Phison E28 2TB SSD Review: A return for vengeance

Phison is back to settle the PCIe 5.0 score.

Phison E28 2TB SSD
Editor's Choice
(Image credit: © Tom's Hardware)

Tom's Hardware Verdict

The Phison E28 is the best Gen 5 SSD to date, with excellent performance and power efficiency in an all-around dynamite package.

Pros

  • +

    Excellent performance

  • +

    Record-setting power efficiency

  • +

    Good random 4K latencies

Cons

  • -

    Mediocre sustained performance

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.

The Sandisk WD Black SN8100 recently sent shockwaves through the enthusiast SSD space with its exceptional performance and power efficiency, cementing SMI’s platform as a serious competitor for the coveted Best SSD crown. Phison is not one to rest on its prior laurels, though, and has worked diligently to one-up the competition with a class-defining storage solution of its own. Today we see the results with the E26's proper successor, the E28 SSD reference design.

If there is one thing that's held back adoption of high-end PCIe 5.0 SSDs, it has been lackluster power efficiency. The first wave of these SSDs required a heatsink for proper operation and had a power draw that made things tricky on any platform, considering idle power consumption was also high. The SMI SM2508 and Phison E31T brought the first signs last year that this would eventually be solved, but it was the Micron 4600 and, finally, the WD Black SN8100, that really took this message home. Now you could have your cake and eat it, too.

Latest Videos FromTom's Hardware
Shane Downing
Contributing Writer

Shane Downing is a Contributing Writer for Tom’s Hardware, covering consumer storage, PC hardware, and AI.

  • james2037
    Terrific review. But the table has an error in the last column! ;)
    Reply
  • John Freiman
    So I'm going to assume that this drive as a dedicated SSD for a Plex Server and it's data hungry needs you fast random reads would make this a fabulous solution...
    The faster the server can load metadata for clients the better. Writes would be secondary since I don't believe they are time sensitive.
    **All of this assumes that your transcode cache is going to a RAM disc.
    I'm in! Sign my up for a 4TB and 8TB!
    Reply
  • Notton
    What is the reasoning behind disabling ASPM in Quarch Idle bench?
    Isn't that what gives extremely low power consumption while idle?
    Reply
  • Stomx
    How much costs Phison E28 in 1000s quantities vs E26 ?
    Reply
  • JarredWaltonGPU
    Notton said:
    What is the reasoning behind disabling ASPM in Quarch Idle bench?
    Isn't that what gives extremely low power consumption while idle?
    For desktop use, ASPM isn't critical. It can reduce power use, but only by up to 1~2 watts in most cases (4W maybe on a "bad" drive with little to no power management), and it can create some slight additional latency at times. If you enable ASPM, in theory it just allows all SSDs to be put in a deep sleep state where they're waiting for the command to wake up, which means most will drop to ~50 mW of power use.

    By leaving ASPM off, we get to see the "still powered on while idle" power use, which is where an SSD will sit if it's getting periodic (like once every 5~60 seconds) accesses. In such a scenario, as I understand it, the SSD won't actually enter deep sleep states and thus continues to draw its idle power. On a desktop, this is still generally not a big deal, but in a laptop, if your SSD keeps getting accesses that limit the deep sleep time, it can end up drawing 1~2 watts of power for a few minutes before going back into deep sleep mode — repeated every time there's an SSD access. That can have a real-world impact on battery life.

    On the Steam Deck, where I've previously done battery life testing with ASPM enabled (because AFAIK you can't disable it), basically every SSD ended up providing the "same" battery life because the drives would just enter deep sleep mode. Over time, changes to SteamOS improved battery life a bit, but without running a battery test that was actively using the SSD on occasion, everything looks identical. In practice, deep sleep modes will basically have the same end-user experience (even a 25 mW vs 100 mW difference just won't really matter, despite the latter being 4X higher), so looking more at the idle but not deep sleep power use makes sense.

    In terms of power use, I'd consider efficiency the most important metric, then average power, then idle power, and finally the max power is only tangentially interesting (depending on how often a drive comes near that level). For laptops, idle power does become a bit more important, but the real consideration is how much power a drive needs on average while in active use, because that impacts cooling considerations. Drives that average over 5W of power when active can run very hot if placed in a laptop, and getting average power use to 4W or less is generally the goal.
    Reply