Identifying Corrected Atom Hardware
I have just purchased, but have not yet received, a brand new SuperMicro 5018A-FTN4 which has motherboard Super A1SRi-2758F. That motherboard has an Intel Atom Processor C2758 which I understand had an issue a couple of months back.
When the hardware arrives on Monday, how do I identify if the hardware is the latest version with whatever fixes are required?
I have found the this PDF from Intel https://www.intel.com/content/www/us/en/processors/atom/atom-c2000-family-spec-update.html, but it's not clear, to me at least, what I should really be looking for to make sure the hardware is the latest version.
Based on my understanding of the document, I'm looking for this code: R3GQ on the processor with the previous version being R1CW. Does that sound correct?
I have the same motherboard, I gave up because I could never find the info to even verify what I did find, does stepping 8 mean B0 or C0, and other things, hopefully you will have better luck
Where do you see stepping 8?
I just SSH in and run dmidecode -t processor
[2.4.2-RELEASE][admin@pfSense.localdomain]/root: dmidecode -t processor
Scanning /dev/mem for entry point.
SMBIOS 2.8 present.
Handle 0x0027, DMI type 4, 42 bytes
Socket Designation: CPU0
Type: Central Processor
Manufacturer: Intel(R) Corporation
ID: D8 06 04 00 FF FB EB BF
Signature: Type 0, Family 6, Model 77, Stepping 8
FPU (Floating-point unit on-chip)
VME (Virtual mode extension)
DE (Debugging extension)
PSE (Page size extension)
TSC (Time stamp counter)
MSR (Model specific registers)
PAE (Physical address extension)
MCE (Machine check exception)
CX8 (CMPXCHG8 instruction supported)
APIC (On-chip APIC hardware supported)
SEP (Fast system call)
MTRR (Memory type range registers)
PGE (Page global enable)
MCA (Machine check architecture)
CMOV (Conditional move instruction supported)
PAT (Page attribute table)
PSE-36 (36-bit page size extension)
CLFSH (CLFLUSH instruction supported)
DS (Debug store)
ACPI (ACPI supported)
MMX (MMX technology supported)
FXSR (FXSAVE and FXSTOR instructions supported)
SSE (Streaming SIMD extensions)
SSE2 (Streaming SIMD extensions 2)
TM (Thermal monitor supported)
PBE (Pending break enabled)
Version: Intel(R) Atom(TM) CPU C2758 @ 2.40GHz
Voltage: 1.6 V
External Clock: 100 MHz
Max Speed: 2600 MHz
Current Speed: 2400 MHz
Status: Populated, Enabled
L1 Cache Handle: 0x0025
L2 Cache Handle: 0x0026
L3 Cache Handle: Not Provided
Serial Number: Not Specified
Asset Tag: ProcessorInfo_ASSET_TAG
Part Number: Not Specified
Core Count: 8
Core Enabled: 8
Thread Count: 8
What really is the most aggravating thing is to pull the thing apart and just pop the heatsink off and just read it directly on the chip to see what s-spec it is and go from there, haven't figured out how to get that info from the command line yet, and that is what is frustrating. According to SuperMicro all motherboards after a certain month of 2017 (January or something, I can not recall) is suppose to have the fix. OK, but how do you know when your MB/Server was built, there is no serial number list that I can find, they make it difficult for sure. Bought mine in June 2017, but who knows when it was built. Figured by the time it shits the bed, C3000 support should be fully functional then in pfsense so not gonna worry about it so much.
Dunno…that pdf I reference doesn't say, as far as I can see, anything about a Stepping of "8". As you suggest they only reference B0 and C0. It seems like those are markings on the processor itself?
My machine will be here in a few days and it is brand new from SuperMicro, so I guess we can compare notes then.
BTW, how is the performance? Are you using gigabit WAN speeds by any chance?
No, not at gig WAN yet, but as performance goes, I see no problem as of yet. Only run 400mbit WAN, but I don't do anything other than basic stuff and pfBlockerNG, have no need for IDS/IPS, as this is for home use. Overkill for sure…
I went on and made a windows pe boot disk and used AIDA64 to get CPU info, here is my info, and it is a B0 stepping…
--------[ AIDA64 Extreme ]–--------------------------------------------------------------------------------------------
Version AIDA64 v5.95.4500
Benchmark Module 4.3.770-x64
Report Type Report Wizard
Operating System Microsoft Windows PE 10.0.15063.0 (Win10 RS2)
--------[ CPU ]–-------------------------------------------------------------------------------------------------------
CPU Type OctalCore Intel Atom C2758, 2400 MHz (24 x 100)
CPU Alias Rangeley
CPU Stepping B0
Instruction Set x86, x86-64, MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, AES
Original Clock 2400 MHz
Engineering Sample No
L1 Code Cache 32 KB per core
L1 Data Cache 24 KB per core
L2 Cache 4x 1 MB (On-Die, ECC, Full-Speed)
CPU Physical Info:
Package Type 1283 Ball FCBGA
Package Size 34 mm x 28 mm
Process Technology 22 nm, CMOS, Cu, High-K + Metal Gate
Die Size 106 mm2
Typical Power 20 W
Company Name Intel Corporation
Product Information http://ark.intel.com/search.aspx?q=Intel%20Atom%20C2758
Driver Update http://www.aida64.com/driver-updates
Motherboard ID A M I ALASKA
CPU #1 Intel(R) Atom(TM) CPU C2758 @ 2.40GHz, 2400 MHz
CPU #2 Intel(R) Atom(TM) CPU C2758 @ 2.40GHz, 2400 MHz
CPU #3 Intel(R) Atom(TM) CPU C2758 @ 2.40GHz, 2400 MHz
CPU #4 Intel(R) Atom(TM) CPU C2758 @ 2.40GHz, 2400 MHz
CPU #5 Intel(R) Atom(TM) CPU C2758 @ 2.40GHz, 2400 MHz
CPU #6 Intel(R) Atom(TM) CPU C2758 @ 2.40GHz, 2400 MHz
CPU #7 Intel(R) Atom(TM) CPU C2758 @ 2.40GHz, 2400 MHz
CPU #8 Intel(R) Atom(TM) CPU C2758 @ 2.40GHz, 2400 MHz
CPU #1 / Core #1 0%
CPU #1 / Core #2 0%
CPU #1 / Core #3 0%
CPU #1 / Core #4 0%
CPU #1 / Core #5 0%
CPU #1 / Core #6 0%
CPU #1 / Core #7 0%
CPU #1 / Core #8 0%
Note that this doesn't mean SuperMicro did not implement a "board fix", that info I just can not find, but they could of, who knows...
So yours is the previous rev then? So, what does that mean? What is actually fixed and what happens if you never get the "fix"?
Supposedly motherboard fix is just that, they do something to alleviate or remove the chance of the bug occurring. Likely has to be some sort of hardware fix I would guess, as that would be cheaper to implement than to replace/reflow motherboards and CPUs. I would think SuperMicro would know by the serial number if it has been fixed or not. In your case popping off the heatsink and checking the cpu s-spec for a C0 stepping, which would be easy to do before putting it online seeing you already have to be in it's insides. Mine is already in the rack so I am not going to worry but so much about it, hopefully it got the fix but if not, if it ever does puke, hopefully C3000 is fully baked in on pfSense by then and will just give me a reason to upgrade. And if the fix hasn't been done, guess it could bork and not restart when I go to reboot and burn down my house…
Have a look here
I have the same A1SRi-2758F motherboard in my pfSense box.
I contacted Supermicro support a couple weeks back, and they confirmed that my board needs the fix based on the serial number.