Cooling Performance

We now have a decent idea of how this system performs in various operating states, and the limitations in squeezing out more performance. I mentioned that the power limits are a big bottleneck, but NVIDIA's dynamic boost also responds to component temperatures in a way that gets the most out of the GPU. Then there is the laptop's noise level being contingent on how fast the fans are spinning, which is dictated by hardware component and ambient temperatures. As such, knowing more about the cooling performance of the XMG NEO 15 (E22) is key. In order to test this, I used GPU-Z to log various metrics under a representative load using the best-performing operating mode (dGPU, Overboost, and Plugged). This was a "short" 35 min test to confirm everything was up to par prior to commencing to a longer stress test. While there are dozens of different graphs I could present, I chose the following as a function of time—average CPU core temperature, GPU core temperature, and GPU load. The mobile GPU doesn't allow for VRAM or real-time power draw logging as a function of TDP as the latter is dynamic and there is no easy way to measure the fan speeds, either. Testing was done in the same environmental chamber I use for all my cooling reviews, with an ambient temperature of 20°C.




Keep in mind that the averaged values are causing some of the spikes and dips compared to the overall trend lines for component temperatures as a function of time, but the largest influencing factor is the ever-changing load and associated power draw. The CPU cores hit a maximum of 92°C at 20°C ambient despite pre-applied liquid metal TIM, and the GPU core temperature went up to 87.6°C under the same conditions. At this point, the fans were at full speed or near, and the system noise level measured separately as with my fan reviews sat at a maximum of nearly 64 dBA at 6" in a 19 dBA anechoic chamber, which dropped to ~50.4 dBA with the probe 1.5' away to better represent where the laptop would be situated from the end user. I've certainly had worse with laptops, but having hot air and loud airflow blasting your desk or lap isn't pleasant either way.


Given things didn't seem too bad under a more realistic scenario with varying load on the CPU and GPU, I pushed things further with a stress test. This time around, I used 20 loops of 3DMark Time Spy Extreme and logged the same metrics as before. While maximum temperatures were about the same, the hardware very quickly hit a thermal and power limit consistently. In fact, the maximum GPU power draw during the realistic load was 151.2 W compared to the 143.2 W here, where you'd expect things to be higher. The GPU was certainly not stable throughout.


89.4% frame rate stability not passed!. There is room for improvement here thus, especially with the GPU hot-spot no doubt exceeding NVIDIA's thermal limits further. Improved cooling might also trigger some free power owing to the laptop fans not consuming as much by spinning slower with components not running as hot—now is as good a segue as any to introduce the XMG OASIS.