What kind of SSD do you actually need for gaming? Today we'll answer that question through a series of benchmarks. Should you go all out on the latest PCIe 5.0 technology, is PCIe 3.0 still sufficient, or can you even get away with just a 2.5-inch SATA drive? We will also test whether mechanical disk drives are still viable, though we think we all know the answer to that already.
The last time we took a deep dive into gaming storage, most blockbuster titles were still designed with older consoles in mind – machines built around slow, spinning disks. Today's landscape is very different.
Modern platforms like the PlayStation 5 and Xbox Series X ship with solid-state drives as standard, and developers are beginning to design worlds around faster storage pipelines. On PC, we have DirectStorage, and while it is not yet widespread, there are a few games that make use of it. Whether this translates into a genuine need for bleeding-edge storage remains to be seen.
Storage Specs and Overview
For this round of testing, storage maker TeamGroup was kind enough to provide us with a range of SSDs, with a broad catalog that conveniently covers just about every mainstream SSD configuration you might want to compare. We also threw in a pair of Western Digital hard drives, one built for data centers, the other a more pedestrian desktop model.
All testing was done on our PCIe 5.0-capable system, featuring a Ryzen 7 9800X3D processor, 32 GB of DDR5-6000 memory, and a GeForce RTX 5090 graphics card.
The goal was to use the fastest components possible to prevent bottlenecking the storage. All M.2 drives were installed in the primary slot connected directly to the CPU at up to PCIe 5.0 speeds, and the operating system was installed on a separate drive to isolate game performance.
The SSDs tested cover a variety of hardware configurations. Representing PCIe 5.0, we have the T-Force Cardea Z540 2 TB and T-Force GE Pro 2 TB, both using 232-layer TLC NAND with a DRAM cache. The main difference is the controller: the GE Pro uses an InnoGrit IG5666, while the Z540 uses a Phison E26.
For PCIe 4.0, we tested the T-Force Cardea A440 2 TB and T-Force A440 Lite 2 TB. The A440 has a DRAM cache and uses 96-layer TLC with a Phison E26 controller, while the A440 Lite has no DRAM cache and uses 162-layer TLC with a Phison E27T.
In the PCIe 3.0 category, we included the MP33 2 TB and MP33Q 2 TB. Both lack a DRAM cache and use the Maxio MAP1202 controller; the MP33 uses TLC NAND, while the MP33Q is a QLC NAND drive.
We also tested a USB SSD, the T-Force M200 Portable SSD 2 TB, which uses a USB Gen 3.2 2x2 interface at 20 Gbps. Inside is a PCIe SSD similar to the Z44Q, which uses 96-layer TLC, a DRAM cache, and a Phison E16 controller. Although the internals are PCIe 4.0, performance is limited by the USB interface.
Two 2.5-inch SATA 6 Gbps drives were also tested: the T-Force Vulcan Z and Vulcan Z QLC. Both use the same SMI SM2256XT controller and lack a DRAM cache. The QLC model uses QLC NAND, while the non-QLC model uses TLC NAND.
To truly stress these games and compare them with outdated systems, we also loaded all titles onto two hard drives. The first is a relatively fast model, the WD Ultrastar DC H550 18 TB, designed for data centers with a 7,200 RPM platter speed and 512 MB of cache. The second is a slower WD Blue 3 TB, which spins at 5,400 RPM and has 64 MB of cache.
Let's get into the testing.
Game Loading Performance
Assassin's Creed Shadows
We begin with Assassin's Creed Shadows, a title that uses DirectStorage and requires an SSD according to its minimum specifications. While loading the game on a hard drive is technically possible, Ubisoft displays a warning, and load times are painfully slow. This is not a recommended way to play.
Outside of this, there is only a small difference in load times between the various PCIe M.2 drives, and this does not appear to be closely tied to drive specifications. Performance drops slightly when using the M200 over USB, but it is with SATA drives that load times become noticeably longer. The Vulcan Z was more than twice as slow as the relatively entry-level MP33 when loading into the game. While it is still fine for gameplay, a PCIe drive provides a noticeably faster loading experience.
Black Myth: Wukong
In Black Myth: Wukong, load times are mildly influenced by storage speed. Moving from a PCIe 3.0 QLC drive to a PCIe 5.0 TLC drive reduced the initial load time by only 10%, which is not very significant. USB storage was 11% slower than the MP33Q, and again we see SATA drives load this title noticeably slower; the Vulcan Z took 36% longer than a PCIe 3.0 SSD.
While the game can load on hard drives, the 5,400 RPM model caused instability, with assets loading slowly and noticeable stuttering. This made for a very unpleasant way to play.
Reloading a save from within the game results in much faster load times, with minimal differences between the various storage types. Even hard drives perform acceptably here, though we still recommend an SSD for the best experience.
Cities: Skylines
Cities: Skylines is unusual in that loading a large map was faster on a SATA SSD than on a PCIe SSD in our testing. The difference was small, and all PCIe SSDs performed similarly overall. Hard drives were slower, but not severely hindered as they are in some other titles.
Cyberpunk 2077
When reloading a save in Cyberpunk 2077, there is no noticeable difference in load times between any SSD, including SATA and USB models. As long as an SSD is present, performance is optimal, with other components becoming the bottleneck. Hard drives, however, can be up to 3.8 times slower, which makes them a poor choice.
F1 25
F1 25 loads in exactly the same amount of time on all PCIe SSDs, so whether you use PCIe 3.0 or PCIe 5.0 does not matter. SATA drives were only about 8% slower, which is negligible. Surprisingly, hard drives also delivered relatively fast load times, making them viable for this title.
Fallout 4
As the oldest game in this set, Fallout 4 shows very little difference between SSDs. The game loads into our test save in around 17 seconds, and it runs just fine on a SATA SSD. However, hard drives – especially on the first load after launching the game – are significantly slower, making an SSD the better choice.
Indiana Jones and the Great Circle
One of the more modern games tested, Indiana Jones and the Great Circle loads extremely quickly despite not using DirectStorage. Fast travel between locations takes under five seconds on any PCIe SSD, with minimal difference between models. Even USB storage is fine here.
SATA drives are slower, with the Vulcan Z taking 70% longer to load than the A440 PCIe 4.0 drive, but this is not a major issue given the infrequency of these loads.
Hard drives, however, are a different story. While the game will load on them, it takes much longer, and visible pop-in persists for several seconds afterward, making the game appear broken. Playing directly from a hard drive is not recommended.
Kingdom Come: Deliverance II
In Kingdom Come: Deliverance II, initial load times are mildly affected by SSD speed. The fastest PCIe 5.0 drive, the Z540, loaded the game 15% faster than the slowest PCIe 3.0 drive, saving about five seconds. SATA drives, particularly the Vulcan Z QLC, were significantly slower, taking 46% longer than the QLC-based PCIe model.
Hard drives are essentially unusable here. The game warns against using them, and with load times exceeding four minutes, it is easy to see why. Reloading a save after entering the game reduces these differences, except for the 5,400 RPM model, which remains slow. The 7,200 RPM model was marginally faster, likely due to caching.
The Last of Us Part II
The Last of Us Part II also uses DirectStorage, but there is no meaningful difference between PCIe NVMe drives in load times. Fast USB storage is also viable. SATA drives are slower, taking at least 60% longer, while the 5,400 RPM drive was slower still. Interestingly, the 18 TB 7,200 RPM hard drive performed similarly to SATA SSDs, though PCIe drives still provided the best experience.
Spider-Man 2
Spider-Man 2 is known for exceptionally fast load times thanks to DirectStorage. Drive type matters less than simply having a PCIe SSD, which ensures the fastest results. SATA drives still work but can be nearly four times slower in the worst cases, which impacts fast travel noticeably.
A side-by-side comparison of the GE Pro PCIe 5.0 drive and the Vulcan Z QLC SATA drive shows the PCIe model loading almost instantly, while the SATA drive pauses for a few seconds. While not a deal-breaker, the PCIe drive clearly offers a smoother experience.
For a better representation of image quality comparisons, check out the HUB video below:
Ratchet & Clank: Rift Apart
Ratchet & Clank: Rift Apart aims for near-instant loading, and PCIe drives achieve this with load times of just 1.3 seconds. SATA drives are nearly five times slower, and hard drives are slower still. While all drives can load the game, anything involving frequent loading is much more sluggish on SATA.
In the portal sequence, the GE Pro PCIe 5.0 drive and MP33 PCIe 3.0 drive performed almost identically, despite large differences in rated sequential speeds. The SATA drive, however, lagged significantly, causing Ratchet to hang mid-air as the game loaded. In some segments, the SATA SSD struggled to keep up, leading to performance issues.
For a better representation of image quality comparisons, check out the HUB video below:
Red Dead Redemption II
Next, we revisit Red Dead Redemption II, a title we also used to test SSD performance five years ago. Once again, there is no difference between any of the SSDs. The game does not care whether you use a PCIe or SATA drive, as long as it is an SSD. All SSDs outperform hard drives in this title.
Civilization VI
In Civilization VI, there is no significant advantage to any particular SSD technology, as long as an SSD is in use. SATA models were up to 7% slower than the best PCIe drives, but this equates to only about a one-second difference, which is negligible in practice.
Stalker 2
Stalker 2 is another example where the type of SSD makes little difference, provided you avoid a physical hard drive. On the 5,400 RPM model, we observed the game loading incorrectly, with missing assets when continuing into the game. This issue did not occur on any SSD. The game lists an SSD as part of its minimum specifications, and based on our testing, that requirement appears valid.
Starfield
The final title in this set is Starfield. Load times here are only slightly influenced by storage speed. The difference between PCIe 5.0 and PCIe 3.0 drives was less than one second, though PCIe 5.0 models consistently loaded a little faster. SATA drives could be over 50% slower when loading the game for the first time, but they remain a viable option. Hard drives, however, are extremely slow and not recommended.
When reloading a save rather than performing a cold load into the game, the differences between SSDs diminish, as the game becomes bottlenecked by other components. The only clear exception is hard drives, which remain far slower and are not officially supported, as the game requires an SSD.
More Storage Benchmarks
Loading games is not the only task you might perform with PC storage. We also ran two additional tests related to gaming: verifying game files and moving games to and from the drive. These are built-in Steam features we use frequently, and faster drives can sometimes deliver noticeable speed improvements here.
Verifying Game Files
The first test involved verifying Stalker 2, a 150 GB game containing around 1,100 files. With PCIe drives, storage speed was not a major bottleneck, with only a 4% difference between PCIe 5.0 and PCIe 3.0 models. The Ryzen 7 9800X3D appears to be the limiting factor in this case. However, when moving to SATA SSDs, the verification process took at least 38% longer, so PCIe storage offers a clear advantage for this task. Hard drives were significantly slower, as expected.
We also tested verification on F1 25, a smaller game at 83 GB, but one spread across 36,000 files. This was the only example where the presence of a DRAM cache made a measurable difference. The three PCIe models with DRAM were 8 – 11% faster than the three without, most noticeably when comparing the A440 to the A440 Lite. Overall storage speed also played a role, with SATA drives being 24% slower than the DRAM-less PCIe models.
Copying Game Files to Another Drive
When copying files to another drive (in this case, a PCIe 4.0 SSD), PCIe models performed similarly overall. The Z540's superior read speeds made it 27% faster than the slowest PCIe model, but all were quick. SATA drives, however, were much slower.
Moving games from a SATA drive to a faster one could take at least five times longer than a PCIe-to-PCIe transfer. This matters if you maintain a secondary archive drive for games you frequently move back to your main SSD. Hard drives were slower by an order of magnitude, with a 100 GB game taking about eight minutes to copy to an SSD before it could be played.
Copying games onto the target drive told a similar story, with two key factors improving speed: a faster interface with faster NAND, and a larger cache. Copying Assassin's Creed, a 123 GB game, to the A440 took 38% less time than to the A440 Lite. While the A440's rated sequential write speed is 6.9 GB/s versus the A440 Lite's 6.4 GB/s, this mostly reflects SLC cache performance.
The A440's cache is around 200 GB, large enough to store the entire game before committing it to TLC NAND. The A440 Lite's 48 GB cache means over half the game must be written directly to slower TLC NAND, reducing performance. Interestingly, the older MP33 PCIe 3.0 drive, with more cache than the A440 Lite, copied the game faster. If you copy large games frequently, it is worth researching SLC cache size and sustained write speeds.
SATA drives were far slower, and those using QLC NAND fared even worse. QLC is inherently slower to write to than TLC, and over a SATA interface, large writes become painfully slow. The Vulcan Z QLC was the slowest drive tested for this task, even slower than a 5,400 RPM hard drive, while the TLC-based SATA model only matched the WD Blue.
QLC on PCIe drives can also be a limitation, though its impact depends on cache size. The MP33Q's reasonable SLC cache handled nearly all of a 123 GB game at full speed, but copying multiple large games would exhaust the cache, forcing much slower sustained writes. Whether this matters depends on how often you perform large file transfers.
When Faster Drives Matter
For gaming on a PC, the storage advice is refreshingly simple: buy a PCIe NVMe SSD.
Beyond that, it generally does not matter whether it is PCIe 3.0, 4.0, or 5.0.
If you want to focus on game loading and gameplay, PCIe 5.0 models occasionally eked out a single-digit edge over PCIe 3.0 in modern DirectStorage titles, but those wins were rare and small. PCIe 3.0 is still perfectly viable and is often just as fast as PCIe 4.0 or 5.0 options.
SATA SSDs can still handle today's games, but they're showing their age. Some titles that load in seconds on PCIe can take two to four times longer on SATA, and in rare moments – like the portal sequence in Ratchet & Clank – they struggle to keep up with real-time asset streaming. As more developers target "instant" loading, SATA risks slipping from fine to frustrating.
We strongly recommend against running games directly from a hard drive, with any form of SSD providing a significantly faster and better experience. While most of the tested titles technically run on HDDs, the experience is degraded with extremely long load times and frequent asset streaming issues. Most new PC releases now list an SSD as a minimum requirement.
Within the PCIe SSD category, any model will work for gameplay, but faster drives can help outside the game itself. Transferring large games is much faster with TLC NAND than with QLC, and large SLC caches or DRAM can further improve performance. These features are worth prioritizing if your SSD will also host your operating system or if you frequently move large games.
SATA drives are increasingly difficult to recommend with their former price advantage now evaporating. In our testing, the 2 TB TeamGroup Vulcan Z SATA drive costs $92, while the MP33 PCIe 3.0 drive is $107. The $15 difference is well worth it for the performance gain. However, upgrading from PCIe 3.0 to PCIe 4.0 or 5.0 often provides no in-game benefit. For example, spending $140 for an A440 PCIe 4.0 or $200 for a GE Pro PCIe 5.0 is not justifiable for gaming alone.
PCIe 4.0 models are now inexpensive enough that if the price difference to PCIe 3.0 is minimal (around 5%), it may be worth the upgrade. PCIe 5.0 drives, starting at $170 for 2 TB, cost at least 40% more than PCIe 4.0 models without offering meaningful benefits for gaming.
These recommendations are not limited to TeamGroup products, of course. For example, it makes sense to spend $8 more to move from a Crucial BX500 SATA SSD to a Crucial P3 Plus PCIe 4.0 SSD. The Samsung 990 Evo 2 TB has nearly the same MSRP as the 870 EVO SATA SSD and is currently cheaper. And the list goes on.
Large-capacity drives show the same trend. The cheapest 8 TB SATA SSD on Newegg, the Samsung 870 QVO, costs $730. In comparison, the WD Black SN850X 8 TB PCIe 4.0 drive is $600, with a DRAM cache and TLC NAND. Doing your research pays off.
Bottom line, as long as you have a PCIe SSD for gaming, you are set to get the best possible experience right now.
Shopping Shortcuts:
- TeamGroup MP33 2 TB SSD on Amazon
- TeamGroup T-Force A440 Lite 2 TB SSD on Amazon
- TeamGroup T-Force Cardea A440 PCIe 4.0 SSD on Amazon
- TeamGroup T-Force GE Pro PCIe 5.0 SSD on Newegg
- TeamGroup T-Force Cardea Z540 PCIe 5.0 SSD on Amazon
- TeamGroup T-Force M200 Portable SSD 2 TB on Amazon
- TeamGroup T-Force Vulcan Z SATA 2 TB SSD on Amazon
- TeamGroup T-Force Vulcan Z QLC SATA 2 TB SSD on Amazon