0

I currently use a 1TB PCIe 3.0x4 (2300MB/s) NVMe SSD as both my OS boot drive and game storage drive. However, it's quickly filling up. I have a 500GB SATA III (560MB/s) SSD sitting in a drawer doing nothing. I'm thinking about wiping the NVMe, reloading the OS on the SATA III drive, and reinstalling my often played (or soon to be played) games on the NVMe SSD.

I know the conventional wisdom is to load the OS on your fastest drive, but I'm wondering if the SATA III SSD is a "fast enough" OS drive so as to not experience any performance degradation when gaming. I don't care if the OS boots a little slower or is a little more sluggish in the UI; I only care about game performance and maximizing my available storage for games.

Note that I'm not asking about PCIe 3.0 vs SATA III SSD performance in general. I understand the difference between the two buses, and I've read a lot of articles and watched a lot of videos that compare loading speeds of different types of drives. However, I haven't been able to find anything that pertains to this specific situation. Before I wipe my drive I'd like to know if anyone has any data about game performance when the gaming drive is fixed and the OS drive differs across hard drive technologies, rather than the other way around, or with both on the same drive.

On SU this is the only question I could find that comes close to mine, but not quite the same. There are also these two questions, but both of those pertain to data migration rather than performance.

MORE RESEARCH:
A different Google query led me to these threads, all concluding that having separate OS and application disks will improve performance, if for no other reason than the OS and application can have separate I/O paths (assuming that the different drives are on different buses, which in my case, they would be).

However, still no data, only speculation.

5
  • Yes, i'm sure people will work for you for imaginary points. Your UI will not be impacted by moving to a sata ssd, nor will game loading times. The only way you will possibly see any performance degradation is if you dont have enough RAM to smoothly game, but you would have noticed that with your nvme drive too. Harrymc's elaborate answer below might not have your exact idea in mind, it still functions as a guideline to see there's already little performance difference in loading times between sata and nvme. The reviewer didnt include the OS drive stats because they dont actually matter.
    – Silbee
    Nov 18, 2021 at 12:40
  • @Silbee Your conclusion sounds reasonable enough, but has anybody actually ran the test I'm asking about in order to confirm with data? As far as I can tell, no. I keep looking; still not finding anything. Hence the question and bounty. Yes, harrymc provided a nice answer, but an answer to a different question. There are at least dozens of videos of similar tests on YouTube. If that was the data I was interested in, I wouldn't have bothered posting the question, because such data is already quite ubiquitous. Nov 19, 2021 at 4:57
  • I think no one tests this because the consensus is that with such fast drives, the difference will be negligible. I know, no numbers right ? :) But that consensus is exactly why no one tests this anymore. Your edit has the right conclusion, that if anything it might have a slightly positive benefit. If i were you; i'd take the plunge. Or maybe end up benchmarking yourself and proving that it does show an impact.
    – Silbee
    Nov 19, 2021 at 8:18
  • @Silbee The video linked in harrymc's detailed answer is an excellent example of why I'm being a snob about getting data. At the beginning of that video the presenter postulated that drives with better technology (e.g. a DRAM chip or TLC v QLC memory) would perform significantly better, but the data showed otherwise. So, there may be a consensus of thought as to how games would perform in the scenario I described, but really nobody knows for sure. Regardless, I'm going to make the move, but I was hoping to have some data to give me confidence before wiping my drive and reinstalling everything. Nov 20, 2021 at 9:03
  • Also, I don't have the hardware or software to benchmark game performance, so even after I make the move, we unfortunately still will not know for sure if there was any empirical performance difference. However, I'll most likely respond to let the community know whether I noticed any perceived difference. Nov 20, 2021 at 9:09

3 Answers 3

1

In general, operating system modules reside mostly in memory. Once your game had been loaded into memory with all its dependent DLLs and libraries, the operating system's disk access should be minimal and not a major factor, meaning that it can reside on a slower disk.

There is only one case where game performance will be affected, which is when the game does lots of disk accesses of temporary files. Such files are by default allocated on the system drive, but are easily moved elsewhere by changing the TEMP environment variable for both System and User.

See for example the article How to Move Windows’ Temporary Folders to Another Drive.

6
  • Why are you posting two answers to the same question? Just take this stuff and add it to the answer that already exists. Nov 19, 2021 at 6:10
  • @Giacomo1968: This answer is from before the bounty. The answers are not answering the same questions. (Not a reason for a downvote.)
    – harrymc
    Nov 19, 2021 at 8:50
  • You have more than enough experience here to know you should not post two answers like this. Just take the text here then and add I to to the answer. Nov 19, 2021 at 15:58
  • 1
    Then in that case, this answer — which reads like an extended comment — should be removed and placed as a comment on the original question. Nov 19, 2021 at 20:31
  • 1
    @Giacomo1968: Thank for the advice, but I don't agree.
    – harrymc
    Nov 19, 2021 at 20:42
1

In answer to the request for data for game performance as vs disk-type, I found it in the YouTube episode of Best SSD for Gaming: PCIe 4.0 vs 3.0 vs SATA vs HDD Load Time Battle.

In this episode, the load-time for seven games was measured on:

  • PCIe 4.0 : 3 SSDs
  • PCIe 3.0 : 4 high-end SSDs and 2 entry-level SSDs
  • SATA 3 : 4 SSDs
  • 1 SATA III HDD

Below are screenshot extracted from the video.

CrystalDiskMark synthetic/artificial performance tests

Sequential Read Performance

The disks here are nicely ordered by technology, in exactly the same order as in the above list.

enter image description here

Sequential Write Performance

enter image description here

Random Read Performance

enter image description here

Random Write Performance

enter image description here

Games load time

Horizon Zero Dawn

enter image description here

Death Stranding

enter image description here

Shadow of the Tomb Raider

enter image description here

Red Dead Redemption 2

enter image description here

Borderlands 3

enter image description here

The Outer Worlds

enter image description here

Tom Clancy's The Division 2

enter image description here

Planet Coaster

enter image description here

Conclusions

  • There are dramatic differences between HDD and SSD for almost all types of games.

  • PCIe SSD drives were on average just seven percent faster than non-PCIe drives for loading into the game, which is pretty negligible. There was no clear advantage to PCIe 4.0 drives at all.

  • For SSDs, there is no clear preference for PCIe over SATA, as the results differ between game types, or at most a negligible advantage for PCIe.

  • In these sorts of games there is far more processing work that goes on when loading a level than copying from storage into system or GPU memory. It's much more important to have a fast system overall than is having a blazing fast SSD.

3
  • The setup and tests in this video are the opposite of my question. I'm looking for performance data when the game drive is fixed (preferably on a PCIe drive) and the OS drive differs across many different hard drive technologies. Nov 17, 2021 at 5:43
  • Although, this is the first test I've seen that uses a separate OS disk, so kudos for that. Too bad though that they don't give the specs on that disk (they classify it only as an "M.2 disk"). Nov 17, 2021 at 5:50
  • The disks specs are available on the web. You may learn from this benchmark about the behavior of "real-world" applications, as it included so many types, but with quite uniform results. The biggest jump in performance is by using SSD. Then you will have a slight improvement going from SATA to PCIe 3.0 and then to PCIe 4.0, also an additional slight improvement by using a higher-end SSD. Besides absolutely needing to use SSD, the rest is up to you according to your budget - the differences will be very slight. My interpretation is that a higher-end SSD may help a bit more than PCIe 3.0.
    – harrymc
    Nov 17, 2021 at 9:43
1

To answer the question if your game performance will be affected by your games from a PCI-E SSD onto a SATA3 SSD:

Technically yes (for asset loading times), but only when you are loading your scenes from your storage medium to be rendered by your GPU.

Loading times for games may become slightly longer due to 2000+ MB/S on a PCI-E SSD compared to only 500 MB/S on a SATA3 SSD, and if scenes are heavy with objects, there may be a slight delay in loading times for these scenes while you are in-game.

Mind you, the differences in loading times are negligible in the real-world and I doubt you would care about them, as your game will probably still load very quickly on even a SATA SSD, but it is still slower.

You will probably only be affected if your storage medium is very slow, and you need to load something big real-time (for example, loading a HUGE chunk of terrain real-time on an old IDE hard drive), in which case, your framerate won't be affected, but you will notice glitched areas due to the asset not loading while in-game yet. This won't happen with any good-quality SATA SSD.

As for framerates, NO, you will probably not notice a difference in framerates with a SATA SSD when compared to a PCI-E SSD. That is determined by how good your CPU and GPU are.

3
  • I'm asking about moving the OS to the SATA III drive, not my games. The games will still be on the PCIe drive. Nov 21, 2021 at 7:37
  • No. Your game performance will not be affected by moving your OS to a SATA drive. Your OS boot performance will suffer slightly, but you will barely notice (only a few seconds slower at most) Nov 22, 2021 at 3:12
  • That's my hope/theory as well. I just wish somebody had the data to support it. Nov 22, 2021 at 4:52

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .