XMG OASIS Performance

In the interest of not overloading your mind with data, I am only showing the results of the conducted 3DMark Time Spy Extreme stress test of 20 loops, done in exactly the same manner as earlier for a direct comparison between recorded metrics.


Look at that! There's a ~20°C decrease (~28%) in the average CPU core temperature and an even larger ~26°C decrease (~30%) with GPU core temperature under exactly the same load with the XMG OASIS compared to without. Not only that, it takes slightly longer for the components to heat up and less for them to cool down owing to the increased heat capacity of water and faster heat dissipation capacity of the larger 120 mm radiator. To further sweeten this deal, it's quieter, with collective sound decreasing from 50.4 dBA from 1.5' away without the OASIS to 39.8 dBA with the OASIS, although idle noise is higher with the OASIS pump continually churning away either way. I noted similar percentage decreases in operating temperatures under gaming and some simulation work, although I would probably turn off the OASIS for lighter office work, such as writing this review.


Given the absence of any thermal throttling now, GPU clock speed is far more consistent with the OASIS used compared to the almost instant drop without. This is direct evidence that the cooler is helping achieve faster, more consistent GPU and CPU performance where it counts.


97.1% frame rate stability—passed! This shows the benefit of the OASIS external watercooling system for longer workloads, and it's now time to see whether the same holds true for more intermittent loads and gaming, which are more realistic scenarios.


Yup, there are appreciable increases in synthetic performance involving both the CPU and GPU. For context, the laptop on its own had a Time Spy score of 13,030, and the 3ds Max 07 composite score was 143.94.


Here's an easier comparison, with two more entries over the relative gaming performance list and the OASIS taking the best spots. There's even headroom to manually overclock, but the gains are only a couple of percentage points.


Now the puzzle is fully solved, with the OASIS not only providing thermal relief, but some power relief to the CPU and GPU limits that were holding them back. In particular, dynamic boost was basically useless on the GPU before, but now results in some of the 150 + 25 W being used. It's not consistently so, but stress tests have the GPU running cooler courtesy the laptop fans consuming less power, which has dynamic boost kick in more. With the CPU, it is more about headroom to overclock, although a cooler processor and quieter system under load is always good.