My Zhitai TiPro9000 is hitting cache limits and causing massive read latency; should I be messing with the virtual memory?

Watching the progress bar jump around while loading 4K textures was giving me actual anxiety, especially when riding fast and seeing the world fail to load. Once the SLC dynamic cache on the Zhitai TiPro9000 gets filled by massive MODs, write speeds crater from 7000MB/s to under 800MB/s, sending I/O response times from 0.5ms up to a brutal 20 - 40ms. I tried setting the page file to half my drive space, but that just caused more disk conflicts and more drops—a total waste of time. I eventually went into Device Manager, pushed the NVMe controller queue depth to 2048, and enabled the forced write cache flush policy in system performance options. In CrystalDiskMark, 4K random reads jumped from 48 - 55MB/s to 70 - 78MB/s, cutting load times by 40%. I had some weird recognition lag at idle after the queue tweak, but switching the power plan to High Performance killed that. Temps are hovering between 45 - 55℃. The I/O curve is finally flat.