Page 2 of 3 FirstFirst 123 LastLast
Results 16 to 30 of 38

Thread: 5 million + polygons...

  1. #16
    Registered User Rayek's Avatar
    Join Date
    Feb 2006
    Location
    Vancouver, BC
    Posts
    1,459
    I, for one, would be deeply interested to hear from you whether that will work. A good friend of mine, owner of a 11 months old Mac Pro workstation, has been in contact with both Nvidia and Apple over the last two months to get clear and obvious information regarding video card upgrade options. So far only the Quadro was suggested, and Apple's support was decidedly less than stellar - he is still confused, and Apple was no help at all. Nor was Nvidia.

    I wish the best of luck to you: please keep us informed whether or not you get that 580 working! It could make all the difference to my colleague as well (he is considering dropping mac for once and for all, although being a fervent mac user since the very first Apple machines)

    Quote Originally Posted by mikkelen View Post
    @rsfd
    I will get my workstation up and running within a week, including all components, ram, gfx etc... if they are not delayed in customs... You are right, it's not a specific Mac version of the GDDX580, but I'll terrorize both Apple and nVidia customer support until it works properly. The drivers are officially made by nVidia and Apple has opened up their GFX-card policies, so there should be no real issues in theory. I'll post my experiences in this thread.
    Win10 64 - i7 [email protected], p6t Deluxe v1, 48gb, Nvidia GTX 1080 8GB, Revodrive X2 240gb, e-mu 1820. Screens: 2 x Samsung s27a850ds 2560x1440, HP 1920x1200 in portrait mode

  2. #17
    Super Member Captain Obvious's Avatar
    Join Date
    Dec 2004
    Location
    London
    Posts
    4,502
    Quote Originally Posted by JonW View Post
    How is it with Modeler?
    What, people still use that? I haven't used Modeler for anything serious since modo 103. Hang on, I'll check... edit: Modeler is pretty slow. Let's just leave it at that. modo is usable up to about 100 million triangles when working on subdiv meshes on my laptop. Modeler up to about 10 million. Yikes.



    Quote Originally Posted by rsfd View Post
    @Captain Obvious
    the difference *is* that big.
    It's due to several reasons: poor graphics drivers from Apple, poor OSX drivers from AMD and Nvidia plus poor implementation in applications like LW and others.
    OSX Lion holds newer drivers, but nearly no application is actually using it. They are all tied to the legacy OGL 2.1 drivers, which are still part of OSX for compatibility reasons.

    With a little luck, we (Apple users) will see improvements with upcoming Mountain Lion and probably rumors will be acknowledged that Apple will use standard PC Graphics cards in the future…
    I don't believe that. modo and Maya are both faster on Windows, but not much faster. I'm not saying you're wrong about Lightwave on the Mac, but I wouldn't be so quick to point the blame. Other software performs fine on OS X. Why doesn't Lightwave?


    And besides, it seems just as crazy to recommend a $4000 graphics card, when a $150 operating system would make a bigger performance difference. There's no shame in dual-booting.
    Last edited by Captain Obvious; 05-12-2012 at 02:32 PM.
    Are my spline guides showing?

  3. #18
    Registered User
    Join Date
    Jan 2009
    Location
    Norway
    Posts
    175
    On Modo and other 3D-softwares... I'm not interested in learning any other package. I'd rather spend my time on other non-technincal things, like writing... If my primary occupation was doing 3D graphics, I would be using Z-brush, Modo, Maya, Motion Builder and other high-end software... But since I had too little to do when I was 14 years old, I know this software, and thrust Newtek to develop it further.

  4. #19
    Super Member Captain Obvious's Avatar
    Join Date
    Dec 2004
    Location
    London
    Posts
    4,502
    Well, if what everyone seems to say about Lightwave's performance on the Mac is true then I have to recommend that you invest in Windows, rather than graphics hardware.

    If you dual-boot and use Windows for Lightwave and nothing else, you won't really have to put up much with Windows' eccentricities.
    Are my spline guides showing?

  5. #20
    Registered User
    Join Date
    Jan 2009
    Location
    Norway
    Posts
    175
    I'm not interested in using another operating system than OSX. I just invested in a second hand MacPro to avoid switching to Windows - it will also be used for color grading, and I just deleted my Windows partition, better used for scratch-disk area. I've been very happy with the performance until this project became complicated. I'm running a first gen 27" i7 iMac... it's quite expected to become slow at this point. Anyways, there is no reason for LightWave to under-perform on OSX, if that is the case, I hope Newtek is working on it...

  6. #21
    Super Member JonW's Avatar
    Join Date
    Jul 2007
    Location
    Sydney Australia
    Posts
    2,235
    Quote Originally Posted by Captain Obvious View Post
    What, people still use that? I haven't used Modeler for anything serious since modo 103. Hang on, I'll check... edit: Modeler is pretty slow. Let's just leave it at that. modo is usable up to about 100 million triangles when working on subdiv meshes on my laptop. Modeler up to about 10 million. Yikes
    I only have my PCs for Lightwave, so LW is pretty privileged to have them all to itself. Modeler is pretty ordinary, but for my purposes LWCad makes all the difference. But if one needs to, eg rotate a 100 layers for an architectural building, I can go & make a coffee & bake a cake!

    It would be nice if LW used more than one core, such a waste having 15 cores sitting of their hands. What does help a bit, but barely, is an SSD.
    Last edited by JonW; 05-13-2012 at 01:19 AM.
    Procrastination, mankind's greatest labour saving device!

    W5580 x 2 24GB, Mac Mini, Spyder3Elite, Dulux 30gg 83/006 72/008 grey room,
    XeroxC2255, UPS EvolutionS 3kw+2xEXB

  7. #22
    Registered User Rayek's Avatar
    Join Date
    Feb 2006
    Location
    Vancouver, BC
    Posts
    1,459
    Just found this Mac Quadro 4000 review:
    http://arstechnica.com/apple/2011/05...ising-start/2/

    Outperformed/equal performance by/with a 5770. Plain sad.
    Win10 64 - i7 [email protected], p6t Deluxe v1, 48gb, Nvidia GTX 1080 8GB, Revodrive X2 240gb, e-mu 1820. Screens: 2 x Samsung s27a850ds 2560x1440, HP 1920x1200 in portrait mode

  8. #23
    Registered User
    Join Date
    Jan 2009
    Location
    Norway
    Posts
    175
    Quote Originally Posted by JonW View Post
    It would be nice if LW used more than one core, such a waste having 15 cores sitting of their hands. What does help a bit, but barely, is an SSD.
    Are you kidding me? Is LightWave only using one core!!!???

  9. #24
    Super Member Captain Obvious's Avatar
    Join Date
    Dec 2004
    Location
    London
    Posts
    4,502
    Quote Originally Posted by mikkelen View Post
    Are you kidding me? Is LightWave only using one core!!!???
    Depends on what you're doing. Certain tasks are incredibly difficult to multi-thread, including many modelling tasks, dynamics, et cetera. The Bullet dynamics library apparently runs faster in a single thead than with multiple threads. The overhead for thread-to-thread communication is greater than the performance boost for splitting the work across many cores, I guess. While I'm sure there are plenty of areas in Lightwave that could do with a bit of multi-threading, you shouldn't get your hopes up too high. Most 3D software is mostly single-threaded. It's only really rendering that threads well, and Lightwave's render engine is fully threaded.
    Are my spline guides showing?

  10. #25
    One Man Army Andy Meyer's Avatar
    Join Date
    Dec 2009
    Location
    Lucern, Switzerland
    Posts
    1,200
    modeler is 100% outdated, but still usefull for my everyday work.
    modelers problem is not only threading. it's the architecture, data structure and the 20 years old program code.
    disassembled amiga source code combined with 16bit barrier hacks will never do a decent job on a 16core 64bit cpu ;-)
    [just kidding, but not so far from the truth i guess]

    i work all day with modeler. if you know how to work around modelers limits it's not bad for many tasks.

    LW needs a new up to date method for modeling.
    NT: just keep modeler as a legancy poly edit tool and present a modern way to model inside LW12 (layout).
    my website: www.andymeyer.ch
    Supermicro 7047A-T/X9DAi: dual E5-2687 8C/16T @3.1ghz | 64GB Ram | nVidia GTX680-4gb (301.42) | Win7 64 Bit (SP1) |
    display 30"+30"+22" | 13 rendernodes, 54 cores

  11. #26
    Super Member JonW's Avatar
    Join Date
    Jul 2007
    Location
    Sydney Australia
    Posts
    2,235
    Quote Originally Posted by Andy Meyer View Post
    modeler is 100% outdated, but still usefull for my everyday work.
    modelers problem is not only threading. it's the architecture, data structure and the 20 years old program code.
    disassembled amiga source code combined with 16bit barrier hacks will never do a decent job on a 16core 64bit cpu ;-)
    [just kidding, but not so far from the truth i guess]

    i work all day with modeler. if you know how to work around modelers limits it's not bad for many tasks.

    LW needs a new up to date method for modeling.
    NT: just keep modeler as a legancy poly edit tool and present a modern way to model inside LW12 (layout).
    Until Modeler is sped up, if one has a spare computer it would be best to have a fast CPU with less cores, & overclocked! Turn off hyperthreading & have an SSD, plus a fast graphics card.

    If you use it for rendering with HT off it will be about 25% slower, but this maybe a price worth paying if one is getting frustrated with Modeler, & Layout for that matter.


    Edit:

    I just tried the scene http://forums.newtek.com/showthread....105742&page=11 here are my benchmarks for LW11.0 1:54 (114) HT on, & 2:25 (145) HT off. (I haven't got around to updating LW11)

    So if you have a 60 minute render with HT off. It's only going to take 47 minutes with HT on. Obviously it's will vary a bit from render to render. Having HT off, is not the end of the world, & will help in other ways!

    LW 10.1: 3dspeed "teapots"

    HT on
    1:03 (63)

    HT off
    1:11 (71)

    53 minutes v 60 minutes. I going to leave HT off unless I have days of renders!
    Last edited by JonW; 05-13-2012 at 07:51 PM.
    Procrastination, mankind's greatest labour saving device!

    W5580 x 2 24GB, Mac Mini, Spyder3Elite, Dulux 30gg 83/006 72/008 grey room,
    XeroxC2255, UPS EvolutionS 3kw+2xEXB

  12. #27
    Registered User
    Join Date
    Jan 2009
    Location
    Norway
    Posts
    175
    I thought Core were supposed be/become a re-write?

    I've not used LW much the last years, but I still like it, BUT too bad that LightWave always feels/is outdated. As long as I've used LightWave, this has been the case, it's not been exactly cutting edge, maybe except for the HyperVoxels feature in LW6?

  13. #28
    Registered User
    Join Date
    Jan 2009
    Location
    Norway
    Posts
    175
    ...to Newtek's credit, LW11 is much faster than LW10! It's moving in the right direction.

  14. #29
    Registered User Rayek's Avatar
    Join Date
    Feb 2006
    Location
    Vancouver, BC
    Posts
    1,459
    Yes, I noticed - I installed the trial, and opengl in Layout is at least twice as fast compared to 9.6. VPR is also nice.

    Quote Originally Posted by mikkelen View Post
    ...to Newtek's credit, LW11 is much faster than LW10! It's moving in the right direction.
    Win10 64 - i7 [email protected], p6t Deluxe v1, 48gb, Nvidia GTX 1080 8GB, Revodrive X2 240gb, e-mu 1820. Screens: 2 x Samsung s27a850ds 2560x1440, HP 1920x1200 in portrait mode

  15. #30
    One Man Army Andy Meyer's Avatar
    Join Date
    Dec 2009
    Location
    Lucern, Switzerland
    Posts
    1,200
    Quote Originally Posted by JonW View Post
    Until Modeler is sped up, if one has a spare computer it would be best to have a fast CPU with less cores, & overclocked! Turn off hyperthreading & have an SSD, plus a fast graphics card.

    If you use it for rendering with HT off it will be about 25% slower, but this maybe a price worth paying if one is getting frustrated with Modeler, & Layout for that matter.
    JonW, if you have HT off then you can overclock better coz HT off uses less voltage!
    my website: www.andymeyer.ch
    Supermicro 7047A-T/X9DAi: dual E5-2687 8C/16T @3.1ghz | 64GB Ram | nVidia GTX680-4gb (301.42) | Win7 64 Bit (SP1) |
    display 30"+30"+22" | 13 rendernodes, 54 cores

Page 2 of 3 FirstFirst 123 LastLast

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •