GL504GM Framerate Issues

[DeadFireNL]

Hi all,

Last week I've acquired a Strix Hero 2 GL504GM (intel i7-8750H, 1060 6GB), and while it is an amazing upgrade over my previous laptop, I've encountered a few very peculiar framerate issues in some games that I've been unable to track down. I suspect there are others with similar problems, but I've found it quite tough to formulate my issue into a google-able search, or to know whether other problems really are similar to this one. I'm really hoping someone can give some insight into my problem.
Running less intensive, more casual games is completely fine. I've ran the Heaven benchmark, my GPU reaches 100% load on a stable 1900MHz (automatic OC), and all the temperatures stay well within acceptable ranges (<65C). No issues either.
Suddenly I've started noticing serious framerate drops during a few games, occurring first in Planetside 2 and Destiny 2. Starting the game, I can hear the fans ramping up to keep the system cool and they do so very effectively, both the GPU and the CPU do not surpass 70C. After approximately a minute or so, every 5-10 seconds, the framerate drops from about 100 fps to 35 fps for 1-2 seconds, then jumps right back to 100+.
Investigating my system with MSI Afterburner as well as in task manager, nothing seems to be out of the ordinary, nothing indicating a CPU bottleneck, except that Afterburner indicates that at every framerate drop, the GPU clock decreases to about 1300MHz, and shows a "Power limit" as reason for the performance cap. Furthermore, the GPU utilisation does not seem to exceed about 60-70% in most cases (unlike the benchmark), but during the clock speed decrease, the utilisation now does hit 100% (due to reduced performance at low clock speeds?), which both seem odd to me.
However, when I launch EA's Battlefront 2, the issue just does not seem to arise (haven't checked average GPU utilisation, but there are no noticeable framerate drops whatsoever).

This odd combination of events suggests to me that there is no hardware power problem going on (feel free to enlighten me however), and ASUS support advised me to refresh the OS when I contacted them.
This seemed to fix the problem, I used the built-in OS refresh option, reinstalled only the games, and I no longer noticed the framerate issues in the games they previously appeared in. Relieved, I reinstalled a few of my other programs, including Afterburner. I launched Overwatch (for the first time), and immediatly the problem had returned, in both that game as well as the previous ones. It seems the problem had narrowed down to the programs that I had installed later, so I removed all of them and restarted again. Problem solved. A quick google search did indicate that Afterburner could indeed be causing issues with the GPU in some cases, although I chose to not reinstall any of the later programs. Spent a few more hours running games, temps still well under control and no further issues arose.
Next day, I launch a few of them again, and did not encounter any oddities for another few hours. I launched another game for the first time, Mirror's Edge Catalyst. Behold, the framerate issues appeared. I launched Destiny 2 again, which ran absolutely fine just half an hour earlier, and to my dismay, the issues arose.
I've tested Battlefront 2 again, for a solid 1-2 hours, and once again did not encounter any issues here.

I'm out of clues, the problem seems so random. My very last theory that this would only happen at full battery charge (for example if it repeatedly stops and restarts the charging to keep the battery at 100%) was also disproved since the issues persist at lower battery levels (while plugged in).
On a final perhaps useful note, when the laptop is running on battery, the framerate is expectedly low, but very close to when a framerate drop occurs when plugged in. Also, there are no further framerate drops when on battery.


Sincere apologies for the wall of text here. What a way to make an entrance on a community forum.
I'm really interested in hearing suggestions. I'll also be contacting support again after the weekend, but my fear is that there is not much they can do at a distance, likely meaning I'll need to RMA the laptop for something that seems more software related, which I fear will not solve the problem (aside from the hundreds of gigabytes of data I'll need to copy to avoid all those redownloads).
Sincerely thanks in advance!


Edit: Ok, I just found out a few things about a hidden Bios update to version 306, I haven't encountered it myself, my laptop is at 304 which is the same version as the website advertises being the latest. It seems to be able to resolve these fps drops, at least I see a lot of people mentioning that in threads. So, if someone has more information on how to go about acquiring a not-really-released bios update, feel free to share 🙂 (Might be due for later here in the Netherlands, I'm aware that different regions can have different release dates for software/firmware)

[silvergem]

hexaae wrote:
Stuttering is something completely different and not the case shown in the videos above.
I do not recommend K5 Pro as it degrades with time BTW. 1mm pads for VRM are long lasting and the best choice (I have a long experience with K5 Pro, many different pads with different W/mK, and many different thermal pastes).

Stuttering = random "hiccups" due to bad frame pacing while gaming
Slow-down = you randomly see the scene slow-motion-like with a fast-forward effect at the end in different games (VRM overheating 99.9%)
FPS drop = constantly lower fps in a specific scene or looking in a specific direction in the game
They have different symptoms and different causes.

Before and after I replaced VRMS (with K5-pro), The FPS dropping issue is still existed. Also based on your comment and TaitOgr which one is true?

Please see below of your comment history in https://rog.asus.com/forum/showthread.php?111503-GL504GW-K5-Pro-Thermal-Paste&highlight=k5pro

"After many tests using Arctic (6W/m²K), Gelid Solutions GP-Extreme (12W/m²K) pads 1mm, 0.5mm, or K5-Pro, and using thermal paste Noctua NT-H2 and Gelid Extreme I can say best results are with 0.5mm pads Gelid Solutions GP-Extreme + Gelid Extreme. K5-Pro wasn't any better: maybe it's good because it's sticky so it's safer for best contact, but in the end the theory of slowing down VRM's heat dissipation using K5-Pro (~5.5W/m²K) wasn't correct. Best solution was using the highest W/m²K THIN pads."

"Gelid GC Extreme for CPU/GPU. Replaced with 0.5mm >4 W/mK pads on VRAM, and 1mm 6 W/mK pads on the VRMs above CPU (those shiny silver Inductors chokes + the small black MOSFETs stripe behind):"

Here you're recommendation is 1mm pad, before it was combination of 0.5mm

[hexaae]

Yep, in my case for my heatsink on GL703GS 1mm for VRM only and 0.5mm for VRAM is the best solution in the end. Yes, in the past I thought 0.5 for both VRM + VRAM were good but after some months I noticed some random slowdowns ( = insufficient pad-heatsink contact for the VRM, again).
ATM after so many experiments (K5Pro, 0.5-1mm Gelid Pads 12W/mK, generic pads with lower 5-6W/mK giving same results, many different thermal pastes like Gelid GC Extreme, NT-H2, MX4, ThermalLight TF8 etc....) I recommend as above + Gelid GC Extreme thermal paste for GL703GS.

VRM: under big 1mm pad stripe are the silver chokes Inductors to "stabilize" voltage as a resistance, and behind are the MOSFETs that regulate voltage and are susceptible to bad dissipation and too high temps causing whole system slow-downs mentioned above and visible also in the videos. I've read they can normally work at very high temps (~100+ °C)


This is how I screw back the heatsink in detail:
- keep gently pressed the heatsink over pads only, don't push down cpu/gpu to avoid "bounce effect" potentially causing micro-holes in the thermal paste when you lift up your hand
- while keeping heatsink over pads gently pressed, follow screw order carved upon heatsink (1 to 😎 but don't screw them tightly yet
- last pass: screw them tighter following the order again while you apply a very small (!) force over CPU and then GPU

Hey... I'm not a Electronic Engineer but that's what I've learned so far that could help others, and I'm not responsible for any damage to your HW... 😉

[silvergem]

hexaae wrote:
Yep, in my case for my heatsink on GL703GS 1mm for VRM only and 0.5mm for VRAM is the best solution in the end. Yes, in the past I thought 0.5 for both VRM + VRAM were good but after some months I noticed some random slowdowns ( = insufficient pad-heatsink contact for the VRM, again).
ATM after so many experiments (K5Pro, 0.5-1mm Gelid Pads 12W/mK, generic pads with lower 5-6W/mK giving same results, many different thermal pastes like Gelid GC Extreme, NT-H2, MX4, ThermalLight TF8 etc....) I recommend as above + Gelid GC Extreme thermal paste for GL703GS.

VRM: under big 1mm pad stripe are the silver chokes Inductors to "stabilize" voltage as a resistence, and behind are the MOSFETs that regulate voltage and are susceptible to bad dissipation and too high temps causing whole system slow-downs mentioned above and visible also in the videos. I've read they can normally work at very high temps (~100+ °C)


This is how I screw back the heatsink in detail:
- keep gently pressed the heatsink over pads only, don't push down cpu/gpu to avoid "bounce effect" potentially causing micro-holes in the thermal paste when you lift up your hand
- while keeping heatsink over pads gently pressed, follow screw order carved upon heatsink (1 to 😎 but don't screw them tightly yet
- last pass: screw them tighter following the order again while you apply a very small (!) force over CPU and then GPU

Hey... I'm not a Electronic Engineer but that's what I've learned so far that could help others, and I'm not responsible for any damage to your HW... 😉

Can u stack both 0.55mm thermal pads? also do you have whole picture where u place all thermal pads

[silvergem]

hexaae wrote:
Yep, in my case for my heatsink on GL703GS 1mm for VRM only and 0.5mm for VRAM is the best solution in the end. Yes, in the past I thought 0.5 for both VRM + VRAM were good but after some months I noticed some random slowdowns ( = insufficient pad-heatsink contact for the VRM, again).
ATM after so many experiments (K5Pro, 0.5-1mm Gelid Pads 12W/mK, generic pads with lower 5-6W/mK giving same results, many different thermal pastes like Gelid GC Extreme, NT-H2, MX4, ThermalLight TF8 etc....) I recommend as above + Gelid GC Extreme thermal paste for GL703GS.

VRM: under big 1mm pad stripe are the silver chokes Inductors to "stabilize" voltage as a resistance, and behind are the MOSFETs that regulate voltage and are susceptible to bad dissipation and too high temps causing whole system slow-downs mentioned above and visible also in the videos. I've read they can normally work at very high temps (~100+ °C)


This is how I screw back the heatsink in detail:
- keep gently pressed the heatsink over pads only, don't push down cpu/gpu to avoid "bounce effect" potentially causing micro-holes in the thermal paste when you lift up your hand
- while keeping heatsink over pads gently pressed, follow screw order carved upon heatsink (1 to 😎 but don't screw them tightly yet
- last pass: screw them tighter following the order again while you apply a very small (!) force over CPU and then GPU

Hey... I'm not a Electronic Engineer but that's what I've learned so far that could help others, and I'm not responsible for any damage to your HW... 😉

So you put 0.5mm in the rest?

[silvergem]

hexaae wrote:
Yep, in my case for my heatsink on GL703GS 1mm for VRM only and 0.5mm for VRAM is the best solution in the end. Yes, in the past I thought 0.5 for both VRM + VRAM were good but after some months I noticed some random slowdowns ( = insufficient pad-heatsink contact for the VRM, again).
ATM after so many experiments (K5Pro, 0.5-1mm Gelid Pads 12W/mK, generic pads with lower 5-6W/mK giving same results, many different thermal pastes like Gelid GC Extreme, NT-H2, MX4, ThermalLight TF8 etc....) I recommend as above + Gelid GC Extreme thermal paste for GL703GS.

VRM: under big 1mm pad stripe are the silver chokes Inductors to "stabilize" voltage as a resistance, and behind are the MOSFETs that regulate voltage and are susceptible to bad dissipation and too high temps causing whole system slow-downs mentioned above and visible also in the videos. I've read they can normally work at very high temps (~100+ °C)


This is how I screw back the heatsink in detail:
- keep gently pressed the heatsink over pads only, don't push down cpu/gpu to avoid "bounce effect" potentially causing micro-holes in the thermal paste when you lift up your hand
- while keeping heatsink over pads gently pressed, follow screw order carved upon heatsink (1 to 😎 but don't screw them tightly yet
- last pass: screw them tighter following the order again while you apply a very small (!) force over CPU and then GPU

Hey... I'm not a Electronic Engineer but that's what I've learned so far that could help others, and I'm not responsible for any damage to your HW... 😉

New Thermal paste(Artic MX-4) + Thermal Pads

On green square border, Is is recommend to put a thermal pad? Based here, it has no stock Thermal Paste?

Per checking there is a gap on VRM (Silver and black square) where I put the 1mm pad beside the VRAM


AIDA64 Test


HW Test


TrottleStop


There is still VR Thermal found in throttlestop and when playing games, the FPS still go down when GPU processing is at 100%.

P.S 1: Nvidia Driver Version is 446.14, Windows 10 Pro 20H2 (19042.844) and CPU undervolted to -0.80mv
P.S 2: Thermal Padd 1mm and Thermal Pad 0.5mm

[hexaae]

Stacking 0.5+0.5=1mm pad you mean? Not at all, definitely not recommended. I can agree since it's just the VRM pad it doesn't have to be super-accurate and I've read some who did it, but it's always better a single compact piece with certified thickness. It's just 6€ 10x10cm 1mm from Amazon...
Be sure the size of the stripes are perfectly fitting the heatsink grooves over the VRM, not longer or wider than the chips.