ASRock C2x50d4i boards now shipping with fixed CPU's

[qwertymodo]

I know that the ASRock C2x50d4i boards are pretty popular for NAS builds, but they have been plagued by the infamous AVR54 "18-month clock death" bug found throughout the entire C2000 line of CPU's. I've had 2 of these boards die on me, and despite hearing that they were planning to update their boards, I hadn't seen any confirmation that they had actually done so. Well, my latest RMA arrived today, and I can confirm that it contains an updated C0 stepping of the chip, which includes the AVR54 fix.




(Note the S-code SR3GR along the bottom edge, click for full-size image if you can't read it... or just take my word for it)

For more information, here's the official doc from Intel

AVR54 is listed as fixed in the C0 stepping on page 10 (Ctrl+F AVR54), and the S-codes for each processor are listed on the Component Identification table on page 16. It looks like the cpuid stepping flags may not have been updated for the C0 stepping, as they are reporting the same stepping code as the one listed for the B0 stepping, so pretty much the only 100% method for checking your CPU is to remove the heatsink, which is really easy on the ASRock boards (small die, and soldered to the motherboard, so it doesn't get stuck and you don't have to worry about ripping the chip out of a socket).

Just thought I'd post this here in case anybody was still wary of the C2000 boards. Other than the high death rate, they've been a solid NAS board for me. Just be sure you update the BIOS to avoid the other major cause-of-death with FreeNAS related to the watchdog timer event logging killing the BIOS flash.

 

[MrToddsFriends]

Edit: You can view your stepping version with CPU-Z

Is it possible for you to show a CPU-Z screenshot from a C0 Stepping Avoton instead of one made with some random Intel i5?

 

[qwertymodo]

That screenshot is from my desktop PC, I'm not running widows on my ASRock board. Sorry for the confusion. The Intel doc I linked indicates that the C2000 boards do provide that info via the CPUID registers, so you should be able to see it just fine. The chip photograph, however, is of the actual board I received, with the S-code visible, matching the S-code listed in the doc for a C0 stepping C2750.

 

[qwertymodo]

Hmm... strange. The CLI cpuid utility on FreeNAS is reporting the same stepping code (0x08) as the B0 stepping. Looking at the Intel doc (table 5, page 12), it shows that both A0/A1 report the same 0x00 code, B0 reports 0x08, and there is no code listed for C0. So, maybe they didn't update the stepping code in the cpuid registers like A0/A1, which is unfortunate. Your best bet might just be to pull the heatsink and verify the printed S-code.

 

[MrToddsFriends]

Your best bet might just be to pull the heatsink and verify the printed S-code.

After reading the C2000 Specification Update it was my impression, too, that the printed S-code is the only feature to differentiate between B0 and C0 stepping Avotons (C0 being not affected by the AVR54 bug). I didn't remove the heatsink on my replaced Supermicro A1SAi-2750F to avoid loosing warranty. Thanks for the die photo.