The SK hynix Gold S31 was sent to me for review about a month ago, but we’re just getting to it at the moment because I had some amazing unrelated hardware issues we’ll talk about another time.
The SK hynix Gold S31 that I’m reviewing is a 1TB (1000 gigabytes,) solid state disk that claims sequential reads up to 560MB/s and sequential writes up to 525MB/s with a world class 1.5 million hours mean time between failures, to about 600 terabytes written before anything goes wrong with the media.
The SK hynix Gold S31 that I’m reviewing is a 1TB (1000 gigabytes,) solid state disk that claims sequential reads up to 560MB/s and sequential writes up to 525MB/s with a world class 1.5 million hours mean time between failures, and about 600 terabytes written before anything goes wrong with the media.
For reference, one million hours is 114 years, so sometime in 2190 it’s probably dead. Is it the drive going to last that long? Probably not. But it’s a nice thought. Other factors such as power supply failures, shocks, first contact with the Vulcan space fleet, etc are going to limit this to reasonable time constraints.
Testing machine specs
Dell T7500, 26 gigs of ram, single Xeon X5675 processor at 3.07GHz, 6 logical cores. Windows 2012 server. I also tested on Windows 10 same setup, double the physical processors, got slightly better results.
SK hynix Gold S31 1tb SSD attached to a LSI Logic SAS9260-8I SGL Raid 8PORT Int 6GB SAS/sata Pcie 2.0 512MB cache card.
Testing apps and settings
- CrystalDiskMark 6.0.2 x64
- ATTO Disk Benchmark 4.01.0f1
- AS SSD Benchmark 2.0.6821.41776
- Anvil’s Storage Utilities 1.1.0
Due to the cache on the LSI SATA 3 card. I set all apps to the largest test size (usually 32 gigabytes,) to make sure the card’s cache was not factored into the testing (I regularly saw 760+ megabytes per second read and write rates when I didn’t, and that’s just the cache)
I got pretty widely different results based on which app, how many threads they were using to write data, but you know how that goes. More threads = more speed.
In testing at 32 gigabytes I get 531 read, 488 write as my best sequential numbers. Or 95% of max claimed read, 93% of max claimed write. As there’s cache on board, their testing is not on sustained but max achievable speeds, I’m going to say this checks out without hesitation.
Now, I’m not going to continue this review until 2190, but the card’s in use at my work and I’ll be sure to let you know if I beat the mean time before failure.