Leaderboard

I want to share my current success with CPU upgrade and heatsink mods. Might be some help to others. Things in this post: -3740QM experience -Fan profile and HPFanControl experience(with dual core i7-3520m) -heatsink mods -3740QM stress test First off the heatsink mods. On top of the heatsink I glued 8 of the 20x20x1mm copper shims. In a 4x2 grid. I couldnt find any 20x20x2mm copper shims on Ebay at the time, if I did then I uused them and put them in a 4x1 grid instead. I also glued 2 of the 15x15x2mm copper shims on each side of the heatsink-pipe. I used Arctic Alumina to glue them. In between the CPU and heatsink I placed a 20x20x1mm copper shim. I used Collaboratory Liquid Ultra between the CPU and copper shim, and between the copper shim and heatsink I used Arctic Silver. I set a custom fan profile using HPFanControl, to remove the annoying constant fan ramp up and ramp down, which is like someone keeps turning a vacuum on and off all the time, Now this annoying RPM jiggle was when I had my Dual Core 3520m... I haven't had enough time to test the standard BIOS fan profile with the 3740QM, during normal laptop usage yet. The standard BIOS fan profile was making me insane when I used my 3520m, so this was really needed. So the custom fan profile makes the laptop more quiet(by allowing it to hover on higher temperatures), when I do my normal stuff like browsing or YouTube. Received the 3740QM today and installed it with the cooling mods, then set the maximum power limit to 99%. I also ran the winsat tool again, just in case. I ran an OCCT stress test to see if the laptop thermal throttles, TDP throttles, or just plainly crashes when the CPU is pushed to the max over a longer time. I disabled my custom fan profile(HPFanControl) for this test, as I have not yet tested the HPFanControl with OCCT stress test, so I did not want my results to be broken by the use of a possibly bugged homemade fan profile. I ran the test for 3.5h, using RealTemp to measure max CPU core temperatures(it has a fast polling rate), and HWInfo to verify there are no "Windows Hardware Error" counts, the max power usage, if any throttling occurs, and if the CPU frequency is stable at 2.7GHz(99% maximum limit in control center) Results: Doesn't seem to be any throttling at all, and max CPU temperature was 93 degrees at an early point in the test, which tells that the maximum CPU temperature was not climbing with time. CPU frequency seemed pretty stable at 2.7GHz the whole test. The max CPU Package Power was 32.973 W(note the iGPU was not stress tested at all). Max CPU voltage was 0.941 V. Other notes: I tried using Noctua NT-H1 but all of the paste would get pushed out of the sides of the CPU die, leaving the actual space between the die and heatsink without any paste...Arctic silver 5 worked better. I've broken the heatsink of my macbook pro once by using Liquid metal on it. The liquid metal literally started to boil the heatsink, like one of those "science exploding foam" experiments, and split it in half, all while i was standing stumped and frozen by the experience. So I can't recommend using liquid metal directly on the heatsink, as a little particle of liquid metal that somehow gets leaked to other non-copper part of the heatsink starts to melt the whole metal block. It should(hopefully) be fine if you use it between a copper shim and the CPU die, like I did here, as liquid metal and copper seems to work fine except a little color change on the copper with time. If the copper somehow gets unusable then just replace the copper shim. Better and cheaper than having to replace the whole heatsink. I haven't tried throttlestop yet as I have not figured out how I should use it. I haven't tried running a stress test with 100% max CPU limit yet, im pretty sure it will thermal throttle when running stress tests, but stress tests does not directly tell the story to most personal use cases. I'm looking for the temperature spikes during short bursts of high CPU usage, which should happen in most cases for me, like when browsing etc, and im pretty sure the CPU won't thermal throttle on those times. However I do watch a lot of YouTube in higher quality(1440p and 4K), and on an external monitor, so the CPU will be taxed over a longer time in that use case. What is most important for me is if my custom fan profile will silence the annoying fan ramping up/down. Bottom line if it runs quiet then I will keep it at 100%, but if the fan needs to ramp up then I will run it at 99%. I got a pair of Bose QC25 Active Noise Cancelling headphones to replace my old ANC headphones which I use almost all day, which silences any fan noise, even at max RPM, so I might allow the standard BIOS fan profile when I'm using the cans, as the profile is more aggressive. When I take my heaphones off while doing some CPU taxing activity I'll probably have to turn the custom fan profile on. Conclusion: My laptop didn't burn down so im glad, also now I can watch 1440p60FPS high bitrate YouTube videos and Twitch 1080p60FPS streams without dropping multiple frames a second.(Couldn't do it with my dual core 3520m) If theres any other tests anyone want me to do please give me a suggestions