Re: Building new Athlon AMD64 Socket 939 or 940 machine

From: Martin Cracauer (cracauer_at_cons.org)
Date: 11/28/05

  • Next message: Martin Cracauer: "Re: Building new Athlon AMD64 Socket 939 or 940 machine"
    Date: Mon, 28 Nov 2005 10:35:20 -0500
    To: Nate Eldredge <nge@cs.hmc.edu>
    
    

    > I put together a machine a couple months ago. I used a Tyan S2875ANRF
    > ("Tiger K8W") dual-Opteron motherboard and a single Opteron 244 (1.8 GHz)
    > cpu (second one may come in the future). I am running 5.4 and have been
    > very happy with it.

    Note that the Tiger K8W (PCI) does not support dual-core CPUs, where
    the newest revisions(!) of the Thunder K8W (PCI-X) do. The Thunder is
    an oversize ATX board, though, and expensive.

    > I decided it was worth going for Opteron over Athlon 64. They seem to
    > have better performance (wider memory bus, etc), obviously SMP ability,

    Sorry, the memory bus is not wider than on socket 939. You do have
    additional Hyperlinks in the socket 940 CPUs (real-life performance
    effect unknown to me) and the better boards have seperate memory banks
    for the CPUs.

    > and as I understand it the regular Athlon 64s don't support ECC memory
    > which I insist upon.

    All AMD64 CPUs support ECC. But only socket 940 requires registered
    memory.

    ECC support in socket 754 and 939 is a function of board, but all or
    almost all Asus boards support it (unregistered ECC of course).

    > The Athlon 64FX appear to be similar to the fastest
    > Opterons and differ mainly for price marketing purposes (sold to gamers).

    The one feature of the FX CPUs is that the LDT->CPU frequency
    multiplier is unlocked, which allows for freer combining of frequency
    settings when overclocking. Plus they have high base clocks and many
    seem to be binned to overclock well.

    Martin

    -- 
    %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    Martin Cracauer <cracauer@cons.org>   http://www.cons.org/cracauer/
    FreeBSD - where you want to go, today.      http://www.freebsd.org/
    _______________________________________________
    freebsd-current@freebsd.org mailing list
    http://lists.freebsd.org/mailman/listinfo/freebsd-current
    To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"
    

  • Next message: Martin Cracauer: "Re: Building new Athlon AMD64 Socket 939 or 940 machine"

    Relevant Pages

    • SMP Timekeeping regression in 2.6.16-rc5 (and 2.6.16-rc4)
      ... Just built and installed a 2.6.16-rc5 kernel on my Powermac G4 Dual ... Essentially the two CPUs have ... # Loadable module support ...
      (Linux-Kernel)
    • Re: NUMA API
      ... I am aware of the code in libnuma by ak but this code ... numa_all_nodes could maybe in future change with node hotplug support, ... NUMA is only about "what CPUs ... but not "sticky memory" effects. ...
      (Linux-Kernel)
    • [SUMMARY] OT: High End Servers comparison
      ... (typical database and banking benchmarks are transactions per second.) ... We just did a comparison of some Sun and IBM servers for a new project. ... Type and number of disks supported: ... IBM, for example, will ship you a system with some CPUs ...
      (SunManagers)
    • Re: Unlock the multiplier
      ... it uses an AMD cpu (socket A) and many ... the "locked" status of most current production CPUs is without doubt. ... multiplier altering software for socket A systems (I've seen it ... |> Do you know what I need to unlock another one. ...
      (alt.comp.periphs.mainboard.asus)
    • [PATCH] x86: remove NexGen support
      ... It is claimed that NexGen CPUs were never shipped: ... the code intended to support NexGen thereby being ... * but fortunately the switch to using exceptions got rid of all that. ...
      (Linux-Kernel)