SK hynix Gold P31 M.2 NVMe SSD Review: High-Performance, Unprecedented Efficiency (Updated)

The first SSD with 128-layer NAND, SK hynix’s Gold P31 sets the bar for efficiency

SK hynix Gold P31 2TB Update
Editor's Choice
(Image: © Tom's Hardware)

Why you can trust Tom's Hardware Our expert reviewers spend hours testing and comparing products and services so you can choose the best for you. Find out more about how we test.

SK hynix Gold P31 (1TB)

(Image credit: Tom's Hardware)

RATING: ★★★★ ½

PROS

+ Top-tier performance
+ Unprecedented power efficiency
+ Competitive endurance & 5-year warranty
+ Single sided-form factor
+ Low cost

CONS

- Lacks AES 256-bit encryption
- Unsightly green PCB
- Lacks SSD toolbox

OUR VERDICT

The first SSD with 128-layer NAND, SK hynix’s Gold P31 delivers great performance and power efficiency for the money.

Comparison Products

As the SK hynix Gold P31 touts high-performance, we find it fitting to compare it against the best SSDs going. From the company's arch-nemesis, we include the Samsung 970 EVO Plus, as well as the 2-bit MLC V-NAND-based 970 PRO. We also threw in one of our favorite SSDs, Adata’s XPG SX8200 Pro and Seagate’s super-fast PCIe Gen4-based FireCuda 520. Crucial’s P5 and MX500 were included as were WD’s Black SN750, Blue SN550, and WD Black 7,200 RPM HDD

Game Scene Loading - Final Fantasy XIV

Final Fantasy XIV Stormbringer is a free real-world game benchmark that easily and accurately compares game load times without the inaccuracy of using a stopwatch.

(Image credit: Tom's Hardware)

SK hynix’s Gold P31 delivers exceptional game load performance. With the fastest load time in the comparison pool, it goes unmatched. Not only did it beat Samsung’s 970 EVO Plus and 970 Pro, it even bested Adata’s SX8200 Pro, one of the most responsive SSDs we have tested thus far.

Transfer Rates – DiskBench

We use the DiskBench storage benchmarking tool to test file transfer performance with our own custom blocks of data. Our 50GB data set includes 31,227 files of various types, like pictures, PDFs, and videos. Our 100GB set includes 22,579 files with 50GB of them being large movies. We copy the data sets to new folders and then follow-up with a reading test of a newly written 6.5GB zip file and 15GB movie file.

With a PCIe 4.0 x4 interface, Seagate’s FireCuda 520 outscores the test pool, eclipsing the performance of SK hynix’s Gold P31. However, the Gold P31 still offers up the second-fastest performance we have seen in our read tests. With a 4-channel controller, it comes up short of fourth place in the the 50GB and 100GB copy tests. Both the Samsung SSDS and the Adata XPG SX8200 Pro take the lead in the smaller transfer while the WD Black SN750 replaces the Adata on the boards at 100GB. Still, the SK hynix Gold P31 delivers very respectable results overall.

Trace Testing – PCMark 10 Storage Tests

PCMark 10 is a trace-based benchmark that uses a wide-ranging set of real-world traces from popular applications and common tasks to measure the performance of storage devices. The quick benchmark is more relatable to those who use their PCs for leisure or basic office work, while the full benchmark relates more to power users.

Both high bandwidth and low latency random performance are needed to do well in PCMark 10’s quick and full system drive benchmarks. Fortunately for SK hynix, the company’s Gold P31 offers up both, scoring third place in both benchmarks. It clearly is superior in performance to Samsung’s offerings under light use cases, but it fell behind both the Phison E16-powered FireCuda 520 as well as the insanely responsive Adata.

Trace Testing – SPECworkstation 3

Like PCMark 10, SPECworkstation 3 is a trace-based benchmark, but it is designed to push the system harder by measuring workstation performance in professional applications.

While SK hynix’s Gold P31 scores third place once more, this time under a bombardment of prosumer workloads, it delivers very strong performance. It completed the benchmark over a minute faster than the Adata XPG SX8200 Pro, two minutes faster than the 970 EVO Plus, and absolutely leaves the Crucial P5 in the dust. Although even with all the SLC caching and tuning and tweaking the company could do, the TLC NAND still can't sustain heavy writes or respond as well to heavy abuse as Samsung's 3D MLC V-NAND on the 970 Pro, which scores first place in this benchmark.

Synthetic Testing - ATTO / iometer

iometer is an advanced and highly-configurable storage benchmarking tool while ATTO is a simple and free application that SSD vendors commonly use to assign sequential performance specifications to their products. Both of these tools give us insight into how the device handles different file sizes.

In ATTO, we tested SK hynix’s Gold P31 at a QD of 1, representing most day-to-day file access at various block sizes and it performed very well. In sequential reads , the Gold P31 maintained a strong lead over the competition from the 4KB file size up to 512KB. However, similar to the Crucial P5, at the 1MB file size, there was a significant decline in performance, falling just below 2 GBps. Sequential write performance was strong regardless of the file size and final sequential performance figures peaked at 3,597 / 3,408 MBps.

Additionally, random performance from the SK hynix is phenomenal. Not only does it offer up random read speed that is nearly on par with the Adata XPG SX8200 Pro, but also it’s random write performance is a bit better. Not to mention, when hammered with a high QD workload, it delivers the second-highest random read result of 505,910 IOPS and a very competitive 519,201 write IOPS.

Sustained Write Performance and Cache Recovery

Official write specifications are only part of the performance picture. Most SSDs implement a write cache, which is a fast area of (usually) pseudo-SLC programmed flash that absorbs incoming data. Sustained write speeds can suffer tremendously once the workload spills outside of the cache and into the "native" TLC or QLC flash. We use iometer to hammer the SSD with sequential writes for 15 minutes to measure both the size of the write cache and performance after the cache is saturated. We also monitor cache recovery via multiple idle rounds.

At 1TB, the Gold P31 delivers some very-strong write performance from empty, outperforming some top competitors. SK hynix’s Gold P31, absorbed roughly 90GB-92GB of data at a rate of 3.4 GBps before degrading in performance. The SSD’s initial after cache performance averages a rather inconsistent 1570 MBps before finally degrading to a more consistent 1,475 MBps until full.

After letting the SSD idle over multiple idle rounds ranging from 30 seconds up to a half an hour, only a 6GB SLC write cache recovered each time. That’s in contrast to the behavior of Samsung, Crucial, and Phison SSD controllers, but similar to the Silicon Motion SM2262EN’s behavior that takes longer idle periods to recover the remaining dynamic cache.

Power Consumption and Temperature

We use the Quarch HD Programmable Power Module to gain a deeper understanding of power characteristics. Idle power consumption is an important aspect to consider, especially if you're looking for a laptop upgrade. Some SSDs can consume watts of power at idle while better-suited ones sip just milliwatts. Average workload power consumption and max consumption are two other aspects of power consumption, but performance-per-watt is more important. A drive might consume more power during any given workload, but accomplishing a task faster allows the drive to drop into an idle state faster, which ultimately saves power.

We also monitor the temperature of the drive via the S.M.A.R.T. data and an IR thermometer to see when (or if) thermal throttling kicks in and how it impacts performance. Bear in mind that results will vary based on the workload and ambient air temperature.

With just a four-channel controller, LPDDR4 DRAM, and some very efficient high-speed flash, SK hynix’s Gold P31 is the current champ to beat when it comes to file copy efficiency. With an efficiency score of 350 MBps per watt consumed, it blows past the competition.

It consumed under 2.43W on average and peaked at just a tad over 4W, under our 50GB copy workload, and at idle it consumes fairly little power, especially in its active state with ASPM disabled. When ASPM is enabled, the Gold P31 could use a bit more idle optimization; Adata’s SX8200 Pro sips just 10mW in its lowest idle state on our desktop test bench.

And, because it consumes so little power, the SK hynix Gold P31 can sustain being written to continuously without much concern of thermal throttling. With an IR thermometer, we measured peak controller package temps of 75 degrees Celsius near the end of a 350GB write transfer with no airflow in a 24 degrees Celsius environment. And at idle, it measured 35 degrees Celsius.

MORE: Best SSDs

MORE: How We Test HDDs And SSDs

MORE: All SSD Content

Sean Webster
Storage Reviewer

Sean is a Contributing Editor at Tom’s Hardware US, covering storage hardware.

  • King_V
    Some nits to picK pick (because, clearly, I'm more likely to fumble when I'm pointing out someone else's fumbles)
    Spec table column headers both say 1TB, the first one should be 500GB (or 512?)
    Paragraph below spec table says "though the 500GB model suffers a minimal performance loss with a write speed that is 100MBps lower" but the table shows the 500GB model having a faster sequential write speed. Which one is correct?That aside - I'm definitely intrigued that it's competitive speed-wise, and manages that impressive level of power efficiency. I think this one might be on my list for a future PC.
    Reply
  • seanwebster
    Thanks, I actually saw that too not too long ago and already alert the team on the errors. The write speeds should be swapped. 3,200 on the 1TB and 3,100 on the 500GB. Correction should go live shortly.
    Reply
  • derekullo
    rRZ-IxZ46ngView: https://www.youtube.com/watch?v=rRZ-IxZ46ng

    Kermit does not agree!
    Reply
  • cknoettg
    One more possible point of contention.

    The article states that the P31 Gold does not support AES-256.

    But SK Hynix's own website for the P31 gold states that it does:
    https://skhynix.freshdesk.com/support/solutions/folders/48000658600#faq_title
    "All SSDs from SK hynix come with the AES 256-bit encryption feature. "

    I have seen manufacturer's documentation be wrong, but is there any way to test to see if it does or not?
    Reply
  • seanwebster
    cknoettg said:
    One more possible point of contention.

    The article states that the P31 Gold does not support AES-256.

    But SK Hynix's own website for the P31 gold states that it does:
    https://skhynix.freshdesk.com/support/solutions/folders/48000658600#faq_title"All SSDs from SK hynix come with the AES 256-bit encryption feature. "

    I have seen manufacturer's documentation be wrong, but is there any way to test to see if it does or not?
    Nice find. This wasn't posted nor articulated at the time of publishing. We will update to include this.
    Reply
  • DZIrl
    And what is benefit for user in using 128 layers SSD? Nothing! Drives even cheaper to produce are not cheaper then any other.
    Why then?
    Reply
  • Rogue Leader
    This thread has been reopened due to the article update
    Reply
  • Soaptrail
    Rogue Leader said:
    This thread has been reopened due to the article update
    Thank you.

    Anyone know what changed in the new firmware. I skimmed the article and must have missed it. Also it appears not to affect the 1TB drive since those numbers were not updated. I know the 500GB page was updated but what were the original benchmarks for that drive?
    Reply
  • retrofitit
    Who cares if it has a green PCB?
    I put it in my laptop, put the bottom back on and it is not visible.
    The desktop another is going in has no glass panel.
    It could have a hot pink PCB or baby poo yellow PCB and I would not care, the efficiency and performance is all that matters.
    Reply
  • retrofitit
    What I would like to know is are they comparing it to the ADATA SX8200 Pro that was provided as an early reviewer sample or the lower spec version actually being sold retail?
    Reply