In our earlier post, we tested the impact of Hardware-Accelerated GPU Scheduling on the top-end GeForce RTX 2080 Ti and found the results rather fascinating. Not only does it reduce the GPU memory consumption but lowers the main memory utilization as well. In this post, we’ll do the same test with a lower-end GPU, namely the GTX 1650 Super (Zotac dual fan) which comes with just 4GB of GDDR6 memory. This part is most likely to run into situations where the performance is limited due to a memory bottleneck. Like before, we’ll first look at the game benchmarks, followed by a lot at the memory and VRAM consumption.
1080p Gaming Benchmarks w/ Hardware Accelerated GPU Scheduling
Memory and VRAM Comparison
In the case of the GTX 1650, the video memory utilization actually increased with Assassins’ Creed Odyssey and Origins upon turning on GPU scheduling, but Borderlands 3 performed as expected with a drop in VRAM consumption with the feature enabled. It’s interesting to note that in the latter, the VRAM usage is maxed out but the former still leaves around 250-400MB unused through the course of the benchmark.
As for the main memory and page file, those seem to be unaffected in this case, with GPU scheduling actually increasing it by a bit across all three titles. We skipped Metro Exodus as it didn’t show any impact on performance or memory consumption upon using GPU scheduling. We’ll be redoing our tests for the 2080 Ti post just to make sure nothing went wrong.