Skip to main content

Intel Optane SSD DC P5800X Review: The Fastest SSD Ever Made

The fastest just got faster

Intel Optane SSD DC P5800X
Editor's Choice
(Image: © Tom's Hardware)

Intel’s initial goal with the DC P5800X was to double the performance of its predecessors, and in our testing, we can see it delivered. The P5800X is an incredible NVMe SSD that leaves everything else in the dust — flash-based SSDs simply can't compete. Due to the inherent advantages of the Optane storage media, like the ability to write data in place, Intel’s Optane SSD DC P5800X sets the new high bar for PCIe 4.0 performance. 

Intel Optane SSD DC P5800X

(Image credit: Tom's Hardware)

Based on what we’ve seen, we can say that the DC P5800X is the fastest storage device we've ever tested. The P5800X's increased bandwidth feeds voracious content creation applications with aplomb. As they say in the storage industry, bandwidth is easy, but latency is hard. The P5800X wins here, too, with its average latency results showing remarkable improvement in low-QD operations and providing exceptional agility and responsiveness in nearly any type of work. Even Intel's previous-gen Optane SSD 905P, which was previously the fastest storage device in the world, pales in comparison. 

The P5800X is also the most resilient SSD we’ve come across, too — nothing else is even close. With up to 100 drive writes of endurance per day, it can handle more write data than your average user will do within a lifetime. 

Given its eye-watering price point, the DC P5800X isn’t a value-oriented drive, but it is the most responsive and durable storage money can buy. If you have heavy workloads that saturate the PCIe 3.0 interface and command PCIe 4.0 speeds, the Optane DC P5800X is your best choice. With speeds of up to 7.2/6.2 GBps read/write and capable of handling up to 1.5 million random read/write IOPS, this SSD gives you the fastest performance at any cost. 

MORE: Best SSDs

MORE: How We Test HDDs And SSDs

MORE: All SSD Content

  • mwestall
    Hmmm. Hello Chia, meet Optane.
    Reply
  • JayNor
    "... with bit-addressable Optane media."

    byte addressable
    Reply
  • domih
    Which is the most upsetting the Apple or the INTEL tax?

    Even for a PC geek like me $3,724 for 1.6TB is not an option :oops:

    For the low-life we are, 2 x Samsung | Corsair | WB in RAID 0 for data and a separate 3rd one for OS bring enough happiness :D

    Optane is definitely not for the populace.
    Reply
  • mdd1963
    $3+K for 1.6 TB?

    (Yeah, that's happening!)
    Reply
  • derekullo
    7.2 gigabytes a second is equal to the slowest DDR3 speed I could find DDR3-800, 6.4 gigabytes per second.

    Does anything weird happen when your storage is faster than your ram?
    Reply
  • domih
    derekullo said:
    Does anything weird happen when your storage is faster than your ram?

    Nothing weird like unnatural, your RAM (if involved) just becomes the bottleneck.

    The same "weirdness" occurs when you try to run 25, 40, 50 or 100 GbE networking on a low-end PC: the PC system itself (buses, memory, CPU, disks) becomes the bottleneck in comparison to the network throughput and you end up with iperf3 returning something like 15 Gbits/sec because that's the max the PC hardware can do.

    I believe these $$$ Optane are good for workstations (e.g. Threadripper) or Servers: matching performance for matching cost. For regular desktops, a "normal" RAID 0 of SSD is preferable, again in terms of matching performance for matching cost.

    PCIe 5 with CXL is going to redefine the bottlenecks over the next few years. But again this is for workstations and servers. It will take several years to see something trickle down to "normal" desktops. Necessity being the mother of invention, one will also have to "invent" desktop applications that would significant leverage PCIe 5 with CXL. Gaming is not part of these applications.
    Reply
  • Dsplover
    I need a low latency/high random device like this.
    I’ll just have to settle for 400GB’s though.
    Reply
  • supremelaw
    We implement a policy of formatting C: up to a current maximum of 100GB.
    When there is unallocated space after formatting C: , we format the remainder
    as a dedicated data partition. This policy has worked out very well for decades.

    As such, I suggest that Intel expand the appeal of this major Optane improvement
    and offer one 128GB and one 256GB PCIe 4.0 U.2 Optane directed to the
    workstation and high-performance desktop marketplaces.

    And, Intel should be open about its commitment to ramp the speed of these two
    smaller drives up to PCIe 5.0, as soon as that standard starts to roll out.

    There is much to recommend the simplicity of a single OS drive,
    particularly when doing a fresh installation of any modern OS.

    This will allow a significant reduction in the MSRP, and the greater demand
    should be very "price-elastic".

    (A widget is price-elastic when a 5% reduction in price
    increases market demand by 10% or more, for example.)

    In simple English, the greater demand these smaller drives will generate,
    should drive prices downward over time. That trend has been true
    historically of many electronic devices e.g. DVD writers.

    The lower unit prices will also inspire Prosumers to experiment
    with RAID arrays of these newer U.2 Optanes.
    Reply
  • watzupken
    supremelaw said:
    We implement a policy of formatting C: up to a current maximum of 100GB.
    When there is unallocated space after formatting C: , we format the remainder
    as a dedicated data partition. This policy has worked out very well for decades.

    As such, I suggest that Intel expand the appeal of this major Optane improvement
    and offer one 128GB and one 256GB PCIe 4.0 U.2 Optane directed to the
    workstation and high-performance desktop marketplaces.

    And, Intel should be open about its commitment to ramp the speed of these two
    smaller drives up to PCIe 5.0, as soon as that standard starts to roll out.

    There is much to recommend the simplicity of a single OS drive,
    particularly when doing a fresh installation of any modern OS.

    This will allow a significant reduction in the MSRP, and the greater demand
    should be very "price-elastic".

    (A widget is price-elastic when a 5% reduction in price
    increases market demand by 10% or more, for example.)

    In simple English, the greater demand these smaller drives will generate,
    should drive prices downward over time. That trend has been true
    historically of many electronic devices e.g. DVD writers.

    The lower unit prices will also inspire Prosumers to experiment
    with RAID arrays of these newer U.2 Optanes.
    I feel they did that at the start, but not gaining much traction in terms of demand. Which is why they pulled Optane from the consumer market. The problem is the price, limited storage options, and high power consumption takes away interest. Now all you find are some left over 16/ 32GB Optane drives more for caching, and they may cost close to some 250/ 256GB drive.
    Reply
  • 2Be_or_Not2Be
    Sadly, Intel didn't buy the Lehi, UT fab that produces these amazing Optane drives from Micron. Micron, instead of creating a new line (Optane 2nd gen) like Intel did, is selling the fab to TI - who isn't producing Optane. Neither Intel nor Micron apparently wanted to stay in the Optane business. It should have been sold to SK Hynix, TSMC, Samsung, or some other mfg that would work on not only continuing the R&D to make better successive generations, but also bringing the price down. I guess Micron sold Intel's production contract to TI also, so Intel will get the last of the Optane drives. But what a waste of some great performance tech!
    Reply