Hello everyone,
Firstly, thank you all very much for the hard work put into finding out the optimal settings and tweaks for this laptopIt's very helpful, and beyond expectations how much the throttling got reduced in your machines reading the experiences.
After reading the thread I've decided to try to tweak my laptop as well, hoping for being able to lightly game on it (I am eyeing Skyrim, which was said to work nicely on MX150, and mostly games having similar requirements).
The undervolting and overclocking settings were very clear to me, however, if it's not troubling, I would like to ask additional questions about thermal pads (I apologize if it is already said as clear as possible.) The thought of opening the laptop brings a bit of anxiety, I have never repasted or added thermal pads and I am afraid of making a mistake.
I understood the thermal pad should be directly sandwiched between the heatsink and the bottom of the lid (please correct me if I am wrong). I have looked online for examples how it should be applied, however I have had no luck. I think people applied them on the bottom of the heatsink instead, which didn't help me much.
I thought about buying an Arctic 1mm thermal pad (it is described as "thermal pad with thermal paste" in the local shop) - would this one be alright? Would it be wiser to go for an another company instead, or a thicker/thinner pad for this laptop and its needs?
I understood I need to cut the thermal pad in the shape of the heatsink, which is a big rectangle. Should I cut it in the shape of the heatsink visible, or so it also covers the thermal pipe? I have drawn over a photo (taken by Andrei Girbea from ultrabookreview.com) to show the areas I am puzzled about (red, or blue?). Am I completely wrong?
![]()
When I cut the thermal pad into the correct shape, should I lay it on the heatsink and close the lid? The description says the pad is self sticking - does it stick on one side, or both? If one sided, would it be correct if I lay the non-sticky side on the heatsink, with the sticky side facing upwards, so when I close the lid, it sticks to the bottom lid of the laptop?
I am very sorry for the amount of questions, and hopefully I didn't take too much time. I am very afraid of messing it up and causing the temperatures to end up even higher.
Thanks so much for reading![]()
-
QuantumCakeIsALie Notebook Enthusiast
Just lay the thermal pad over the CPU/GPU sockets, cuting it to match their respective outlines. When you close the case again, it will stick on both the heat pipes and the case; re-opening the laptop will probably tear them and require a reapplication. You can experiment with putting pads at the end of the heat pipe like you described, but I'd start with the sockets to remove some heat from the pipes ASAP.
I personally used Arctic 1.5mm 6 W/mK light blue pads. I didn't bother with repasting, I assume it could help but not that much given that the original paste is brand new.
If you stay calm and don't panic, it's not very risky to open a laptop. In doubt about something? Just stop what you're doing and look it up; don't rush. Avoid rug floors or touching fabric. Work on a solid surface, and discharge yourself by touching a big piece of metal, like your oven, prior to touching electronics.Xff34 likes this. -
It was of great help and relieved some of my anxiety too. Opening it doesn't seem so scary anymore lol. I just have to remember to be careful. Thanks so much again, it's really helpful.
I've received my laptop two days ago. The date of manufacturing seems to be August 2018, I live in a central European country (I'm not sure if it makes any difference though). I've tried checking for PWM, dimming the display as much as possible and looking through my phone camera.
PWM seems to be non existant. HWiNFO says the display is Chi Mei (the colors are indeed a bit warm when I think about it - it's as if I had a very slight orange filter on the screen). Perhaps ASUS permanently changed the display in later units? Since PWM was a very important factor to many users and a game breaker, it would be great news to have it removed.
I believe the coil whine is present, when I put my ear close to the left side of the laptop, I hear a quiet high pitched noise and some "pulsating" kind of electronic buzzing (SanDisk SD9SN8W512G1002). It is only noticeable when I look for it, however I have the laptop for a very short time and haven't put it through heavy workload yet. Which also means I haven't tested for throttling, however I wouldn't expect a change.
My experience with the laptop is very good so far, Kubuntu works great on it, keyboard and touchpad feel pleasant. Booting from USB was a bit of a problem even after disabling fast and secure boot as BIOS wouldn't see my USB drive. Strangely enough it worked only after I entered BIOS while restarting the PC instead of powering off and on and inserted the drive it into USB 2.0 socket.
----------
EDIT: I've run a fairly demanding 3D animation program while having OpenHardwareMonitor open, trying to push the GPU close to the limits.
The laptop was plugged to the AC Adapter since 2-3 restarts. The temperatures raised to 74C rather quickly, however, I've had stable fps without any sight of stuttering. I've had it running for a few minutes, the temperature bounced between 72C and 74C. I added an another 3D model to the scene, I experienced very slight stutter every half of a minute or so, but nowhere as bad as described earlier - it might have been caused by physics in a certain moment of the animation instead. (I've had around 25 fps)
As soon as I unplugged the laptop I got heavy throttling and my GPU stopped working for around half a minute (<1fps). I will be testing more. I might have not stressed the GPU enough to throttle on AC power, although the fans seemed to run on full speed and it reached the maximum temperature. I'm positively surprised though - I expected a complete disaster.
I updated to the newest BIOS (December) before, the laptop is still "raw" and not tweaked. What was a surprise to me - I examined HWiNFO's log closer and my model is apparently UX430UN R which is also confirmed in BIOS, although the laptop had UX430UN-GV115T on the sticker on the brown cardboard box.Last edited: Jan 8, 2019 -
QuantumCakeIsALie Notebook Enthusiast
-
One more thing: Intel XTU broke my power limits setting them to 28w for PL2 and up to 22W for PL1...i've tried everything, even XTU can't set it back to where it was originally and throttlestop seems to be unable to modify them...i've even tried to reinstall windows using the recovery partition...any idea on how to fix it?Last edited: Jan 17, 2019 -
QuantumCakeIsALie Notebook Enthusiast
70°C is what I get at sustained undervolted 2.7GHz with thermal pads and no repasting on a flat surface. -
-
QuantumCakeIsALie Notebook Enthusiast
The issue with TB to 4.0 GHz is that it heats the CPU fast and then it has to underclock in order to cool down, overall we're loosing versus a sustainable moderate boost for a longer time. Kind of like you wouldn't run a marathon as a succession of full-speed sprints. -
-
@QuantumCakeIsALie i've managed to set throttlestop to control TDP. I can now sustain 3.1Ghz stable @75C° (100% fan) setting a power target of 25W, but after some time i'm limited by "VR Thermal"...i suspect VRMs are overheating.
Have someone here discovered if there are any SMDs on the backside of the board? Do anyone know a way to sink CPU VRMs to the back panel? -
QuantumCakeIsALie Notebook Enthusiast
Last edited: Jan 26, 2019 -
-
Well I finally was able to get my GPU to stop doing the massive throttles and maintain a stable-ish FPS.
Here is my solution and it involves the use of 2 software +/- notebook fan control (Set to UX430UQ)
1) GPUtweak (to LOWER temperature limit rather than increase)
2) MSI Afterburner (set up a custom frequency to voltage curve)
3) Notebook fan control (maintain high fan speed to stabilize frequency)
The main issue is no matter what you do, at 73C the laptop will hard downclock, and generate massive stuttering, regardless of whether you increase the temp limit with GPUtweak. Therefore, the important step is to DECREASE the target temp with GPUTweak to 72C, preventing the massive downclock via ASUS, which results in stuttering, and instead relying on nvidia to downclock and downvolt for you.
Steps:
1) Use GPUtweak to set the target temperature to 72C. That's it with this software.
2) Use MSI afterburner to generate a custom frequency to voltage target. Maximum clock is about 1645hz, so I set it to about 1650hz at 0.881V, which did the trick for me. Default is about 1650hz @ 1.08V. Therefore you are getting a 0.2V decrease! This significantly decreases the temperature. See below. The sharp drop off at 0.875V is because my GPU would crash at this voltage with max frequency. Why use MSI afterburner instead of GPUtweak for the custom curve? Because GPUtweak only goes down to about 0.95V, and as you can see from above, my GPU was able to downvolt a lot more than the lower limit of GPUtweak. Also its important to note that my curve is also higher and steeper at lower voltages, so that means that when it "naturally" downclocks/volts (versus the sharp downclock with ASUS), it does so at a higher frequency. However the flip side is that when it downvolts, and your frequency is too aggressive, it can still crash so this portion of the curve also needs optimization.
Thats the major steps.
3) This last step is just to maintain a more consistent frequency. I use notebook fan control and set it at 100% when gaming so that the frequency is more consistent.
That's it! So now you get as high of a clock as possible, staying below 73C, and avoiding that jolting downclock with ASUS.
Hope this helps people. -
This profile as shown here : https://ibb.co/cy9nKHx
I don't know what you mean but I've just tried GPUteak II and set to the default gaming profile which caused the laptop to no longer thermal throttle, though GPU temps reach 94-98 °C (201.2-208,4 °F) and I've tried playing overwatch on full HD and low settings (previously it used to stutter hard and was some what unplayable) was as smooth as butter; though I'm afraid of frying my PC, so dunno if this temp is safe :/ .
It should be noted that my UX430U is from Austria, has 16GB RAM, a MX150 GPU and an i7-7500U CPU was manufactured on 7/2017, and finally has the BOE display which flickers. (very weird to have the 7th gen CPU but the new GPU)
Finally, I would like to ask the following:
1) Would the Dual core i7 7th gen differ in real world performance when gaming and productivity than the Quad core i5 8th gen? (because I may return if it affects gaming performance or real usage)
2) Is 94-98 a safe temp to play on for the GPU?
3) Did you guys try playing overwatch on full HD and low settings without modifying the laptop or gpu settings? Did it stutter heavily as well and make it unplayable? (was the case for me in 1360x768 as well so is my unit faulty or is it cause the 7th gen i7 is a bottle neck?)Last edited: Feb 21, 2019 -
Ok so decided to boot up this baby for some gaming and was getting the annoying fps drops. Then I realized something if I use throttlestop and disabled turbo the problem was gone. Actually in rocket league, evil within , and soul Calibur 6 got way better fps if I disabled turbo (+8 to +10 fps). I also undervolted in -74.2 (core and cache) and Intel gpu -49.8 and it's running great .Finally !
Remember, disable turbo boost!!
I think it has to do with a single heat pipe for both gpu and cpu, and the random drops coming from actually the CPU overheating. -
QuantumCakeIsALie Notebook Enthusiast
***
The issue for gaming on ultrabooks, is that the ULV CPU are designed to boost a LOT for short amount of time so that everything feels snappy, but have a low long-term frequency to keep thermals and power consumption in check.
This is a very good design for productivity and part of why it's so great to do light work on an ultrabook for extended period of time on the battery. I can code and write documents for upwards of 8 hours on a single charge on mine under linux without feeling like it's crippled in any way.
Now if you launch heavy sustained programs, like games, that's where the aggressive boost is detrimental. The CPU boosts a LOT but for a long period of time, so it ends up downclocking very aggressively to keep the heat/power in check, typically way before the program/game is finished. That's why limiting turbo boost is a good idea for such situations, if you don't buld up heat to begin with, you can stabilise at a much higher sustainable frequency.
I suggest doing that for anything really heavy that has to run for extended period of time, like gaming or rendering or encoding videos. See my tests here for an example of long term temps/clock attainable with this approach.Last edited: Mar 4, 2019alejo099 likes this. -
I have a HP spectre x360 with an i7 8550u, I would like to increase its TDP, because my CPU is bottlenecking my eGPU (RTX2060). However I cannot do it using XTU or throttle stop because the maximum TDP is locked to 15W, despite this CPU being able to work at 25W.
During stress tests, after the turbo boost turn off, I'm power throutled (PL1) at 15 W, despite my temperatures being low, and the fan being not at full speed.
The HP BIOS does not have the advanced options to configure the TDP, so I wonder if this msr method could be used to increase my TDP (for instance to 20W) without having to flash a modded BIOS.
I can use the RWutility to run msr commands, but those you posted do not work for me (for instance wrmsr 0x610 0x42016000dd8160 > "parameter error"), yet the rdmsr works:
>rdmsr 0x610
Read MSR 0x610: High 32bit(EDX) = 0x00420098, Low 32bit(EAX) = 0x00DD8160
63 56 55 48 47 40 39 32 31 24 23 16 15 8 7 0
00000000-01000010-00000000-10011000-00000000-11011101-10000001-01100000
which commands can I use to set my TDP 20 W or 25 W ?
I'm on windows, I also have ubuntu, but for gaming I use windows.
thank you very much for your help -
QuantumCakeIsALie Notebook Enthusiast
You might've to disable secure boot in order to be able to write to the MSR.c_henriques likes this. -
-
It worked, I ran a benchmark and TDP was about 8W after the turbo turn off
then I ran this command WRMSR 0x610 0x0 0x00DD8088 to get a TDP of 17W
It worked during some time, then TDP drops to 15W again.
This is the same behaviour I get If I change the turbo boost in XTU, I can get a good TDP during the first seconds but then it drops always to 15W, even with lower temperatures (PL1 throttling).
I guess I will have the same limitation on ubuntu
Do you think I can get a stable TDP of for example 20 W for long workloads if I mod the BIOS in order to be able to configure TDP settings? -
QuantumCakeIsALie Notebook Enthusiast
On my side, the MSR values are overwritten when the CPU throtlles (too hot) or on waking up from sleep.Last edited: Apr 8, 2019c_henriques likes this. -
I will see if I can get something working, and share if I am successful. -
QuantumCakeIsALie Notebook Enthusiast
Code:wrmsr 0x610 0x42016000dd8160 devmem2 0xFED159A0 w 0xDD8160
-
-
QuantumCakeIsALie Notebook Enthusiast
You can use the following python3 functions to understand the MSR values better:
Code:#!/usr/bin/python # -*- coding: utf-8 -*- def get_power_unit(x): x = format(x,'064b') ret = dict( power_unit = 0.5**int(x[-3-1:None],2), energy_unit = 0.5**int(x[-12-1:-8],2), time_unit = 0.5**int(x[-19-1:-16],2) ) return ret def _get_power_limits(x,y): x = format(x,'064b') pl = dict( pkg_power_limit_1 = y['power_unit'] * int(x[-14-1:None],2), pkg_power_enabled_1 = bool(int(x[-15-1],2)), pkg_clamping_limit_1 = bool(int(x[-16-1],2)), pkg_power_limit_time_window_1 = 2**(int(x[-21-1:-17],2))*(1.0 + int(x[-23-1:-22],2)/4.0)*y['time_unit'], pkg_power_limit_2 = y['power_unit'] * int(x[-46-1:-32],2), pkg_power_enabled_2 = bool(int(x[-47-1],2)), pkg_clamping_limit_2 = bool(int(x[-48-1],2)), pkg_power_limit_time_window_2 = 2**(int(x[-53-1:-49],2))*(1.0 + int(x[-55-1:-54],2)/4.0)*y['time_unit'], pkg_msr_lock = bool(int(x[-63-1],2)) ) return pl def get_power_limits(x,y): y = get_power_unit(y) return _get_power_limits(x,y) # Python3 function to compute hex from mV for undervolt def undervolt(mv): return format(0xFFE00000&( (round(mv*1.024)&0xFFF) <<21), '08x')
e.g. in my configuration when I do long term 2.7GHz computations:
Code:In [1]: get_power_limits(0x42016000dd8160, 0xa0e03) Out[1]: {'pkg_clamping_limit_1': True, 'pkg_clamping_limit_2': False, 'pkg_msr_lock': False, 'pkg_power_enabled_1': True, 'pkg_power_enabled_2': False, 'pkg_power_limit_1': 44.0, 'pkg_power_limit_2': 44.0, 'pkg_power_limit_time_window_1': 28.0, 'pkg_power_limit_time_window_2': 0.00244140625}
Last edited: Apr 8, 2019 -
thank you very much
this will be a big help, now I need some time to understand how all of these stuff works, because I'm not used with it. first I will try in ubuntu, as a proof of concept.
-
Last days I was very busy, but now I get time to try your suggestions. So I'm using ubuntu now.
I used these commands:
# 3.0 GHz Turbo for 4/3/2/1 cores
wrmsr 0x1ADH 0x1e1e1e1e
# Unlock full 44 watts TDP
wrmsr 0x610 0x42016000dd8160
devmem2 0xFED159A0 w 0xDD8160
it seems to work, I made a stress test the the CPU clock was capped at 3GHz, and power was about 19W. However, after some seconds, the power decreases down to 15 W, and consequently the clock decreases down to 2.6GHz.
I already tried several clock caps and TDP settings, but my 8550u will allways slow down after a while, even if the temperatures are 85ºC
any ideas?
thank you -
Bought UX533FN last week with MX150. The MX150 in the laptop differs from UX430UN. How come? I thought UX430UN had the best version of MX150?
-
QuantumCakeIsALie Notebook Enthusiast
The MSR can get overwritten when you wake up from sleep or hit high temps too.
It may also be overtook by the bios. -
yes, you may be right, it is probably the bios that overtook the MSR. if I configure TDP to 16W, it will be always capped to 15W after 70 sec, even with very low temperatures, under 80ºC.
Even so, the values I read in devmem2 and msr are not changed, they are only automatically changed if my temps are very high (as you said), which did not happen with a TDP below 19W. Despite these values remain the same ones that I set, TDP is always capped to 15W after a while no matter what I do. Perhaps there is other place in memory where CPU checks for these limits.
Probably there is no way to solve this problem without flashing a modded bios
Asus UX430UN i7-8550u, 150MX GPU, 16GB RAM, 512GB SSD $999 Costco.com
Discussion in 'ASUS Reviews and Owners' Lounges' started by HTWingNut, Nov 3, 2017.