Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Sign In Register

Quick Links


Categories

In this Discussion

› › Intel Atom C2000 series dying at 18 months due to fault - Page 2
New on LowEndTalk? Please Register and read our Community Rules.

All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.

Intel Atom C2000 series dying at 18 months due to fault

2

Comments

  • Vova1234Vova1234 Member, Patron Provider
    edited February 2017

    http://not-install.hostsuki.org/2017/01/atom-c2750.html

    62.210.106.150
    Datacenter: DC2, Room: 101, Zone: Salle 101, Line: M, Rack: 15, Block: D, Position: 21
    Atom / 8 / 160 ssd
    Intel Atom C2750 2.40GHz
    SSDSC2BB160G4T - 1.2 year work

    62.210.106.154
    Datacenter: DC2, Room: 101, Zone: Salle 101, Line: M, Rack: 15, Block: E, Position: 1
    Atom / 8 / 160 ssd
    Intel Atom C2750 2.40GHz
    SSDSC2BB160G4T - 1 year work

    62.210.106.184
    Datacenter: DC2, Room: 101, Zone: Salle 101, Line: M, Rack: 15, Block: F, Position: 7
    Atom / 8 / 160 ssd
    Intel Atom C2750 2.40GHz
    SSDSC2BB160G4T - 2 year work

    163.172.13.22
    Datacenter: DC2, Room: 101, Zone: Salle 101, Line: N, Rack: 11, Block: B, Position: 21
    Atom / 8 / 128 ssd
    Intel Atom C2750 2.40GHzSAMSUNG MZ7LN128HCHP - 1.5 year work

    163.172.10.213
    Datacenter: DC2, Room: 101, Zone: Salle 101, Line: N, Rack: 8, Block: C, Position: 20
    Atom / 8 / 1 sata
    Intel Atom C2750 2.40GHz
    HGST HTE721010A9E630 - 0,5 year work

    62.210.169.2
    Datacenter: DC2, Room: 101, Zone: Salle 101, Line: N, Rack: 4, Block: A, Position: 9
    Atom / 8 / 128 ssd
    Intel Atom C2750 2.40GHz
    SAMSUNG MZ7LN128HCHP - 1 year work

    163.172.13.60
    Datacenter: DC2, Room: 101, Zone: Salle 101, Line: N, Rack: 11, Block: D, Position: 11
    Atom / 8 / 128 ssd
    Intel Atom C2750 2.40GHz
    SAMSUNG MZ7LN128HCHP - 1,2 year work

    163.172.12.198
    Datacenter: DC2, Room: 101, Zone: Salle 101, Line: N, Rack: 11, Block: E, Position: 7
    Atom / 8 / 128 ssd
    Intel Atom C2750 2.40GHz
    SAMSUNG MZ7LN128HCHP - 1,2 year work

    Check which will not work after 19 months without reboot?

    Who does backups that are not afraid of such things.

  • hawkjohn7hawkjohn7 Member

    my server c2750 auto off if installed with windows server,
    i don't know if this processor problem

  • doughmanesdoughmanes Member

    Vova1234 said: Check which will not work after 19 months without reboot?

    IPs posted on this forum have a history of being DDOS attacked. You were warned.

    Thanked by 1vimalware
  • lbftlbft Member

    Online.net have made a blog post about an issue that is likely this one:

    https://blog.online.net/2017/02/09/component-erratum-potentially-affecting-dedibox-sc-xc-and-scaleway-c2-vc1/

    The TL;DR as I see it is that they don't know what effect the issue will have, but if your server breaks they'll replace it.

    Most interesting:

    We'd like to be more open but this erratum is related to a component protected by NDAs, and whose communication is deliberately restricted by the manufacturer.

  • myhkenmyhken Member

    So Intel hides behind their NDA agreement, most likely the same as they did with Cisco, that did not tell that it was the C2000 CPU that was the issue with their switches. But we know better.

  • claudiofclaudiof Member

    This also affets several Synology Models

    https://www.theregister.co.uk/2017/02/06/cisco_intel_decline_to_link_product_warning_to_faulty_chip/?mt=1486651372531

  • GamerTech24GamerTech24 Member

    NDAs inside

  • DamianDamian Member

    lbft said: but if your server breaks they'll replace it.

    Well, at least that's an improvement over what happens when a kidéchire dies.

    Thanked by 1lbft
  • GamerTech24GamerTech24 Member

    @yomero said:
    OK. So, anyone can explain what's going on here?

    I mean, this is a CPU, it has a bunch of transistors and all that. But... it has some kind of ROM and battery to store something? Or, how it would fail after exactly that time? I can't understand that

    Sounds like the CPU just destroys itself

  • ExpertVMExpertVM Member, Host Rep

    Cisco affected too

    http://www.cisco.com/c/en/us/support/web/clock-signal.html#~faqs

  • myhkenmyhken Member

    ExpertVM said: Cisco affected too

    lol...did you click on the link in the first post? It's goes to Cisco. Nothing new there...

  • WilliamWilliam Member
    edited February 2017

    This happens when power is removed by outage or similar very quickly, normal reboots seem to cause this far more rarely.

    Asrock is replacing boards for free, HP will be one of the larger affected also (moonshot).

    Lots of NAS system use this CPU also and ironically it was IIRC the first Atom to do out of order execution (= MUCH faster) and i always liked the power:core:performance ratio..

    Replacement CPUs have new stepping, so DIE changes (or just code, but doubt it). Special bummer here as SoC and locked down for user side microcode updates...

    Thanked by 2vimalware Maounique
  • rm_rm_ IPv6 Advocate, Veteran
    edited February 2017

    yomero said: OK. So, anyone can explain what's going on here?

    I mean, this is a CPU, it has a bunch of transistors and all that. But... it has some kind of ROM and battery to store something? Or, how it would fail after exactly that time? I can't understand that

    The errata message talks about "SoC LPC_CLKOUT0 and/or LPC_CLKOUT1 signals (Low Pin Count bus clock outputs)".

    LPC controller is a separate chip on most regular motherboards, which does a lot of tasks: https://en.wikipedia.org/wiki/Low_Pin_Count Maybe it is also separate on those Atom C2000 boards, and the
    "LCLK: 33.3 MHz clock, provided by the host" is what goes missing over time. Or conversely, maybe it's built in to the SOC, and whatever generates the 33 MHz signal for it externally, stops doing so.

    Thanked by 1yomero
  • xyzxyz Member
    edited February 2017

    William said: IIRC the first Atom to do out of order execution

    Silvermont is largely in-order. Integer units have an 8 entry ROB (which pales in comparison to CPUs like Haswell with a ~192 entry ROB), but from memory, everything else is completely in-order (including FPU).

    First gen Atom was largely crap as it probably was quickly rushed out of the door, so greatly improving on it probably wasn't too hard.

  • WilliamWilliam Member
    edited February 2017

    xyz said: First gen Atom was largely crap as it probably was quickly rushed out of the door, so greatly improving on it probably wasn't too hard.

    Keep in mind that Silvermont is not second gen (which is 300, after 200, followed by the N stuff) - it is later.

  • zilchzilch Member

    Please, dont die. I need few days to create full backup

  • jvnadrjvnadr Member

    zilch said: Please, dont die. I need few days to create full backup

    "Die" does not mean "lose data"... it means, bricked cpu that can be changed... And what is this that needs... days to be created as backup? How many disks do you have there? 50?

    Thanked by 1netomx
  • WilliamWilliam Member
    edited February 2017

    jvnadr said: it means, bricked cpu that can be changed

    It cannot. The Avoton series is SoC only, not socket. You have to return the entire mainboard (Asrock), IOT thing (Intel) or server (SM/HP).

    Thanked by 1Janevski
  • jvnadrjvnadr Member
    edited February 2017

    William said: It cannot. The Avoton series is SoC only, not socket.

    SOrry, you're right, I meant to write "change of the board".

  • xyzxyz Member

    William said: Keep in mind that Silvermont is not second gen (which is 300, after 200, followed by the N stuff) - it is later.

    In terms of general uArch design, it is. Cedarview etc were just process improvements using the same general uArch. The rest is just marketing (which Intel seems to be a little inconsistent with).

  • WilliamWilliam Member

    xyz said: In terms of general uArch design, it is. Cedarview etc were just process improvements using the same general uArch. The rest is just marketing (which Intel seems to be a little inconsistent with).

    Right, yea that's all the same base (and slow, plus old also DDR2 based still) before - second gen in tech makes sense...

    We'll see what comes next, i'll eye core-M and x5- as replacement for some things, nowadays mostly used in tablets due to lack of PCIe lanes though.

  • netomxnetomx Moderator, Veteran

    Thinking about this, a MiniPC died some months ago, and never found out why.... until this :o

  • WilliamWilliam Member

    netomx said: Thinking about this, a MiniPC died some months ago, and never found out why.... until this :o

    some sort of NAS? The Avotons tend to be pretty rare outside of that and server boards...

  • netomxnetomx Moderator, Veteran

    @William said:

    netomx said: Thinking about this, a MiniPC died some months ago, and never found out why.... until this :o

    some sort of NAS? The Avotons tend to be pretty rare outside of that and server boards...

    Like this one... isn't it?

    https://www.aliexpress.com/item/Wintel-Pro-MINI-PC-intel-atom-X5-Z8300-quad-core-2GB-32GB-with-WIFI-BT4-0/32613906196.html

  • WilliamWilliam Member
    edited February 2017

    No, that are x5 "atoms" which are made of another entirely different CPU design (Cherry-Trail) in 14nm (Avoton is 22nm).

    Thanked by 1netomx
  • netomxnetomx Moderator, Veteran

    @William said:
    No, that are x5 "atoms" which are made of another entirely different CPU design (Cherry-Trail) in 14nm (Avoton is 22nm).

    Oh, ok ok. Well, that sucks anyway hjahah

  • xyzxyz Member

    William said: We'll see what comes next, i'll eye core-M and x5- as replacement for some things, nowadays mostly used in tablets due to lack of PCIe lanes though.

    Goldmont is officially the successor to Silvermont. A few small devices using Goldmont (Apollo Lake) exist, but not that many yet, despite the uArch having been launched around 6 months ago.

    Denverton (Atom C3000 series) is supposed to succeed Avoton, but apart from some leaked rumors, haven't heard much about it.

  • tomcbtomcb Member

    lbft said: if your server breaks they'll replace it.

    They will not replace it!

    I required a Hardware check,

    They only guarantee that your server can start in rescue mode.

    Does not guarantee that your server can work in normal mode.

    They don't think it's a issue .

    4WhFo.png
    4WZe6.png
    4Wj3C.png
    4WIdE.png
    4WR6j.png

    There is a video that records the whole process from installation satrt to crash @1:55:13

    https://drive.google.com/file/d/0B8agXO4adxsTWldtTmlZY0RrZkk/view

  • netomxnetomx Moderator, Veteran

    @tomcb said:

    lbft said: if your server breaks they'll replace it.

    They will not replace it!

    I required a Hardware check,

    They only guarantee that your server can start in rescue mode.

    Does not guarantee that your server can work in normal mode.

    They don't think it's a issue .

    4WhFo.png
    4WZe6.png
    4Wj3C.png
    4WIdE.png
    4WR6j.png

    There is a video that records the whole process from installation satrt to crash @1:55:13

    https://drive.google.com/file/d/0B8agXO4adxsTWldtTmlZY0RrZkk/view

    Check last crash log?

  • WilliamWilliam Member
    edited February 2017

    tomcb said: They will not replace it!

    If your system still boots in rescues this CANNOT BE THIS ISSUE.

    Systems affected by this are UNABLE TO POST, EVEN TO IPMI OR BIOS, EVER AGAIN

    Thanked by 7lbft yomero Janevski rocket sin Mathias tomcb
Sign In or Register to comment.