When you build a PC and pay top dollar for components, you expect it to handle just about everything you throw at it for at least the next five (or maybe seven) years, depending on how light your wallet felt after assembling the rig. That's the kind of expectation I had when I built my system around an RTX 4070 Ti Super and a Ryzen 5 7800X3D. With that kind of performance headroom, there were no compromises expected, and the last thing I expected to encounter was micro-stuttering during gameplay of The Last of Us Part II.

Naturally, I blamed the usual suspects. The last Windows update was at the top of the list, and honestly, there's no reason as to why it wouldn't be. Windows has earned that sort of disrepute after a long list of update-related fiascos this year alone. A few driver rollbacks and an entire clean re-installation of the OS told me the problem wasn't where I thought it was. After all, no one looks at their storage configuration with suspicion, not at first glance anyway.

My boot drive was busier than I'd thought

It was bogged down by background indexing

Windows and its endless background processes don't exist in a vacuum, and the OS continues doing what it has always done behind the scenes. It took me a little while of troubleshooting to acknowledge that, but I've labeled it as one of the reasons that can lead to micro-stutters. Search indexing, Windows Update, antivirus scans and page file constantly adjusting based on memory requirements aren't exactly a symptom of a problem, but they do tend to compete for access to the same storage device your games rely on streaming assets.

As such, the micro-stutters do make sense. The moment a graphically intensive title such as The Last of Us Part II enters an asset-heavy scene (which is once every minute when it comes to TLoU), it sends a burst of read requests to the SSD. If Windows happens to be indexing files or writing data at the same time, those requests can end up competing for the same queue. Those requests enter the same I/O queue and wait, and the game "hitches." For the user, the assets load a fraction of a second late, frame times spike, and the experience feels like a GPU bottleneck or a driver issue, perhaps because as gamers we are conditioned to point fingers at the chip whenever the experience is disrupted in this way.

The architecture behind NVMe drives makes another case in point

There has always been a cache 22

What complicated matters much more in this respect further is the fact that most consumer NVMe drives are either designed around TLC or QLC NAND flash memory. To achieve the impressive random and sequential write speeds advertised on the box, drives rely on a dedicated SLC cache that only temporarily behaves like faster memory. Data hits this cache first, and transfer speeds look impressive for as long as it lasts. Once it is exhausted, the drive falls back to writing directly to TLC or QLC cells, and the performance drop will be obvious to any user like myself who is privy to changes in drive speed.

Since my boot drive was carrying the bulk of Windows, my full Steam and Xbox game library, update caches and shader storage simultaneously, this very well presented ideal conditions that shrink the SLC buffer and trigger cache exhaustion during sustained write operations.

When the theory made sense, I decided to split the load

Putting the spare WD Blue to work solved the problem

It sounds just as implausible as it looked, but my testing with reinstalling The Last of Us Part II on the dedicated SSD drive managed to eliminate the micro-stuttering altogether, to an extent that, after a while, it completely skipped my mind that I was testing if that continued to be an issue. It wasn't until later that I found that the solution had nothing to do with storage speed and everything to do with how the operating system and games share resources.

The test also showed me that even high-end systems are susceptible to this behavior. You can own a GPU with a huge VRAM budget, plenty of system memory, and a bleeding edge Gen 4 or Gen 5 NVMe drive and still experience the same symptoms every once in a while. Since most of a game's rendering pipeline depends on the RAM and VRAM, storage performance only becomes apparent when new assets need to be streamed from the disk.

Meanwhile, the OS continues carrying out its housekeeping tasks in the background, which means paging activity, temporary file writes, and routine tasks continue regardless of what's running in the foreground, all of them hitting the same drive. Small and scattered file operations are an IOPS problem rather than a bandwidth problem, and even a Gen 4 or Gen 5 drive does not have infinite QD (Queue Depth). Enough simultaneous, small requests hammering the NAND can and certainly will exhaust it, given enough time, and affect your latency.

But there's one more advantage worth mentioning

I couldn't help but notice that this approach may also be healthier for the longevity of the boot drive itself. Having experienced both a boot drive failure and cache exhaustion in the past, I can say that neither is a pleasant problem to have. Keeping a dedicated gaming SSD means fewer writes, less resource contention as I described, and more breathing room for the drive that keeps your OS "snappy" to feel. Even if the performance gains aren't a good argument to do so, the peace of mind knowing that your SSD isn't being subjected to constant read and write operations would've made the change worthwhile.