Over the last 10 years I've found laptops throttle the processor speed in order to keep fan noise down. I prefer the fan to run noisily, and the processor to run at full speed. I've developed tried and true mods for lappies and have demonstrated their effectiveness.
I have two Lenovo W541 laptop workstations with I7-4940MX processors and discrete nVidia graphics. On one I overhauled the cooling. Copper shim between CPU and hsf, AS5 on both sides. Shims between hsf and inductors near CPU and gpu, and shim on the gpu.
Used 'thinkpad_acpi' Linux kernel module to set fan to max speed. Ran mprime (prime95 for Linux) AND Unigine Heaven, alternating, to cure the AS5. Then ran mprime. Temps were slightly high. But laptop processors tend to run hot.
The I7-4940MX is a 3.1 GHz, 4.0GHz. Turbo processor. The modded machine ran at 3GHz. continuously running mprime, for weeks. The unmodded machine with the same max-fan-speed software tweak and stock thermal-interface material, throttled to 1.7GHz. at the same running temps.
Clearly, laptop manufacturers are robbing performance off the top, by inferior cooling hardware and materials. Granted, getting the cooling mod to all fit together correctly with smooth, flat surfaces requires hours of work and materials not normally found sitting around the house.
I normally do not mod the CPU die. I abrade copper surfaces down to 1500 grit paper, and then polish with scapings from a bar of buffing compound. Everything is done on 1/2" plate glass to maintain flatness. The final mirror polish is done with the compound granules on a clean, damp cloth.
Desktop processors mostly have slightly raised corners, so smoothness is not as much of concern. You're going to get thermal-interface compond completely covering the center of the processor. The ideal is to fill the microscopic gaps between the cpu die and the hsf, but where there is metal on metal, have it be.
Metal to metal provides 100s to 1,000s times the heat transfer of thermal paste. But, on a desktop, there are more effective cooling solutions than moving air. Most user are satisfied with their laptops, because the processor doesn't run full speed long enough to make a difference.
But I enjoy tweaking, and I do processor-intensive work, like batch editing 100s of large images, video rendering, security testing, password recovery, network testing; that are all severely compromised by poor, stock laptop cooling solutions.
Well I am back, sort of.
Reminder of Stats
*Alienware M17x R4 (bought June 2012) w/ Intel i7 3720QM@2.6ghz (Ivy Bridge) including the igpu Intel 4000 HD series
8GB Ram, 17 Inch, SoundBlaster software only Recon 3DI, and the normal stuff that comes with it
*Graphics Card: Originally shipped 660m, upgraded to 7970m, 7970m died early, with community help moved to MSI 4GB 680M (flashed to Dell thanks to @svl7) and installed I think 314.xx drivers with help from many (sorry can't remember everyone but notable was @J95), was running 7970m in whatever they called their hybrid drivers, the 680m is run in Optimus)
*OS: Windows 7 64bit Service Pack 1
*Storage: Msata 1tb Samsung 850 Evo, 2 Hitachi 1TB 7200 rpm 2.5" drives (yes I used all 3 GB or will)
So what will you read. What got me started on this project and then using all the resources on the net, from @J95 and legions of others I owe all credit for the project too. Please note I take no credit, take no rights, this is simply something to help others out based on working with what has already been laid down for us. Installing the latest Nvidia 417.67 mobile/notebook drivers for a 680m for Windows 7 Service Pack 1 on Alienware M17x R4
When I got things up and running back in 2016/2017 with the noted people above I saw a new driver and with the STO game having a driver issue I upgraded to it. Now I didn't notice anything wrong, the game worked and so paid no attention. When I started to get unable to eject 680m error messages I thought great who got this, ejecting a gpu, wrong. I then tried the uninstall Intel, Nvidia, use DDU in Safe Mode, disable AV/Firewall, disable driver enforcement, etc then reinstall intel, nvidia (removing certain folders) with the modified inf but now the game said it couldn't find any feature set and wouldn't let me play. Finally removing the Nvidia drivers allowed at least the iGPU to run the game. Used Unigine Heaven 4.0 to encounter more errors with Nvidia.
I moved back to my msata install of windows 7 that I haven't had a chance to work on since 2017 (i know pathetic) that only had system drivers, expired Norton, and not yet registered/activation. Seems fine. Tried to run Heaven, same error. Well after 3 days of trying to duplicate what I had been instructed to do, I finally sat down with a nice big bottle of Fiji Water (courtesy of lovely lady at the Oscars in almost every background pic) and decided I was determined to fix things. I downloaded the 419.67, copied the nv_dispi.inf & nvdmi.inf to get to work. After reading the content and using comparisons once again from the noted individuals and the legions of other, I had them modified and decided to give it a try on the clean win 7 install from 2 years ago. After the uninstall, ddu, deleting folders, copying mod. inf to their directory (also included the change to setup.cfg - credit once again goes to individuals like @J95), gyrxiur, DeathAngel , creator of Notepad++, sora and many others) I was able to launch the Heaven 4.0 Benchmark. So I am assuming that I have the new 419.67 working on my system. I did a benchmark of the 680m in 419.67 with unigine Heaven 4 to go with the files I am posting. Please note the bios was provided by I believe @svl7 (sorry if I got wrong person) and no overclocking of it other than what the changed vbios did.
I am attaching below the following:
Nvidia 419.67 modified inf and config - nvdmi.inf, nv_dispi.inf, setup.cfg (no credit to me all credit belongs to others)
The Unigine benchmark results (txt file).
Now the shortened legalize warning
Please remember you are to take all risk when using modified inf files, I take no responsibility (no virus or other dildrus included) if they don't work with your GPU. The usual.
This was tested and used on Alienware M17x R4 running the last bios they released A12 or A13 and so it the system bios is not modified at all, the card is a MSI 680m 4Gb flashed vbios to Dell 680m 4GB, and installed in Windows 7 64bit Service Pack 1. Hopefully if you have only a different card you can make any minor adjustments you need to.
Please let me know how they work. I will also be reporting back on the reinstall on my other laptop install I use to play Neverwinter/ Star Trek Online - was suppose to be a temp solution till new install was finished, and if they worked on it as well.
Alienware M17x R4 680m 419.67 Mod Inf and Setup and Heaven 4 Benchmark.zip
I picked up a Dell 680m - vBIOS 80.04.5B.00.02 - I think it is a 2GB version after doing some photo comparisons online.
After doing some research I fitted it into my Clevo P150HM hoping it would work, but and I get no POST, the LCD screen is blank, the HDD light does show activity but after ~ 15 seconds or so, 3 blinking lights appear on the left (power light, caps and insert buttons?) and the computer beeps about 4 times before it completely powers off.
I'm guessing that I'm going to need to flash it to a clevo bios since the clevo bios doesn't recognise the dell vBIOS?
I am running the latest 1.01.14 bios and 1.00.13 EC firmware for my P150hm.
Does anyone have any ideas on what I can do? I thought i would at least get a POST but then struggle with the driver installation but i cant even get it to POST...
My issue: one of the two 680M GPUs started displaying an "Windows has stopped this device because it has reported problems. (Code 43)". My hardware: Clevo P37EM / Sager NP9370 My OS: Windows 10 Enterprise 2016 LTSB x64 I keep the OS always up-to-date (Windows LTSB only ships security and fix updates). I also have very few softwares installed. Basically, my computer was working normally, but today I was playing Overwatch (which I always play) and after some hours, I suddenly heard a Windows beep, the game froze and the computer restarted. When I was back at the desktop, I noticed the resolution was not right and then I opened the Device Manager to see that one of the 680M was giving a Error Code 43. Both GPUs appear on my BIOS setup, the BIOS correctly identifies them as 680M. I found it strange, so I check my GPU VBIOS and noticed something odd, one of them was at 80.04.33.00.10 and the other was at 80.04.67.00.01 (I get these drivers here). I know for a fact that since some older NVIDIA Driver, the VBIOS must be 80.04.33.00.10 or else I got some errors on this SAGER model, so I flash 33 on both a while back. I then proceeded to flash the correct version again. When I run DDU I notice it only detects the VBIOS version for one of the GPUs, the other one doesn't show the VBIOS version. My results are: NV Driver 398.36 ROM 33 PCI BUS 2 working PCI BUS 1 error code 43 ROM 67 PCI BUS 2 working PCI BUS 1 error code 43 NV Driver 391.35 ROM 33 PCI BUS 2 working PCI BUS 1 error code 43 ROM 67 PCI BUS 2 working PCI BUS 1 error code 43 Can a Nvidia update or Windows update brick a GPU? How can I know if it's a software or hardware issue? I'm afraid that one GPU died as I don't have the money to replace it now and it would be difficult to find a spare new 680M (which is now old hardware) where I live. I'm panicking a little...
Hey guys, first post, but I thought I’d discuss my heatsink fan retrofit in my old M14x r1.
Old fan was loud and not very efficient: the air transfer rate was around 2 cfm (I’m assuming cubic feet per minute). Finally had enough and decided to swap out the crappy r1 fan/coughing pinwheel for a r2 fan (rated at 10.4cfm).
I found an old r2 heat sink on eBay (beware though that the heat sink for the r2 will NOT fit the gpu/cpu configuration for the r1. HOWEVER, the fan of the r2 heat sink it is attached to is a carbon copy fit into the r1 MoBo. A few philips head screws and a seal of good electric tape and BAM, you’ve got a significantly cooler and MUCH quieter intake fan in your old R1.
My CPU temps from before were around 87-88 max load in FSX, 65 for the GPU; bear in mind this is after a repaste. Now with the new fan, I’m not going over 77 degrees on the CPU, and the gpu stays around 58, all while being MUCH quiter. I’ll post pics of the walkthrough if anyone is interested.
Just thought I’d share with all you fellow loonies trying to squeeze every last frame and degree out of our “portable toaster ovens with a screen and keyboard”.