First look at the new Intel Xeon D-1541 Supermicro SuperServer SYS-5028D-TN4T

Posted by Paul Braren on Feb 7 2016 (updated on Mar 3 2016) in
  • ESXi
  • Virtualization
  • HomeLab
  • HomeServer
  • Windows
  • Supermicro has a reputation for stable servers, thankfully devoid of BIOS bells-and-whistles that could adversely affect system stability. My last 7 months of Supermicro SYS-5028D-TN4T mini-tower ownership have held up to those expectations. Had my hands on 3 other identical loaner SuperServers for a few weeks back in October and November, those tests also worked out well. Even with the initial BIOS 1.0a, turns out this little guy allowed boot-from-NVMe, but I suspect most folks have moved to 1.0b by now, done for you in certain bundles, where up to 128GB of added DDR4 RAM is also pre-tested for you.*

    Beginning the week of Monday, February 22 2016, placing your order for one of these bundles will likely get you one of these newer Xeon D-1541 CPUs.
    [Feb 09 2016 Update - These estimated availability dates are not yet confirmed, see update below.]

    So today, you're getting a TinkerTry exclusive sneak preview of what's in store this month. I strongly suspect many blogs will also have similar information available very soon. Why? Because the popularity of this rising-star in the Xeon line continues to grow.

    It is with great joy that I present to you these first-in-world looks at the long anticipated arrival of the Intel Xeon D-1541. Yeah, it's an admittedly modest speed boost of perhaps 5%, but there is more to discuss. My initial impression is relief, for the following reasons:

    Here's a view of the factory's internal cable-routing. Note the new finger grip/pull tab on the air baffle, where it's attached to the underside of the drive cage. Installing that Samsung 950 PRO took about 45 sec.
    • this also means the process of retesting OSs is negligible, especially if it turns out that we can use this same BIOS 1.0c on our existing Xeon D-1540 systems (that cannot be assumed), note that SR-IOV won't likely ever show up on the Xeon D-1540, even with 1.0c

    • it's only been 7 months since the Xeon D-1540 first arrived, yet folks like me who proudly own that first revision aren't likely to be kicking ourselves for not waiting, since it's such an incremental change

    • the hardware appears identical in all significant ways, even the same CPU fan part number

    • the only changes I've noticed so far are:

      • the new finger pull tab on the side of the internal air baffle to make removal and re-install easier (pictured above)
      • a new serial # sticker arrangement on the rear
      • a new MAC address vendor ID/range
      • new number on CPU heat sink
    • the most significant change in function is likely SR-IOV, unlikely to affect home lab folks, but some folks putting these into production may care, discussed at some length here:

    I will have more information coming up a TinkerTry soon, including SR-IOV tests using both:

    • VMware ESXi 6.0 Update 1b
    • Hyper-V / Windows Server 2016 Technical Preview 4
    Samsung 950 PRO inside, FLIR image outside.

    I also plan to have a look at the thermal characteristics inside the PC when it's booted at factory default settings, see also:

    What I have for you today is some barely edited lab footage of my first looks at the new system powering up for the first time, just hours after the FedEx box arrived right after a decent snowstorm. In this video, I capture all differences in my systems 1.0b BIOS, versus the new system's 1.0c BIOS, along with taking a brief look at CPU-Z and ATTO on Windows Server 2016 Technical Preview 4.


    Supermicro Xeon D-1540/1541 BIOS 1.0b/1.0c BIOS differences. Boot-from-NVMe, CPU-Z, ATTO tests.
    Supermicro SYS-5028D-TN4T Xeon D-1540 running Windows Server 2016 Tech Preview 4 off NVMe
    Supermicro SYS-5028D-TN4T Xeon D-1541 running Windows Server 2016 Tech Preview 4 off NVMe
    At left, Xeon D-1540 is powered off, at right, Xeon D-1541 CPU 16 threads at 100% out, courtesy of Prime95, run at default balanced setting.

    Note that over the past few weeks, my brief tests of the boot-from-NVMe have gone quite well using Windows 10, Windows Server 2016 Technical Preview 4, SuSE 12, and Red Hat 7.2. Therefore, I won't be doing extensive re-testing, since those results are likely to be functionally identical anyway.

    My own use-case for my SuperServer in my home lab is for full time VMware ESXi 6.0 Update 1a virtualization duties. I admittedly only briefly tinker with things like Intel RSTe RAID, which has never been compatible with VMware anyway.

    I hope to have more information on the 12 and 16 core versions coming up soon. Of course, those even more impressive CPUs may turn out to put the price way above what many folks can justify. We'll have to wait and see. Remember, Skylake only has 4 cores/8 threads, the Xeon D-1540/1541 has 8 cores/16 threads.

    Prime95 keeps things interesting.

    The current Xeon D-1540/1541 CPUs share the same 45 watt rating. While the exact ship dates are not yet known, soon we should have a 12 core variant that may be called Xeon D-1557 or Xeon D-1567. It should works in this same exact system, since the motherboard is the same size, and the same 45 watts, explained here:

    The 16 core model will likely be called the Xeon D-1577, and will have a 65 watt rating, and still shares the same Mini-ITX 6.7" x 6.7" form-factor.

    Feb 07 2016 Update 11:08 PM

    I just noticed a slightly different part on the motherboard. In the left picture of the Xeon D-1540, sandwiched in between the CPU heat sink, and the tall aluminum VRM heat fins behind it. Notice the visible silver-colored leg on the small black component called the SMD (Surface-Mounted-Device) power inductor, with no such leg seen next to the Xeon D-1541 pictured at right.

    Addition details on the CPUs:

    • the Xeon D-1540 heat sink is labeled SNK-C0057A4L 10110215
    • the Xeon D-1541 heat sink is labeled SNK-C0057A4L 10113415, in a new font
    • the difference between the two is 3,200, which likely means very little

    Because of the limited time I have with this loaner system, and because a mere 5% boost in CPU speeds is barely statistically significant in benchmarks, my time will be more focused on things like SR-IOV, RST, and vSAN testing. To prepare, my home lab is now equipped with my 2nd Samsung SSD 950 PRO M.2 NVMe drive, to make for some very interesting testing possibilities. Stay tuned!

    Feb 08 2016 Update

    Curiosity got the better of me, had to know what's up with the slight differences in the PCB design. Didn't really want to remove the motherboard entirely to get a better view, especially on my own system. I did manage to get a smartphone with a macro lense attachment in there, getting a shot of them innards deep inside both systems. Tricky to get decent lighting and focus, but the photos are good enough to easily determine that the PCB been redesigned, for whatever reason Supermicro felt like doing so. Given this is a company that doesn't even do BIOS release notes, I doubt we'll really know why exactly, but it's a pretty common in the industry. As long as there are no functional differences, it likely doesn't really matter.

    It's quite possible folks who bought their system recently already have this Rev. 2.00 in their Xeon D-1540 system. Will be interesting to see if the future 12 and 16 core Xeon D systems ship with an even later PCB revision.

    Looking forward to using my FLIR ONE inside as well, recording video during boot-up. That's for another day, and another article.

    Here's the space near the center of the motherboard, right next to the CPU, it's a Rev: 1.02 PCB.
    Yep, it's clearly Rev. 2.00 PCB.

    Anybody got a Xeon D-1540 motherboard that has a Rev. 2.00 PCB like this Xeon D-1540? I'm curious, drop a comment below for us. It's easy to determine which you have at a glance:

    • remove the PC cover
    • look toward the CPU from the side of the case with the SATA cabling
    Rev 1.02 PCB at left, Rev. 2.0 PCB at right.

    Rev. 1.02's SMD Power Inductor right between the CPU and VRM is the VITEC PR72-221 1435S, pictured below.

    Rev. 1.02's SMD Power Inductor near VRM heat sink is the VITEC PR72-221 1435S.
    Rev. 2.00's SMD Power Inductor near the VRM heat sink is a bit of a mystery, only labeled as R15 545 CDL.

    Feb 09 2016 Update

    I have now been informed by Supermicro that my PCB Rev. 2.00 may be pre-release, and that BIOS 1.0c is not yet finalized. I noticed I couldn't get into Ctrl+I (Intel RSTe RAID), this could be why. General availability of SYS-5028D-TN4T with Xeon D-1541 is also not yet solid, and as any IT pro knows, release dates can change for any number of reasons.

    There's a lot of content and observations in the long video I posted above. In case you haven't gone through it, here's the firmware screens from the IPMI web UI, pictured below. Since the rest of the BIOS menus may still be undergoing changes, I'm not likely to go through and screenshot all the differences.

    Xeon D-1540 GA levels, BIOS 1.0b, and IPMI 02.14.
    Xeon D-1541 Pre-release levels of BIOS 1.0c and IPMI 03.26, and Redfish 1.0.0.

    Wondering what Redfish is? See:

    Feb 10 2016 Update

    Not really suprising a new motherboard revision might be needed to support both DDR3 and DDR4, see differences between the Xeon D-1540/1541 at Intel ARK:

    ARK Menu - Compare Intel Products:

    Feb 12 2016 Update

    Intel Rapid Storage Technology enterprise

    New RSTe RAID menus spotted in 1.0c, which I managed to gloss over in the video.

    New Intel RSTe BIOS UI replaces Ctrl+I for all BIOS modes [Legacy/Dual/UEFI].
    Create RAID Volume. ________________________________________________________
    Select Disks. ________________________________________________________

    Note what the BIOS calls it, "Intel RSTe SATA Controller." Thus, any hopes of RAIDing two M.2 NVMe drives together are dashed, since they're not SATA.

    What if you took two AHCI M.2 drives, like the SM951. Then you get yourself a Lycom DT-120 pass through adapter for $20ish, to get yourself that 2nd M.2 socket. Then you might be able to RAID your M.2, but I no longer own any SM951 drives, so I'm currently unable to test this out. See also:

    Feb 13 2016 Update

    SR-IOV Works! See it at TinkerTry:

    Mar 03 2016 Update

    Doh! It's super easy to determine at a glance whether you have a PCB that's Rev: 1.02 or Rev. 2.00, with needing to peer in to see what's near the CPU. Instead, just look at the white fan headers at the front corner of the motherboard, ahead of the PCI slot:

    • One fan header means you have PCB Rev. 1.01 (Xeon D-1540)
    • Two fan headers mean you have PCB Rev. 2.00 (Xeon D-1541)

    See also at TinkerTry

    If you wait until the week of Monday February 22nd to place your order from Wiredzone, you are likely to receive your system with the new Xeon D-1541 on board, and there's 30 days to request a return. Here's all the ordering details:

    See also


    *Disclosure, LLC is an independent site, has no sponsored posts, and all ads are run through 3rd party BuySellAds. All equipment and software is purchased for long-term productive use, and any rare exceptions are noted. The Xeon D-1541 equipped SuperServer featured in this article is on temporary loan from Supermicro, equipped with 128GB of Samsung RAM and two 128GB Intel DC S3510 SSDs.

    TinkerTry's relationship with Wiredzone is similar to the Amazon Associates program, where a very modest commission is earned from each referral sale from TinkerTry's SuperServer order page. I chose this trusted authorized reseller for its low cost and customer service, and a mutual desire to help folks worldwide, including a new way to reduce EU shipping costs. Why? Such commissions help reduce TinkerTry's reliance on advertisers, while building a community around the Xeon D-1500 chipset that strikes a great balance between efficiency and capability.

    I personally traveled to Wiredzone near Miami FL to see the assembly room first-hand, and to Supermicro HQ in San Jose CA to share ideas and give direct product feedback.

    I'm a full time IT Pro for the past 23 years. I've worked with IBM, HP, Dell, and Lenovo servers for hands-on implementation work across the US. Working from home lately, I'm quite enjoying finally owning a lower-cost Supermicro solution that I can recommend to IT Pro colleagues, knowing it will "just work." That's right, no tinkering required.