By Walter Jobs, Technology writer and editor · Published 8 October 2026
Short answer: your drive does not change your average frame rate, and every benchmark that measures average frame rate correctly finds almost no difference. But average frame rate is the wrong metric for this question. Storage causes stutter, and stutter does not show up in an average.
That is why the usual answer, “an SSD only improves load times”, is both technically correct and misleading. It improves a great deal more than load times; it just does not improve the number on your frame counter.
What storage does and does not touch
| Symptom | Is storage responsible? | What actually controls it |
|---|---|---|
| Average frame rate | No | Processor and graphics card |
| Load times | Yes, enormously | Drive speed |
| Texture pop in while moving | Yes | Drive speed and VRAM |
| Hitching when crossing an open world | Yes | Drive speed, and the engine’s streaming |
| Stutter in the first minutes of a session | Partly | Shader compilation, which is CPU bound but cached on disk |
| 1% lows | Sometimes | Processor, and streaming stalls |
| Stutter when memory runs out | Yes | Paging to disk, caused by too little RAM |
Why the benchmarks say no and your experience says yes
The standard test runs a built in benchmark on an SSD, then the same benchmark on a hard drive, and reports the averages. They come out within a frame or two of each other, so the conclusion is that storage does not matter.
That test is measuring the wrong thing. A built in benchmark usually runs a fixed camera path through already loaded geometry, which is the one scenario where storage is not being asked to do anything.
What actually happens in a real session is that you sprint across a city, the engine needs assets it has not loaded yet, and the game waits for the drive. That produces a frame time spike: one frame takes 120 ms instead of 14 ms. You feel a hitch. Across a 60 second average, one spike moves the number by a fraction of a frame.
Averages hide stutter by design. If you want to know whether storage is hurting you, watch a frame time graph while you move through the world, not the frame counter.
Where a hard drive genuinely falls apart
A mechanical drive has to move a physical head to reach data. Random reads, which is what asset streaming is, take milliseconds rather than microseconds. Modern engines assume otherwise.
- Open world traversal. Fast travel, sprinting or driving across a large map requires a continuous stream of assets. This is where hard drives hitch most visibly.
- Games that require an SSD. A number of recent titles list one as a minimum requirement, and they mean it. Running them from a hard drive produces stutter no settings change fixes.
- Texture pop in. Surfaces load in blurry and sharpen a second later, because the drive could not deliver the high resolution texture before the camera arrived.
- Shader cache and first run stutter. Compiled shaders are written to and read from disk. A slow drive makes the first session rougher than it needs to be.
If you still run games from a mechanical drive, moving them to any SSD is the largest felt improvement available for the money, and it will not move your frame counter at all.
NVMe versus SATA: the honest answer
A SATA SSD reads at roughly 550 MB/s. A modern NVMe drive reads at 3,000 to 7,000 MB/s and beyond. That is a tenfold difference on paper, and it produces almost nothing in games.
| Upgrade | Load times | Traversal stutter | Average FPS |
|---|---|---|---|
| Hard drive to SATA SSD | Transformative | Largely fixed | No change |
| SATA SSD to NVMe Gen 3 | Modestly faster | Marginal | No change |
| NVMe Gen 3 to Gen 4 or 5 | Barely measurable | None | No change |
The reason is that games issue relatively small random reads rather than large sequential ones, and random performance between SATA and NVMe is far closer than the headline sequential numbers suggest. The jump that matters is the first row. After that you are buying numbers rather than experience.
One real exception: DirectStorage, which lets a game load compressed assets straight to the graphics card and decompress them there. In the handful of titles that implement it properly, drive speed can affect frame rate rather than just loading. It remains rare enough that it should not drive a purchase today, but it is why “storage never affects FPS” is stated too confidently elsewhere.
DirectStorage, and why the blanket answer is now slightly wrong
Every article on this topic, including the first half of this one, says storage does not affect frame rate. That has been true for the whole history of PC gaming and it is beginning to stop being true.
DirectStorage lets a game load compressed assets straight from the drive to the graphics card and decompress them there, bypassing a trip through the processor and main memory. In titles that implement it fully, drive throughput becomes part of the frame pipeline rather than only part of loading.
| Traditional loading | DirectStorage | |
|---|---|---|
| Path | Drive to CPU to RAM to VRAM | Drive to VRAM |
| Decompression | Processor | Graphics card |
| Processor cost | Significant | Small |
| Affects frame rate? | No | Can do, in demanding streaming scenes |
The honest position today: this affects a handful of titles, and it is not a reason to buy a faster drive right now. It is the reason to stop saying storage can never affect frame rate, because that statement has an expiry date and it is approaching.
Shader compilation, the stutter everyone blames on storage
A large share of what people experience as storage stutter is not storage at all. Modern games compile shaders for your specific graphics card and driver combination, which is processor work, and the results are cached to disk.
- First session on a new game or after a driver update is when this happens, because the cache is empty.
- It presents as hitching at new effects or new areas, which looks exactly like an asset streaming problem.
- A faster drive helps a little, because the cache is read and written, but the compilation itself is the processor.
- The real fix is patience. If a game offers a shader pre compilation step at launch, let it finish.
Diagnosing this correctly matters because the fixes are completely different. If the stutter goes away on the second playthrough of the same area, it was shader compilation. If it happens every time you cross the same bridge, it is streaming.
When the real culprit is memory, not the drive
If your system runs short of memory, Windows pages data to disk. That produces constant drive activity and severe stutter, and it looks exactly like a slow drive because the drive genuinely is the bottleneck at that moment. The cause is upstream.
| Symptom | Storage problem | Memory problem |
|---|---|---|
| Stutter while crossing the map | Likely | Possible |
| Stutter in combat at a fixed location | Unlikely | Likely |
| Constant disk activity throughout | Possible | Very likely |
| Alt tabbing takes many seconds | Unlikely | Very likely |
| Gets worse the longer you play | Unlikely | Very likely |
On a system with 8 GB of memory, add memory before you replace the drive. It is usually cheaper and it is far more likely to be the actual cause.
What a drive does for load times, which is what you will actually feel
The frame rate answer is no. The time you spend waiting answer is enormous, and that is the number worth optimising, because it is the one you experience dozens of times per session.
| Drive | Typical large game load | Fast travel | Shader cache read |
|---|---|---|---|
| Mechanical hard drive | 60 to 120 seconds | Slow, with hitching | Slow |
| SATA SSD | 10 to 25 seconds | Smooth | Fast |
| NVMe Gen 3 | 8 to 20 seconds | Smooth | Fast |
| NVMe Gen 4 or 5 | 7 to 18 seconds | Smooth | Fast |
The gap between rows one and two is the upgrade. Everything after that is diminishing returns, which is the opposite of how these drives are marketed.
Is your drive the problem? A two minute test
- Watch frame times, not frame rate. Turn on a frame time graph in your overlay and move quickly through a large area. Spikes that line up with crossing into new territory point at storage.
- Check disk activity during a hitch. Open Task Manager on a second screen. If the drive hits 100% active time exactly when the game stutters, that is your answer.
- Rule out memory first. If you have 8 GB of RAM, the stutter is more likely to be paging to disk than the drive being slow. Memory is the cheaper fix.
- Compare installs. If you have both drive types, install the same game on each and play the same area. The difference is immediate and obvious, in a way a benchmark run is not.
What to actually buy
- Any SSD beats any hard drive for games, by a margin you will feel every session.
- A good SATA SSD is genuinely fine if the price gap is meaningful. You are not missing much.
- Buy capacity over speed. A 2 TB Gen 3 drive serves you better than a 1 TB Gen 5 one, because the worst storage outcome is having to keep games on a hard drive for lack of space.
- Keep a hard drive for storage, not for games. Media and backups do not care.
The load times and transfer figures described here are drawn from published drive specifications and from our own model, not from a timed run on each drive. Where the model stops being reliable is explained on how we calculate this.
Frequently asked questions
Does an SSD increase FPS?
No. Average frame rate is set by your processor and graphics card. An SSD improves load times, reduces texture pop in, and removes the hitching that happens when a game has to wait for assets, none of which the frame counter shows.
Can a hard drive bottleneck gaming?
Yes, but not in the way people mean. It will not lower your average frame rate. It will cause stutter during traversal and loading, long load times and visible texture pop in, which many people experience as the game running badly.
Is NVMe better than SATA for gaming?
Slightly, and much less than the specifications suggest. The meaningful upgrade is hard drive to any SSD. Going from SATA to NVMe shortens load times modestly and does nothing for frame rate.
Does storage affect 1% lows?
It can. A streaming stall produces a single very long frame, which is exactly what 1% lows measure. This is the metric where storage shows up and average frame rate does not.
Will moving a game from HDD to SSD stop stuttering?
If the stutter happens while moving through the world or loading new areas, usually yes. If it happens in combat or in crowded scenes at a fixed location, that is a processor limit and a drive will not help.
Do I need a Gen 5 SSD for gaming?
No. Games do not issue the large sequential reads that Gen 5 is built for. Spend the difference on capacity or on a better graphics card.
Does an SSD help with shader compilation stutter?
A little. Compilation itself is processor work, but the shader cache is read from and written to disk, so a fast drive smooths the first run. The real fix is letting the game finish compiling before you play.
Does DirectStorage make SSDs affect FPS?
In the small number of games that fully implement it, yes, because assets travel from the drive to the graphics card and are decompressed there rather than going through the processor. It is still rare enough that it should not drive a purchase today, but it is why the blanket claim that storage never affects frame rate is starting to age.
Is my stutter shader compilation or storage?
Play the same area twice. If the stutter is gone on the second pass, it was shader compilation, which is processor work cached to disk. If it happens every time at the same place, it is asset streaming and therefore storage.
Should I buy more RAM or a faster SSD?
If you have 8 GB of memory, buy memory first. A system that is paging to disk shows constant drive activity and severe stutter that looks exactly like a slow drive, and no drive upgrade fixes the underlying cause.
Find your actual bottleneck
If your frame rate is low rather than stuttery, storage is not the cause. Check your processor or your graphics card against what it should be delivering.
If your frame rate is low rather than your experience being stuttery, storage is not the cause. Check which of your two main parts is the limit.