Part because of a mistake and part for fun, I purchased an ASRock J3455-ITX. I tried to boot a couple different OSs without much luck, I could not get CoreOS to boot or even a Gentoo LiveCD. Granted, it is low power and slow, so it might have been my impatience. Finally, I tried a Gentoo minimal install DVD and it booted, so in the spirit of making something easy much harder this would be yet another Gentoo build.
System make up.
The system make up by the time it was done:
- ASRock J3455-ITX Intel Quad-Core Processor J3455 (up to 2.3GHz) Mini ITX Motherboard/CPU Combo (Newegg)
- Silverstone 0.8mm Steel Body Tek Acrylic Front Panel for Mini-ITX Media Center/HTPC Case Cases ML05B (Amazon)
- SilverStone Technology 300W SFX Form Factor 80 PLUS BRONZE Power Supply with +12V single rail, Active PFC (ST30SF) (Amazon)
Timetec Hynix IC 16GB Kit (2x8GB) DDR3L 1866MHz PC3-14900 Unbuffered Non-ECC 1.35V CL13 2Rx8 Dual Rank 204 Pin SODIMM Apple Memory RAM Module Upgrade (16GB Kit (2x8GB)) (Amazon)Returning to Amazon
- G.SKILL 8GB (2 x 4GB) 204-Pin DDR3 SO-DIMM DDR3 1333 (PC3 10666) Laptop Memory Model F3-10666CL9D-8GBSQ (Newegg)
- SAMSUNG 850 EVO 2.5″ 250GB SATA III 3D NAND Internal Solid State Drive (SSD) MZ-75E250B/AM (Had this from an old system)
BIOS Version J3455-ITX, BIOS P1.60 01/16/2018 – Out of the box BIOS version.
Here is the catch, on the ASRock site it lists 1.40 then jumps to 1.70 as a bridge before installing 1.80. I tried the 1.70 from a fat32 USB and the board would not recognize the firmware. So screw it, I just went to 1.80 and it worked. I have to assume the 1.60 is already a bridge to 1.80.
Probably will not put a fan in because the passive cooling seems to work well. There are two fan connections on the board (CPU and chassis) and settings in the board configuration. Compiling webkit, which is a grind, the temps 30+ minutes into compiling webkit:
Adapter: ISA adapter
Package id 0: +131.0°F (high = +221.0°F, crit = +221.0°F)
Core 0: +131.0°F (high = +221.0°F, crit = +221.0°F)
Core 1: +131.0°F (high = +221.0°F, crit = +221.0°F)
Core 2: +129.2°F (high = +221.0°F, crit = +221.0°F)
Core 3: +129.2°F (high = +221.0°F, crit = +221.0°F)
Adapter: Virtual device
temp1: +132.8°F (crit = +212.0°F)
This is a real BS point. ASRock and a number of other board manufacturers advertise 16 Gig of RAM usable. I read all over the internet this thing can run with 16 Gig. I gave it a try and everything SEEMED to work.
I gave it a shot (see above in System makeup) using 16 Gig (8×2) DDR3L 1866 RAM. Like I already said, it seemed to work, but under heavy loads I got random kernel panics (Kernel 4.18.20), usually SMP NOPTI and sometimes even more spectacular blow ups. I was thinking bad RAM module? Anyway, I pulled a module and ran 8 Gig with the result that I cannot kill it with any load. A quick check of the Intel J3455 Datasheet says only an 8 Gig MAX:
In ASRock’s Memory Support List, only DDR3 Crucial 1600 16GB CT204864BF160B is listed for this configuration. So, in conclusion it might run with 16 Gig, but not always. I do not care, I really only need 8Gig and can return the other RAM to Amazon, saving some money.
EDIT: I have since read that the board might not be automatically powering the RAM or the RAM is not quite right. A possible fix might be to take the board out if AUTO and increasing the voltage or lowering the speed. I have not tried this …
This is my real hard spot with this board:
mce: [Hardware Error]: CPU 0: Machine Check: 0 Bank 4: e600000000020408
mce: [Hardware Error]: TSC 0 ADDR fef13b80
mce: [Hardware Error]: PROCESSOR 0:506c9 TIME 1543272637 SOCKET 0 APIC 0 microcode 32
This seems to be common on all the ASRock boards and a lengthy Redhat (here) post gets to the source. The most interesting:
Tony Luck 2017-08-02 13:15:00 EDT
I wrote a trivial EFI program to dump the machine check banks and put it on a USB stick a \EFI\BOOT\BOOTX64.EFI, then booted that USB stick.
The machine check was already logged at this point.
So Linux (and grub boot loader) aren’t involved. Tracking with someone who supports BIOS on this platform now.
So, this is some kind of issue with the BIOS and during boot it picks up on the already existing error. Intel NUC Kits also had this problem, but fixed it in the BIOS. It seems to be a problem that probably will not go away, but does not cause issues.