6-core CPU on the horizon - VMS

This is a discussion on 6-core CPU on the horizon - VMS ; For anyone watching competing technologies, Intel has a 6-core CPU on the horizon. http://arstechnica.com/news.ars/post...processor.html Neil Rieck Kitchener/Waterloo/Cambridge, Ontario, Canada. http://www3.sympatico.ca/n.rieck/...

+ Reply to Thread
Page 1 of 3 1 2 3 LastLast
Results 1 to 20 of 45

Thread: 6-core CPU on the horizon

  1. 6-core CPU on the horizon

    For anyone watching competing technologies, Intel has a 6-core CPU on
    the horizon.

    http://arstechnica.com/news.ars/post...processor.html

    Neil Rieck
    Kitchener/Waterloo/Cambridge,
    Ontario, Canada.
    http://www3.sympatico.ca/n.rieck/

  2. Re: 6-core CPU on the horizon

    Le 26-02-2008, à propos de
    6-core CPU on the horizon,
    Neil Rieck écrivait dans comp.os.vms :
    > For anyone watching competing technologies, Intel has a 6-core CPU on
    > the horizon.
    >
    > http://arstechnica.com/news.ars/post...processor.html


    Yes and ? My Sparc T1000 uses 8-cores CPU (and 4 threads by core)
    for a long time...

    JKB

    --
    Le cerveau, c'est un véritable scandale écologique. Il représente 2% de notre
    masse corporelle, mais disperse à lui seul 25% de l'énergie que nous
    consommons tous les jours.

  3. RE: 6-core CPU on the horizon


    > -----Original Message-----
    > From: Neil Rieck [mailto:n.rieck@sympatico.ca]
    > Sent: February 26, 2008 7:33 AM
    > To: Info-VAX@Mvb.Saic.Com
    > Subject: 6-core CPU on the horizon
    >
    > For anyone watching competing technologies, Intel has a 6-core CPU on
    > the horizon.
    >
    > http://arstechnica.com/news.ars/post...p-on-upcoming-
    > intel-dunnington-six-core-processor.html
    >
    > Neil Rieck
    > Kitchener/Waterloo/Cambridge,
    > Ontario, Canada.
    > http://www3.sympatico.ca/n.rieck/


    Well, with regards to the HW, one needs to remember that without the SW to
    support these new capabilities i.e. App's written to support multi-threading
    and SMP etc. then these new capabilities are moot.

    As an example, my young lad has part time job working in an Internet gaming
    lounge and one of the first things they typically do when they receive a
    new PC is disable the 2nd core as it negatively impacts most of the current
    games on the market today.

    This may be a select niche, but how many applications out there today have
    really been designed with 2 way multi-threading, cache thrashing and SMP in
    mind, let alone 4-way and 8-way?

    Note this is not a new issue. When VMS V4.7 was out, it was asymmetrical
    processing and some applications actually ran better if you halted the
    secondary.


    Regards

    Kerry Main
    Senior Consultant
    HP Services Canada
    Voice: 613-254-8911
    Fax: 613-591-4477
    kerryDOTmainAThpDOTcom
    (remove the DOT's and AT)

    OpenVMS - the secure, multi-site OS that just works.




  4. Re: 6-core CPU on the horizon


    >As an example, my young lad has part time job working in an Internet gaming
    >lounge and one of the first things they typically do when they receive a
    >new PC is disable the 2nd core as it negatively impacts most of the current
    >games on the market today.


    This is also common practice in the Linux high performance cluster world.
    At my current employer, on dual socket/dual core systems we run finite
    element simulations using only one core per socket. May seem like a waste,
    but we get better performance.

    Regards, Jim



  5. Re: 6-core CPU on the horizon

    On Feb 26, 7:33*am, Neil Rieck wrote:
    > For anyone watching competing technologies, Intel has a 6-core CPU on
    > the horizon.
    >
    > http://arstechnica.com/news.ars/post...lip-on-upcomin...
    >
    > Neil Rieck
    > Kitchener/Waterloo/Cambridge,
    > Ontario, Canada.http://www3.sympatico.ca/n.rieck/


    More info:

    http://gizmodo.com/360476/intels-6+c...cessors-leaked

    Neil

  6. Re: 6-core CPU on the horizon

    On Feb 26, 9:01*am, "Main, Kerry" wrote:
    >

    [...snip...]
    >
    > Well, with regards to the HW, one needs to remember that without the SW to
    > support these new capabilities i.e. App's written to support multi-threading
    > and SMP etc. then these new capabilities are moot.
    >
    > As an example, my young lad has part time job working in an Internet gaming
    > lounge and one of the first things they typically do when they receive a
    > new PC is disable the 2nd core as it negatively impacts most of the current
    > games on the market today.
    >
    > This may be a select niche, but how many applications out there today have
    > really been designed with 2 way multi-threading, cache thrashing and SMP in
    > mind, let alone 4-way and 8-way?
    >
    > Note this is not a new issue. When VMS V4.7 was out, it was asymmetrical
    > processing and some applications actually ran better if you halted the
    > secondary.
    >
    > Regards
    >
    > Kerry Main
    >


    Don't take this the wrong way because I'm a big fan of OpenVMS but
    OpenVMS it is not the only OS capable of supporting SMP. I just
    purchased an HP-Pavilion A6245N Quad-core (Q6600) from FutureShop for
    $650. It came with Windows-Vista preinstalled (which I'm not thrilled
    about) and all 4 cores are up and running. This is a personal WINDOZ
    computer which means that some of the cores maybe doing mundane things
    like virus and malware scans but they also are contributing to
    folding@home

    http://www3.sympatico.ca/n.rieck/doc...g_at_home.html

    I have always worried that with various UNIX + Linux offerings being
    able to run on these multi-core Intel machines, that OpenVMS (which
    has not been ported to x86-64) might be priced out of the market
    place. Just my two cents worth.

    Neil Rieck




  7. Re: 6-core CPU on the horizon

    Neil Rieck wrote:
    > For anyone watching competing technologies, Intel has a 6-core CPU on
    > the horizon.


    Is this Moore's law that is coming to an end ?

    If increased capacity comes via more and more CPUs, it no longer makes
    each CPU faster as was the case in the past.


    Seems to me that anything between now and when CSI becomes available is
    just stop gap measures. Once CSI comes to the 8086, then the real future
    will start to happen. Will HP start to build superdome-type systems
    based on the 8086 ?

    Will Dell start to enter the real enterprise computing market by
    building some large "mainframe" type servers ?

    And if so, how quickly.

  8. Re: 6-core CPU on the horizon

    JKB wrote:
    > Le 26-02-2008, ? propos de
    > 6-core CPU on the horizon,
    > Neil Rieck ?crivait dans comp.os.vms :
    >> For anyone watching competing technologies, Intel has a 6-core CPU on
    >> the horizon.
    >>
    >> http://arstechnica.com/news.ars/post...processor.html

    >
    > Yes and ? My Sparc T1000 uses 8-cores CPU (and 4 threads by core)
    > for a long time...


    and each "thread" is about the speed of a 486.

  9. Re: 6-core CPU on the horizon

    Le 26-02-2008, à propos de
    Re: 6-core CPU on the horizon,
    Cydrome Leader écrivait dans comp.os.vms :
    > JKB wrote:
    >> Le 26-02-2008, ? propos de
    >> 6-core CPU on the horizon,
    >> Neil Rieck ?crivait dans comp.os.vms :
    >>> For anyone watching competing technologies, Intel has a 6-core CPU on
    >>> the horizon.
    >>>
    >>> http://arstechnica.com/news.ars/post...processor.html

    >>
    >> Yes and ? My Sparc T1000 uses 8-cores CPU (and 4 threads by core)
    >> for a long time...

    >
    > and each "thread" is about the speed of a 486.


    Nope. This server is a database server and runs faster than all x64
    I have tried.

    JKB

    --
    Le cerveau, c'est un véritable scandale écologique. Il représente 2% de notre
    masse corporelle, mais disperse à lui seul 25% de l'énergie que nous
    consommons tous les jours.

  10. Re: 6-core CPU on the horizon

    On Feb 26, 1:43*pm, JF Mezei wrote:
    > Neil Rieck wrote:
    > > For anyone watching competing technologies, Intel has a 6-core CPU on
    > > the horizon.

    >
    > Is this Moore's law that is coming to an end ?
    >
    > If increased capacity comes via more and more CPUs, it no longer makes
    > each CPU faster as was the case in the past.
    >


    I think Intel admitted as much when they turfed out NetBurst. If
    memory servces, P4-Prescott and Pentium-D 31-stage instruction
    pipelines (NetBurst) which was only good for running benchmarks but
    really bad for real world OS annoyances like interrupts. So with Core2
    they dropped back to between 10-12 stages where AMD had been all
    along. Meanwhile, Intel also admitted that each processor improvement
    was an exponential effort for a linear return so it only made sense to
    do multiple cores.

    Now we all know that some Alpha engineers went to AMD while the
    majority ended up at Intel. So it is anyone surprised that we are
    seeing really cool chips come out right about now?


    Neil Rieck
    Waterloo, Ontario.

  11. Re: 6-core CPU on the horizon

    JKB wrote:
    > Le 26-02-2008, ? propos de
    > Re: 6-core CPU on the horizon,
    > Cydrome Leader ?crivait dans comp.os.vms :
    >> JKB wrote:
    >>> Le 26-02-2008, ? propos de
    >>> 6-core CPU on the horizon,
    >>> Neil Rieck ?crivait dans comp.os.vms :
    >>>> For anyone watching competing technologies, Intel has a 6-core CPU on
    >>>> the horizon.
    >>>>
    >>>> http://arstechnica.com/news.ars/post...processor.html
    >>>
    >>> Yes and ? My Sparc T1000 uses 8-cores CPU (and 4 threads by core)
    >>> for a long time...

    >>
    >> and each "thread" is about the speed of a 486.

    >
    > Nope. This server is a database server and runs faster than all x64
    > I have tried.
    >
    > JKB
    >


    Each thread is horribly slow, there's just lots of them. shut off 30 cores
    and see how that machine handles.

  12. Re: 6-core CPU on the horizon

    Le 26-02-2008, à propos de
    Re: 6-core CPU on the horizon,
    Cydrome Leader écrivait dans comp.os.vms :
    > JKB wrote:
    >> Le 26-02-2008, ? propos de
    >> Re: 6-core CPU on the horizon,
    >> Cydrome Leader ?crivait dans comp.os.vms :
    >>> JKB wrote:
    >>>> Le 26-02-2008, ? propos de
    >>>> 6-core CPU on the horizon,
    >>>> Neil Rieck ?crivait dans comp.os.vms :
    >>>>> For anyone watching competing technologies, Intel has a 6-core CPU on
    >>>>> the horizon.
    >>>>>
    >>>>> http://arstechnica.com/news.ars/post...processor.html
    >>>>
    >>>> Yes and ? My Sparc T1000 uses 8-cores CPU (and 4 threads by core)
    >>>> for a long time...
    >>>
    >>> and each "thread" is about the speed of a 486.

    >>
    >> Nope. This server is a database server and runs faster than all x64
    >> I have tried.
    >>
    >> JKB
    >>

    >
    > Each thread is horribly slow, there's just lots of them. shut off 30 cores
    > and see how that machine handles.


    I have a lot of T1000 and I have done a lot of tests before buying
    this kind of material. I have made some tests to see degradation due
    to SMP architecture. This degradation is not important and when you
    know what you do with this material, you can achieve very good
    performances in parallel computation (or with multithreaded
    softwares). Each thread is not horribly slow as you say, and its
    performance is good even when load average is high. I think you
    haven't tried this material.

    JKB

    PS: [OT], thus for me,

    --
    Le cerveau, c'est un véritable scandale écologique. Il représente 2% de notre
    masse corporelle, mais disperse à lui seul 25% de l'énergie que nous
    consommons tous les jours.

  13. Re: 6-core CPU on the horizon

    On Feb 26, 7:57*am, JKB wrote:
    [...snip...]

    > * * * * Yes and ? My Sparc T1000 uses 8-cores CPU (and 4 threads by core)
    > * * * * for a long time...
    >
    > * * * * JKB
    >


    That's a really nice system but how much did it cost?
    I just bought an HP-Pavilion with a quad-core (q6000) @ 2.4 GHz for
    $650
    A full install of Microsoft Office only took a minute.

    http://www.tomshardware.com/cpu_2007.html

    Neil Rieck
    Waterloo, Ontario.


  14. Re: 6-core CPU on the horizon


    "JF Mezei" wrote in message
    news:47c45e32$0$30973$c3e8da3@news.astraweb.com...
    > Neil Rieck wrote:
    > > For anyone watching competing technologies, Intel has a 6-core CPU on
    > > the horizon.

    >
    > Is this Moore's law that is coming to an end ?
    >
    > If increased capacity comes via more and more CPUs, it no longer makes
    > each CPU faster as was the case in the past.
    >
    >
    > Seems to me that anything between now and when CSI becomes available is
    > just stop gap measures. Once CSI comes to the 8086, then the real future
    > will start to happen. Will HP start to build superdome-type systems
    > based on the 8086 ?
    >
    > Will Dell start to enter the real enterprise computing market by
    > building some large "mainframe" type servers ?
    >
    > And if so, how quickly.



    "Once CSI comes to the 8086, then the real future will start to happen. "

    So Intel have been telling us. Have HP said much on the subject so far? You
    might not expect to hear much about it from companies with no interest in
    Itanium...


    "Will HP start to build superdome-type systems based on the 8086 ? Will
    Dell start to enter the real enterprise computing market by building some
    large "mainframe" type servers ?"

    Don't know about those two questions as such, but Unisys tried "mainframe
    x86" a long time ago with their ES7000 family, which Compaq briefly rebadged
    (as the Proliant 9000?). Unisys still sell it:
    http://www.unisys.com/products/enterprise__servers/ and today it offers up
    to 32 processors (choice of Xeon or Itanium), as well as up to 512GB memory,
    as well as hardware partitioning and a reported 99.99% availability.
    Unfortunately for Unisys, it doesn't have a real enterprise-class OS to run,
    just Linux or Windows (and presumably lots of VMware instances). The
    hardware sounds interesting but an enterprise-class machine probably needs a
    more suitable operating system (and the things that go with it) than those
    two.

    Regards
    John



  15. Re: 6-core CPU on the horizon

    Neil Rieck wrote:

    > I just bought an HP-Pavilion with a quad-core (q6000) @ 2.4 GHz for
    > $650


    Out of curiosity, does HP's wintel box offer anything that Dell or
    Lenovo or others don't have ?

    Do you feel comfortable buying from a company that is destroying VMS's
    chances ?




    > A full install of Microsoft Office only took a minute.


    Quick, someone call 911 to send paramedics to Mr VAXman's house, he's
    just had a heart attack after reading this :-) :-) :-) :-)

  16. Re: 6-core CPU on the horizon

    JF Mezei wrote:
    > Neil Rieck wrote:
    >> For anyone watching competing technologies, Intel has a 6-core CPU on
    >> the horizon.

    >
    > Is this Moore's law that is coming to an end ?


    No.

    But instead of increasing GHz they increase the number of cores.

    Arne

  17. Re: 6-core CPU on the horizon

    "Arne Vajhøj" wrote in message
    news:47c4ba59$0$90272$14726298@news.sunsite.dk...
    > JF Mezei wrote:
    >> Neil Rieck wrote:
    >>> For anyone watching competing technologies, Intel has a 6-core CPU on
    >>> the horizon.

    >>
    >> Is this Moore's law that is coming to an end ?

    >
    > No.
    >
    > But instead of increasing GHz they increase the number of cores.
    >
    > Arne
    >



    Moore's "Law" was that the transistor count would double every 18 months or
    so. Not the clock speed. There are real issues of radio frequency
    interference when push clocks past the 3 to 3.5 GHz range. This is why
    newer processors keep the clock around 3 GHz and increase the number of
    cores.

    Mike.



  18. Re: 6-core CPU on the horizon

    On Feb 26, 6:29*pm, JF Mezei wrote:

    [...snip...]

    First off, here is an overview of my recent PC activities:

    In 2007, I purchased the following from Factory Direct:
    1) Compaq Presario (Pentium-D 820) running Windows-XP
    2) Compaq Presario (Pentium-D 925) running Windows-Vista

    Then in September I heard a piece on NPR (Science Friday) about
    folding[at]home so decided to contribute to it. I bought 5 "used" high
    powered ATI Radeon cards on eBay for installation in some 24/7 PCs at
    my place of employment. (ATI cards are used as stream processors to
    increase science throughput by 7000% over CPU clients)
    http://folding.stanford.edu/

    In 2008, I purchased the following from Future Shop (a.k.a. Best Buy):
    1) HP-Presario based upon AMD 6000+ running Windows-Vista
    2) HP-Presario based upon Core2quad (Q6600) running Windows-Vista

    BTW, the 2007 systems are in my basement doing folding[at]home (a tiny
    "folding farm" if you will).
    http://www3.sympatico.ca/n.rieck/doc...g_at_home.html

    >
    > Out of curiosity, does HP's wintel box offer anything that Dell or
    > Lenovo or others don't have ?
    >


    I work for a group that does PC installation and repair. I hate to
    paint with a wide brush BUT, in general and only based upon things
    I've personally seen, HP/Compaq PCs are (currenly) higher quality than
    either Dell or Lenovo equivalents. Caveat: maybe my employer only
    purchases cheaper lower end products from these companies; maybe Dell
    and Lenovo make better products but I just haven't seen them yet.

    > Do you feel comfortable buying from a company that is destroying VMS's
    > chances ?
    >


    Are you referring to HP? :-) :-) :-)

    Seriously, I only purchased systems that could run folding[at]home
    clients. I would have considered buying Alpha or Itanium systems if
    folding and/or BOINC clients were available for them but they weren't.
    Remember back in the old days when Alpha led the way in SETI analysis?
    Compaq was only interested in this when they had the best crunching
    stats. Then when SETI converted over to the BOINC framework (and PCs
    got faster due to memory technolgies like DDR2), HP didn't bother to
    help port BOINC to either Alpha or Itanium.

    Meanwhile, companies like IBM, have the BOINC platform running on
    every product they sell to help with medical reseach.
    http://en.wikipedia.org/wiki/World_community_grid
    They make server clients available to their customers and you just
    know their PR people are going to take partial credit for curring
    Alzheimers or Cancer. (does anyone remember what happened when
    DeepBlue beat Kasparov? IBM stock jumped 35% even though DeepBlue was
    never ever going to be marketed)

    Many servers (Alpha and Itanium as well) do much less at night and I
    see nothing wrong with companies diverting these resources to medical
    research when possible. So why hasn't HP produced BOINC (and or
    folding) clients for their server products?

    > > A full install of Microsoft Office only took a minute.

    >
    > Quick, someone call 911 to send paramedics to Mr VAXman's house, he's
    > just had a heart attack after reading this :-) :-) :-) :-)


    Neil Rieck
    Kitchener/Waterloo/Cambridge,
    Ontario, Canada.
    http://www3.sympatico.ca/n.rieck/

  19. Re: 6-core CPU on the horizon

    In article <47c4a117$0$31251$c3e8da3@news.astraweb.com>, JF Mezei writes:
    >Neil Rieck wrote:
    >
    >>{...snip...}
    >> A full install of Microsoft Office only took a minute.

    >
    >Quick, someone call 911 to send paramedics to Mr VAXman's house, he's
    >just had a heart attack after reading this :-) :-) :-) :-)


    Nah. Nothing revolutionary there JT. I can flush **** down my toilet
    and into the storage tank in roughly the same time or less, so dumping
    M$ **** onto a PeeCee's storage device in a minute is quite believable.

    As for any heart attack, let's see what today's duplex ultrasounds says
    for my chances.

    --
    VAXman- A Bored Certified VMS Kernel Mode Hacker VAXman(at)TMESIS(dot)COM

    "Well my son, life is like a beanstalk, isn't it?"

    http://tmesis.com/drat.html

  20. Re: 6-core CPU on the horizon

    JKB wrote:
    > Le 26-02-2008, ? propos de
    > Re: 6-core CPU on the horizon,
    > Cydrome Leader ?crivait dans comp.os.vms :
    >> JKB wrote:
    >>> Le 26-02-2008, ? propos de
    >>> Re: 6-core CPU on the horizon,
    >>> Cydrome Leader ?crivait dans comp.os.vms :
    >>>> JKB wrote:
    >>>>> Le 26-02-2008, ? propos de
    >>>>> 6-core CPU on the horizon,
    >>>>> Neil Rieck ?crivait dans comp.os.vms :
    >>>>>> For anyone watching competing technologies, Intel has a 6-core CPU on
    >>>>>> the horizon.
    >>>>>>
    >>>>>> http://arstechnica.com/news.ars/post...processor.html
    >>>>>
    >>>>> Yes and ? My Sparc T1000 uses 8-cores CPU (and 4 threads by core)
    >>>>> for a long time...
    >>>>
    >>>> and each "thread" is about the speed of a 486.
    >>>
    >>> Nope. This server is a database server and runs faster than all x64
    >>> I have tried.
    >>>
    >>> JKB
    >>>

    >>
    >> Each thread is horribly slow, there's just lots of them. shut off 30 cores
    >> and see how that machine handles.

    >
    > I have a lot of T1000 and I have done a lot of tests before buying
    > this kind of material. I have made some tests to see degradation due
    > to SMP architecture. This degradation is not important and when you
    > know what you do with this material, you can achieve very good
    > performances in parallel computation (or with multithreaded
    > softwares). Each thread is not horribly slow as you say, and its
    > performance is good even when load average is high. I think you
    > haven't tried this material.
    >
    > JKB
    >
    > PS: [OT], thus for me,


    bull****.

    run a compile on a t1000. it's like stepping back into the 1990s. You can
    read the compiler jabber as it slowly scrolls by.

    threaded stuff seems happy on such a machine, but for single processes
    that want CPU, it's really really slow.

+ Reply to Thread
Page 1 of 3 1 2 3 LastLast