Rainbow Scenery
Valid HTML 4.01 Transitional

Intel NUC5i5RYH
Setting Up Iris

Jim Carter, 2016-01-18, upated 2018-01-16


In the Box

Check the box contents and record serial numbers. Try to get the ethernet MAC address so it can be registered with the firewall.

Beware: This machine has no memory, no disc and no operating system. You will need to obtain your operating system from another source. This of course is simple, inexpensive and legal for Linux.

What's in the box:

Wonder of wonders, the MAC address (and the serial number) is on a sticker on the bottom of the machine.

The power supply has interchangeable pins. None of them have a grounding pin. For the type letters see the IEC World Plugs website. These pin assemblies are provided:

Assembly Steps

Quirks when Booting

When you turn on power, the machine may send video on HDMI, or VGA, or both. It seems random. This means that you may not see the splash screen, particularly since the NUC5i5RYH doesn't have a VGA connector (the NUC6CAYH does). When you are doing a special boot procedure, like setup or booting from USB, when you see the Numlock light on the keyboard and no splash screen, hit your function key anyway, perhaps several times. If it doesn't show you video, press power off and try again.

Booting from USB storage: Get into BIOS setup by pressing F2 during boot. On the left side of the start screen is the Boot Order panel. Use your USB mouse; shift from the UEFI to Legacy tab. Double click on whichever device, to boot from it immediately. Alternatively, press F10 during boot, scroll to the desired device, and press enter.

In a normal BIOS there is a line item for generic USB devices and another for PXE boot, and you can adjust their boot order in the usual way. In the NUC's BIOS there may be checkboxes for Boot USB First or Boot PXE Last. But not on the NUC5i5RYH. If you configure the order of the specific USB stick, then remove it, it is forgotten. You will have to use the F10 menu to boot from it next time.

BIOS Update

How to obtain the latest BIOS version:

Now that I have a flasher in the form of a MS-DOS EXE file, what am I going to do with it?

Update history:

2015-08-12: BIOS 0350 (2015-08-12)

This is the BIOS that the machine was delivered with, and its date. Not too badly back version.

2016-01-27: BIOS 0353 (2015-12-20)

When received, the machine was updated to this version, the latest.

2018-01-13: BIOS 0368 (2017-12-20)

When the NUC5i5RYH was converted to the Baobei role (Windows), I updated to the latest BIOS. It is believed to have a patch partially mitigating the Meltdown (CVE-2017-5754) and Spectre (CVE-2017-5753, CVE-2017-5715) exploits.

2018-01-31: BIOS 0367 (2017-09-29)

Oopsie! On two occasions (and another with 0368) the machine appeared to boot normally, but after about 10 minutes of operation Windows froze and BIOS showed an error message. Reverting to 0367 factory defaults (F9 in Setup) was helpful. See below for investigation of this issue.

Lots of users are reporting weird reboots on a variety of Intel CPUs after updating the BIOS to versions with Meltodown mitigation. Intel has withdrawn BIOS 0368; it's no longer on the download site. (And similarly for the NUC6CAYH.) I downloaded and installed BIOS 0367, but the freezeup continued. I tried to downgrade further, but 0366 and earlier are rejected as being incompatible. Forum posters attribute it to a security fix to the Intel Management Engine (that allows SSH into it without a password), and it will not downgrade to a version lacking that fix.

Meltdown/Spectre mitigation and BIOS horkage: Here's a good guide to the present very complicated status of Meltdown/Spectre mitigation: A Clear Guide to Meltdown and Spectre Patches by Jonathan Crowe (latest update 2018-01-29).

After BIOS 0358 was installed, and also after it was reverted to 0357, I had three mysterious crashes. The machine booted normally and was in normal use for about 10 minutes (measured 12 mins, the third time). The machine froze without warning, and the BIOS showed this message:

BIOS has detected unsuccessful POST attempt(s).
Possible causes include recent changes to BIOS
Performance Options or recent hardware change.
Press 'Y' to enter Setup or 'N' to cancel and attempt
to boot with previous settings.

BIOS flashing had no error messages at the time, nor were there any POST error reports. The only thing in the event log was repeated BIOS flashing failures, which I attribute to trying to get various incompatible BIOS versions onto the machine; see above for the reason. The consensus of observations appears to be that when it boots from S5 (power off), it always gets this result after about 10 minutes. If you then take the 'N' choice to reboot, it will continue for a considerable period; the longest so far was 5 hours, and no freezeup occurred.

Google searches yield very little information.

Ben suggested trying to revert to factory defaults. When you re-flash BIOS it sets some but not all settings to default, and perhaps there is something from 0368, that may not even be visible in the BIOS GUI, that is poisonous and that 0367 does not revert. OK, I hit F9 to revert. No freezeup yet: cross fingers…

For the record, here is how BIOS was set, after the third freezeup. Most of these are default values; in fact at this point I didn't change anyhing.

BIOS Setup

Power and Speed Measurements

Memory test using memtest86+ comparing just stick 'B' in the bottom socket, labelled 'A' in the BIOS, vs. both sockets populated.

Speed test on Jimc's benchmark. Columns in the output:

The test was done first with one memory slot occupied, then with both. It was run 3 times in each condition and the last one is reported. Actually the test is designed to be reasonably immune to buffer cache effects and scores vary only about 3% between repetitions. Numbers are in kbytes/sec.

Slots SHA512 SHA512*cores Disc read Composite
1 132815 265630 8352 87185
2 132504 265008 8379 87004

I'm surprised to find that populating the second socket has no effect at all on speed.

Comparing with my other machines, the NUC with Core i5-5250U@1.6GHz (overclocked to 2.7GHz in Turbo mode) falls between the i7-3517UE @1.7GHz (Diamond) and the i7-3632QM @2.2GHz (Xena), making it the second fastest machine in the house, and clearly the winner in work units per joule.

On another task which does checksums on a big block of memory, where the memory fits in the L3 cache, comparing 1 vs. 2 simultaneous processes, the one core test finished about 8% faster. With one core the CPU governor set a clock rate of 2.7GHz. With both, it used 2.5GHz, 8% less. The nominal clock rate of this CPU is 1.6GHz, but Turbo overclocking can be configured in the BIOS (see above). Likely a much longer running task would warm up the CPU and the clock rate would be turned down more.

Power measurements, in watts, with a Kill-a-Watt meter. Most of these measurements are without the Hauppauge HTV-950q; it adds 2 watts.

Wakeup behavior, indicating what was used to wake the machine:

I have a wireless keyboard which communicates by Bluetooth to a dongle that connects to a USB port and presents the payload as a HID keyboard and mouse, so it can be used before the OS is running. Sometimes it gets into a strange state and sends a lot of keycodes for no reason, making the host wake as soon as it goes into S3 state. But for this sequence of tests the dongle was well behaved and the host could sleep.

Brain Transplants

The present assignment of machines and roles is:

Fit-pc3 Pro Iris Video playback Works well.
Zbox AD03BR Kermit Video recording Works, but 12W idle
Zino 400HD Aurora Video playback Works
Zino 400 Piki Audio playback Works, but can't sleep

I will move the discs of Iris, Aurora and Piki to get this configuration:

NUC5i5RYH Iris Video record+playback New macine
Fit-pc3 Pro Aurora Video playback Formerly Iris
Zbox AD03BR Kermit Audio playback No transplant
Zino 400HD Piki Hot spare Formerly Aurora
Zino 400 -- On medical leave No disc

I will do the transplants in this order:

Here are details of the procedure to swap discs and IPv6 addresses.

Revised brain transplant plan:

Zino 400HD


Initially the NUC was going to receive directly the disc out of Iris, but that plan had to be revised. I resurrected a 250Gb laptop drive and called it Orion.

Issues with NUC5i5RYH as Iris:

FitPC Pro

Issues with FitPC Pro as Aurora:

Zino 400

Zino 400HD again

Issues with Zino 400HD as Piki:

Zbox AD03BR

Installing OpenSuSE 42.1

Initially I intended to do a bunch of brain transplants, and the NUC5i5RYH would never need a fresh installation of the operating system. However it turned out that I needed to set up a laptop drive, in the NUC, that would eventually host Aurora and which would be transplanted into the Fit-PC Pro.

For MythTV, create a directory /s1/video and a link to it from /video. [Done]

USB Bus Structure on NUC5i5RYH

Unlike on the Fit-PC Pro and LP, the NUC has only one USB bus and all of the ports connect to it. This bus is provided with a XHCI (USB-3.0) and a EHCI (USB-2.0) controller.

Rainbow Scenery