SK hynix

SK hynix Gold P31 Spec and Price in Italy

NVME2TBM.2 22803500 MB/s read

Released August 17, 2021 ·

Where the Platinum P41 chases peak throughput, the Gold P31 sits a generation back on PCIe 3.0, capping sequential read at 3,500 MB/s — plenty for mainstream desktops but well short of the PCIe 4.0 drives in this catalog. It still carries a DRAM cache and the SK hynix Cepheus controller, delivering 570,000/600,000 random IOPS at a $280 MSRP that undercuts the Platinum P41's launch price. SK hynix marketed this drive around low power draw, making it a fit for laptop upgrades and thin-and-light builds as much as desktops.

Tied 61st of 106 models in the SSD catalog on Sequential read — 7 models share this figure. 60th of 106 models in the SSD catalog on Sequential write.

Current best price

No tracked price — this page compares specifications alone.

Specifications

Capacity
2TB
Interface
NVME
Sequential read
3500 MB/s
Sequential write
3200 MB/s
Random read
570000 IOPS
Random write
600000 IOPS
PCIe generation
3
Form factor
M.2 2280
NAND type
TLC
DRAM cache
Controller
SK hynix Cepheus controller
Endurance (TBW)
1200 TB
Warranty
5 years

Capacities

ConfigurationPart numberEndurance (TBW)Sequential writeSequential read
500GBSHGP31-500GM-21200 TB3100 MB/s3500 MB/s
1TBSHGP31-1000GM-2
2TBSHGP31-2000GM-21200 TB3200 MB/s3500 MB/s

Guides for this category

What owners and reviewers say

Summarised from 1 source(s), reviewed September 13, 2026. A source is counted only where it names this model exactly. How we track prices

Commonly praised

  • Excellent real-world and synthetic performance, costing much less than next-generation drives while performing similarly.
  • Excellent sustained write speed with onboard cache and no thermal throttling, under a five-year warranty.

Commonly criticised

  • Only the previous generation of the interface, with a somewhat small cache and a top capacity of 1TB.

Sources behind this summary

Sources

Every spec above was pulled from one of these — a field a source didn't state was left off rather than estimated.