core using NVidia Inspector, which will enable you to get a higher stable core overclock, and given your very low GPU temperatures under 60 degC, then I think it's safe to increase the voltage & the core speed Mhz. If it's not voltage adjustable, then you're just stuck with the highest stable overclock you've currently been able to achieve. 933Mhz (or whatever exact overclock you mentioned) is a great overclock at stock voltage, so that's good anyway.

You have very old information. It's completely unnecessary to arbitarily set your max temperature of your GPU to 60 degC, it's completely fine up to say 85 degC. Won't shorten the life of the card, unless you want to keep it for 10 years or something. Your Quaddro is based on the GTX 675MX, which can hit insanely high overclocks compared to stock for the reason you mentioned (these particular cards are underclocked by NVidia excessively). I have the 670MX which is essentially the same card, and it also overclocks to ridiculous levels for the same reason. Your Quaddro isn't clocked that high, but it's a good clock on stock voltage, I have 1124Mhz at 1.05V though.

The way that guy is writing (the user you quoted named 'Masked'), how it's all constructed, I wouldn't believe anything he says.

I don't have the memory voltage reported in GPUz, it just reports the core voltage for me. I can't get any higher than the 110GB/s and 1150Mhz VRAM frequency. I should think it's safe for you to increase the VRAM frequencies if your card is padded up properly, say if you increased it to 1050Mhz on the VRAM you'd end up with 134GB/s - if I had your GPU I'd up the VRAM frequency - some games will give you a performance boost from a greater memory bandwidth, I'm sometimes limited by the 110GB/s. 69 degC max core temperature in the Summer. Normal temperatures 66 degC max.

Haha, that's terrible, especially in a beast of a laptop like The Batman! Sacrilege, especially when that laptop is all about enthusiast performance!

Is this applicable to any specific vendors, or in cards just bought separately off the internet?

I use Arctic Silver 5 on my M17xR3. I do find that temperatures tend to rise after 6 - 12 months, but only by less than 5 degC, so I just re-paste every 6-12 months. I've used the spread method (as recommended on Arctic website), the line method I've tried on the CPU, as well as the central blob method. I've not really noticed any difference in temperatures between the different methods, although for the CPU I've noticed slightly better results with an ultra ultra thin spread method of the paste. I use it on both my CPU & GPU. (I use cans of compressed air to blow out my heatsinks, so that's not a factor). I've not noticed any pumping out, paste is always well distributed where it should be, just noticed that very slight less than 5 degC rise of temperatures over that 12 month period. No motivation to try other pastes as GPU temps are 69degC max and less than 80 degC on CPU during Prime95 at a constant 50-56W CPU power.

Yes, those 675MX and 670MX have a load of untapped power! I have the 670MX and it's the same chip as your 675MX, except you have a 256bit memory bus in comparison to my 192bit bus. I use 1124Mhz on the core at 1.05V, and run 1150 Mhz on the VRAM to give me 110GB/s memory bandwidth. There are some games that would benefit from a greater bandwidth than my 110GB/s, but I can't push that figure any higher - if I was you I'd make sure your memory bandwidth (seen in GPUz) is at least 110GB/s, so overclock VRAM to make sure you're at least at that figure. (So much untapped performance on the 670MX and 675MX because the stock clocks are so low at only 600Mhz!)