Workstation Specs

Discussion in 'SolidWorks' started by Donegalforever, Jan 15, 2007.

  1. Donegalforever, Jan 15, 2007
    #1
  2. Donegalforever

    JKimmel Guest

    I got the 490. After upgrading its video card, its processor and its
    memory, it's still a piece of shit, and the warrenty is void when you
    open the box. I built up a custom computer with an AMD Athlon x2 that
    benchmarks TWICE as fast for 1/4 the money.

    --
    J Kimmel

    www.metalinnovations.com

    "Cuius testiculos habes, habeas cardia et cerebellum." - When you have
    their full attention in your grip, their hearts and minds will follow.
     
    JKimmel, Jan 15, 2007
    #2
  3. Donegalforever

    iQ Guest

    i just finished testing the Dell M65 (core2duo) and Dell690 (xeon
    core2duo 3GH, 4GB ram, single 4500 nvidia) workstations. in WinXP (x32
    mode) running SWx 2006 SP5.0. (baseline dell 670 specs, 3.6GH, 4GB
    ram, nvidia 4500). the M65 took 20% less time than my previous Dell
    670. the Dell690 took 30% less time then my previous Dell 670. i
    almost went for the laptops. this was testing on our product line.

    no test on other computers that you specified.

    one thig of note is that our IT manager had pre-agreed with Dell to
    send one of the test units back as we had a 30 waranty for returns.
    this allowed me to test hardware configurations on my models and
    drawings. helped me select what hardware would give bang for buck and
    which one would be not compatible. of note: dual video cards seemed to
    slow down computer in XP32. iQ
     
    iQ, Jan 15, 2007
    #3
  4. What hard drives did you have on the 670 & the 690? Were they both SCSI, or
    one IDE, or whatever?

    WT
     
    Wayne Tiffany, Jan 16, 2007
    #4
  5. Thanks for the replies.

    I actually wasn't looking at buying a Dell because I've got too many
    horror stories from friends of mine.

    I was looking at the site to get a specification to start off with, and
    get a custom unit built.

    Brian
     
    Donegalforever, Jan 16, 2007
    #5
  6. Donegalforever

    iQ Guest

    670 had SCSI but i got the SAS on the 690. noise level on the 670 was
    disturbing, more of a vibration to desktop than anything. the 690 is
    real quiet, you can hear the hard drive a little when it hits.

    the only problem on the 690 that i had was the size of the box (and the
    stabilization rack below the unit. actual size with the rack is 12"
    wide x 22.5" tall x 22" deep (deep not including the 4" of space behind
    unit for cables). and we opened it up and it is full. what ever
    happened to the shrinking PC? iQ
     
    iQ, Jan 16, 2007
    #6
  7. Donegalforever

    Bo Guest

    There are always a number of issues that define a particular person's
    work and needs, and then what 'Extras' they would like. Everyone is a
    bit different.

    1. Workstation today has different meanings: I consider my Dell M60
    Laptop a workstation, and for my size of assemblies, it does not limit
    me in any way but the 1900 across pixel screen, and I have learned to
    live with that as being entirely adequate for me. My understanding is
    the m90 has made a considerable jump in performance.

    2. XP Pro vs Vista: SolidWorks will no doubt have to support XP Pro
    for at least 2 more years. Early corporate users doing evaluation are
    frank in their note that Vista is snail slow and prone to numerous
    problems just with things they normally use. It is going to take a
    year or two for Vista to shake out, and probably likewise for
    SolidWorks and its Patners add-ons.

    3. CPU box compatibility w/Vista DRM: It sounds like Microsoft is
    going to make future Intel/AMD boxes comply with some draconian sets of
    requirements which mean Vista is going to be permanently crippled by
    the cycles required to constantly checke everything in the OS for
    'banned' or unauthorized transmissions of data. My personal ultimate
    worry, is that we are heading into an era where the OS & Hardware are
    constantly checking your applications and work with the intent to
    deprive you of the use of something if it 'detects' you are 'not
    authorized' to be say using that DVD from SolidWorks, or maybe could it
    be, even SolidWorks itself at some time in the future when SolidWorks
    changes the terms of its license to make it mandatory to keep
    maintenance fees every year or your SolidWorks is zapped (plus if you
    do not connect to the internet every week or so, the OS &/or SolidWorks
    could simply tell you to connect or your OS will go down to low level
    functionality. I do not like the direction I see the
    Vista/DRM/authorization/mandatory connections to the Internet going.
    ---------Thus I recommend buying a PC now for running Win XP only which
    is not optimized for Vista, which assumes that I might avoid some of
    the Vista cost, complexity, and potential slowdowns.

    WinXP, for all its faults is known, can be stripped down to minimize
    processes, does NOT have to be connected to the Internet to operate
    (more than once), and has limited behind the scenes DRM crapware stuff
    hidden in the OS. I'll look at Vista when it is production ready, but
    XP will rule for at least 2 years.

    4. Working in a multi-user office versus alone: If you must plug into
    and use network servers etc, then it looks like you will have to
    "follow the pack", but if you are a single user and work as a mobile
    consultant, then a different set of requirements evolve.

    I think the Windows world is getting more complex, yet I wonder if it
    will be better any time soon, as it seems the audio-visual
    mega-corporations & Microsoft have come up with a way to string people
    up into a VistaBox that looks a bit more like a prison than a free
    computing environment.

    Bo
     
    Bo, Jan 16, 2007
    #7
  8. Many thanks all, I think I have enough to get something sorted out now
     
    Donegalforever, Jan 17, 2007
    #8
  9. Donegalforever

    TOP Guest

    I just had a chance to run my STAR2.1 benchmark on a brand new out of
    the box Dell 390 with a:

    Dual-Core IntelĀ® CoreTM 2 Extreme X6800 (2.93GHz, 1066MHz FSB, 4MB L2
    Cache) and 2mb of RAM.

    It had times averaging around 56 seconds compared to a AMD64 FX53 that
    runs the same in the 35 second time frame and AMD64 4400+ that run in
    the 40s time frame. I'll try to get Ship in a Bottle times tommorrow.
    It isn't much improvement over the old 2.6Ghz Pentium it replaced.
    That machine was also a Dell and four years old.

    This particular test is not affected by either graphics card or
    memory. It is almost pure CPU. I'll try a more scientific test in
    coming days.
     
    TOP, Feb 13, 2007
    #9
  10. Donegalforever

    matt Guest

    Paul,

    Does that have hyperthreading on? And just to clarify, 32 or 64? XP or
    Vista? I thought the Extreme line was supposed to be the FX60 killer.

    Anyway, I guess I'll stick with AMD for a while. I'll be interested to
    see other results.
     
    matt, Feb 13, 2007
    #10
  11. Donegalforever

    TOP Guest

    Matt,

    This was right out of the box. I didn't have a chance to really dig
    into it. I squeezed the test in during a dbWorks install. Unless Dell
    is in the habit of turning on HT I would say it doesn't have it turned
    on. And if it does I would have to wonder if they have a switch in the
    BIOS to turn it off. 32 bit XP with an Nvidia Quadro high end card,
    although STAR2.1 really doesn't need much memory or graphics. SW2007
    SP2.2
    published benchmarks with other software, but none of them have done
    SW. SW has traditionally fared quite well on AMD in comparison with
    Intel. The Extreme might be good if overclocked, but the results I saw
    would make me wonder. I am surprised we haven't seen some results
    posted here based on the Duo platform.

    Since top end AMD will get below 30 seconds on this and I have seen
    even lower posted here I fully expected this box to crack the 10
    second barrier. Either the machine has a problem or that is what it
    is. Last year's SWW boxes did way better than 55 seconds.

    The other possible caveat is that I haven't added SW2007 support to
    the STAR benchmark and ran it in 2006 mode. I wouldn't think that
    anything in the API changed that radically but it is something else to
    look into.

    In coming days I'll try some real world comparisons and other
    benchmarks.
     
    TOP, Feb 13, 2007
    #11
  12. Donegalforever

    Dale Dunn Guest

    Core2Duo does not support Hyperthreading.
     
    Dale Dunn, Feb 14, 2007
    #12
  13. Donegalforever

    TOP Guest

    Ship in a Bottle times for 50 iterations:

    AMD64 FX53 29/23s
    FX3000 AGP graphics card

    Core2 X6800 23/17s
    FX4500 PCI graphics card

    That's 21%/26% faster for image quality from high to low. Not much of
    an improvement over a single processor machine that is three years
    old.

    So the Core 2 is a bit faster on the Ship in a bottle, but much slower
    on the STAR benchmark. Given the Core2 performance on other benchmarks
    it is rather surprising that it is not twice as fast as the old AMD
    chip. What is it about AMD that makes them do so well with SW?
    Upgrading to the newest AMD chips would almost guarantee a doubling of
    the old FX53 performance.
     
    TOP, Feb 23, 2007
    #13
  14. Donegalforever

    Dale Dunn Guest

    chip. What is it about AMD that makes them do so well with SW?

    Interesting data. From everything I've read comparing the two
    architectures, the only thing AMD is supposed to be doing better right now
    is memory throughput. Is there something about the STAR benchmark that
    would cause a lot of traffic out to main memory instead of being able to
    work in the L2 cache?

    One of the criticisms of Ship-in-a-bottle and the STAR benchmark is that
    they are focussed on just a few functions of SW. The difference between
    these two tests seems to highlight the fact that different parts of SW have
    different needs, and will therefore perform differently on different
    platforms. Do you have any plans to run something more "comprehensive" like
    SPECapc 2005?
     
    Dale Dunn, Feb 23, 2007
    #14
  15. Donegalforever

    jimsym Guest

    See http://www.digit-life.com/articles2/cpu/intel-conroe-2-13-ghz.html
    for a very thorough comparison of the Core2 Due and Athlon X2
    architecture. You have to get into the spreadsheet to see the actual
    SPECapc for SolidWorks 2005 scores.

    The 2.66GHz Core2 Duo E6700 outperforms the 2.8 GHz Athlon FX62 on
    both ProE and SolidWorks benchmarks. (SPECapc benchmarks are CPU
    bound and scale directly with processor speed. There's actually very
    little spread between different "professional" graphics cards. )

    The AMD architecture still holds an edge in pure floating point
    operations, thus the Athlons excelled on the MATLAB tests.

    There is virtually NO DIFFERENCE in performance between the old
    Athlon64 platform and the new AM2 platform. While I haven't seen any
    SPECapc scores, the new 3.0GHz 6000+ is NOT proving to be a great
    performer. The E6700 beats it on nearly every benchmark and even the
    E6400 is right on its tail - and a fair amount cheaper.
     
    jimsym, Feb 23, 2007
    #15
  16. Donegalforever

    solid steve Guest

    Hi Brian

    If you are in the UK I can recommend,

    www.xworksinteractive.com

    I have a Dual opteron 254 from them 12 months old now but still quick.

    They will build you anything you want. Intel or AMD.

    steve
     
    solid steve, Feb 23, 2007
    #16
  17. Donegalforever

    TOP Guest

    The next step is to run actual assemblies of the large variety. That
    is what we really do. I run the quick tests to get a feel for the lay
    of the land. I don't go much for the SPEC benchmarks. In the past they
    were heavily weighted to graphics. I haven't looked at the code for
    the newer benchmarks. They may have changed, but I believe many
    graphics card companies were on the committee that controls this
    test. What I don't like about SPEC is that there is a new benchmark
    every year which means that you can't test the software itself from
    release to release. STAR2.1 gets past that by generating it's own
    geomtry in whatever version it is running in.

    The Core2 may outperform the FX62 in those benchmarks, but I am
    comparing an out of the box Dell with an older FX53 that has been
    optimized. It isn't overclocked, just optimized.

    The next thing I will post is comparisons of assembly open/rebuild and
    then open and rebuild drawing.

    TOP
     
    TOP, Feb 23, 2007
    #17
  18. Donegalforever

    TOP Guest

    What kind of times do you get on a benchmark like STAR2.1 or Ship in a
    Bottle?
     
    TOP, Feb 23, 2007
    #18
Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.