Listen up

Viewing 15 posts - 16 through 30 (of 32 total)
  • Author
    Posts
  • #209417
    -k
    Participant

    Once again of course you can compare… what’s the problem with that? I did’nt say they are the same, cause a comparison would be rather boring in that case.
    They are really different that’s why you have to compare!

    I totally agree with demand on realtime playback but you can do the same with combustion. After rendering to memory you put the contents of the cache to disc and that’s it… you know stones are not the only highspeed disc arrays in the world… we just got ab Xraid on a g5 which does 2k playback. I don’t like them particulary but they work.. granted I would not want to work on lots of footage with this in a supervised session. But for short sequences it does the job…

    What does bandwith help if your CPUs are too slow to deliver… The only discreet product which does alot in realtime is lustre which runs on a windows PC and apparently not even with StoneFS on the raid (anybody knows this? from what I understood it’s actually NTFS???)… so much for the bandwith…

    And now with the increasing power of GPUs already or soon to be used for real realtime image processing…

    I think discreet has to look at these trends and better make toxic a real good thing…

    -k

    #209423
    eltopo
    Participant

    I think for the prices of a FFI you should get more. The speed issue seems to be the selling point here, but it is very limiting into what it can do. For example about colour grading, the FFI doesn’t work with 16-bit or 32-bit images, just 12-bit. Also there is the question that you need those plug-ins to do a lot of things and the rendering of those takes some time and with the price of burn…

    Another point has to do with the fact that if discreet wanted they could develop their systems to work with different computers that would give you a cheaper solution. An SGi machine is not faster than a Xeon, Opteron, Itanium or Power based computer system. The fact is that when you look at the numbers it shows how old the MIPS-SGI technology really is and how sticking with old technology is hurting FFI and the reason is very simple. Each day the gap closes. First nothing could do what the FFi did, nowadays the difference is only about speed. But will that be an issue tomorrow?

    I wouldn’t count on that 😕

    #209419
    Anonymous
    Inactive
    eltopo wrote:
    Another point has to do with the fact that if discreet wanted they could develop their systems to work with different computers that would give you a cheaper solution.

    As has been pointed out in these forums countelss times, the real cost of FFI is the software and support, not the hardware. Do you seriously think that when/if discreet changes the underlying platform for FFI, that they will/would pass any savings in hardware costs on to the customer? Think again. It wouldn’t give customers a cheaper soluton, but discreet’s profit margin might increase. Also keep in mind that discreet probably doesn’t pay what you or I would pay for new SGI hardware.

    Since SGI workstations are such a niche market, SGI is more likely to offer special discounts to partners like discreet than some PC vendor who sells millions of boxes a year. Discreet is a big feather in SGI’s cap, and I’m sure they price their hardware attractively to discreet in order to keep such a highly visible marketing opportunity. Also, any PC configuration that discreet would consider as a host for FFI would be a top-notch workstation configuration (dual Xeons/Opterons, Quadro FX 4400, etc.). The cost of disk arrays and audio/video I/O hardware would remain more or less the same, so in the end the difference to discreet in hardware costs between the SGI boxes or high end PCs is negligible.

    I would be very surprised if two years from now discreet is selling flame on PCs for under $150K. If people are willing to pay $200K or more, why would discreet lower their prices, regardless what box is lurking under the hood?

    -zolo

    #209425
    patdawg
    Participant

    This thread is getting ridiculous. So what if combustion can render a spark a little faster than flame on an Octane? It can’t render faster on my dual-Xeon machine than it does in flame on my 4P Tezro. I’ve got both, and I’ve compared, and the SGI is, in fact, faster. Not just on effect rendering speed, but on disk throughput, and interactivity as well. How many of you have a PC, or Mac with a disk array that can push sustained 1.6Gbit worth of data straight to four processors, on two of four independant PCI-X busses? If I added a second disk array and controller I’d be able to push 3.2Gbit…without a hiccup. At that rate you can do uncompressed 3K clips…without proxies. The answer is no one…there isn’t a PC, or Mac in existence that can pull that off. Nor do any of the graphics accelerators out there come even close in realtime OpenGL performance when compared to the SGI systems. I can load 100 layers of uncompressd NTSC clips into the DVE in smoke 6.5.3 and interact with ANY of them in realtime. You can’t do that with ANY desktop app…with ANY graphics accelerator. I’ve got an Nvidia Quadro FX4000 in my combustion workstation, and I’ve tried it.

    #209412
    majik
    Participant

    You hit it on the head Patdawg, as I said earlier its all about bandwidth not CPU speed (which is where sparks get their power from). Bandwidth, bandwidth, bandwidth, they don’t call them supercomputers for nothing.

    #209407
    luc
    Participant

    Humm…

    i have seen running 4k+ at 5 place at IBC on pc system and most of the time over xp 32bits: Assimilate scratch, nucoda, lustre, flint linux, iridas.

    Why di you said a pc can do that ? I know you can get a better score on bandwidth over sgi but do you need them… For my part a prefere a good mix over bandwidth and cpu power.

    Discreet are not moving is solution over pc box ? License and diskarray. that what you paid for your system. The sgi box cost nothing.

    luke
    Waiting for toxic

    #209426
    patdawg
    Participant

    Graph, you did NOT see any of those PC based systems running realtime uncompressed 4K. All of them were playing proxies.

    graph wrote:
    Humm…

    i have seen running 4k+ at 5 place at IBC on pc system and most of the time over xp 32bits: Assimilate scratch, nucoda, lustre, flint linux, iridas.

    Why di you said a pc can do that ? I know you can get a better score on bandwidth over sgi but do you need them… For my part a prefere a good mix over bandwidth and cpu power.

    Discreet are not moving is solution over pc box ? License and diskarray. that what you paid for your system. The sgi box cost nothing.

    luke
    Waiting for toxic

    #209431
    chrise
    Participant

    >Graph, you did NOT see any of those PC based systems running realtime uncompressed 4K. All of them were playing proxies.

    I’m pretty sure that the Baselight system does playback realtime
    4K – but it does use 8 pcs to do this:
    A quote from the filmlight website:

    In contrast to other parallel processing platforms, the multi-node Baselight architecture provides true parallel operation of all hardware components. Each node processes a strip of the input frame and a proprietary video engine then combines the processed strips to form a single frame. This method of parallel processing removes the latency inherent with other approaches and allows the full power of each node, including its graphics processor, to be used all of the time.

    Chris

    #209408
    luc
    Participant

    Hi

    i agree witht you, what i have seen is realtime rezise or crop via fx4000 pixels shading engine + 3Dlut visualization.

    That look ok for me and that fit my budget too.

    luke


    fx3000 flint
    fx4000 flame ?

    #209427
    patdawg
    Participant
    chrise wrote:
    >Graph, you did NOT see any of those PC based systems running realtime uncompressed 4K. All of them were playing proxies.

    I’m pretty sure that the Baselight system does playback realtime
    4K – but it does use 8 pcs to do this:
    A quote from the filmlight website:

    In contrast to other parallel processing platforms, the multi-node Baselight architecture provides true parallel operation of all hardware components. Each node processes a strip of the input frame and a proprietary video engine then combines the processed strips to form a single frame. This method of parallel processing removes the latency inherent with other approaches and allows the full power of each node, including its graphics processor, to be used all of the time.

    Chris

    Wow, and it’s cheaper than flame too! NOT!!! The thing starts at around $80,000 for plain old Baselight on one CPU…to get realtime 4K you have to get Baselight Eight…which is about eight times the cost…just for the system, not including storage. At that price flame is a hell of alot cheaper, and does a hell of alot more. If you just wanted grading you would save assloads of money if you got Lustre.

    #209420
    Anonymous
    Inactive
    patdawg wrote:
    chrise wrote:
    >Graph, you did NOT see any of those PC based systems running realtime uncompressed 4K. All of them were playing proxies.

    I’m pretty sure that the Baselight system does playback realtime
    4K – but it does use 8 pcs to do this:
    A quote from the filmlight website:

    In contrast to other parallel processing platforms, the multi-node Baselight architecture provides true parallel operation of all hardware components. Each node processes a strip of the input frame and a proprietary video engine then combines the processed strips to form a single frame. This method of parallel processing removes the latency inherent with other approaches and allows the full power of each node, including its graphics processor, to be used all of the time.

    Chris

    Wow, and it’s cheaper than flame too! NOT!!! The thing starts at around $80,000 for plain old Baselight on one CPU…to get realtime 4K you have to get Baselight Eight…which is about eight times the cost…just for the system, not including storage. At that price flame is a hell of alot cheaper, and does a hell of alot more. If you just wanted grading you would save assloads of money if you got Lustre.

    No, you’re missing the point! PCs are cheaper! That means that big bad discreet is fleecing its customers by making them buy expensive proprietary old slow SGI systems! The (not so) hidden subtext here:

    SGI=bad
    PC=good

    Discreet’s engineers can’t possibly know better than discreet’s customers what hardware will work best for their systems. Plus, using SGI hardware makes it too hard for all of the kewl d00ds to share it along with their other warez on BitTorrent.

    Seriously, some people will never be happy until flame costs $500, runs on a $200 Celeron PC and does realtime 4K with a single SATA drive. You can’t argue with these people, and you can’t please them. Just look at the combustion forum–everyone wants flame-level performance from a $1000 software application.

    Is FFI perfect? Of course not. Could discreet improve the price/performance ratio of its systems? Probably. Are people who think that FFI is expensive because of the SGI hardware severely deluded? Certainly. Until a competitor comes along that can offer FFI performance for half the price of discreet’s systems, discreet has no incentive to lower its prices. Porting FFI to PCs won’t change that. Smoke and flint are available on Linux PC hosts now, but they are limited in resolution comared to the SGI versions, and they’re still not at a price point that most single users could afford. Trust me, even if discreet ports everything to PCs, there will still be people bitching about how expensive they are.

    -zolo

    #209432
    Anonymous
    Inactive

    Full disclosure – I work at Filmlight

    A data point (since some of the quoted figures aren’t quite right)

    A Baselight Eight with 12 Terabytes disk delivers 2.5Gbytes per second from its disks. This is equivalent to approx 2 x 4K streams uncompressed, no proxies. Cost of this system including disks and software is $460K. Of course this isn’t the whole story, it also matters what can be done interactivly with this massive amount of data, what the client attend expectation is and whether the application (in our case grading) needs such high sustained data rates.

    #209424
    wolf
    Participant

    I love the fxguide forums because of the democratic aspects and a general
    respect for artists both very experienced and starting out as junior compositors.

    I’ve seen and participated in quite a few arguments of epic proportions (always the “my hardware is doper then yours !” kind of.)

    don t misunderstand me i love the subject but there moments when some
    of the valued posters forget or disregard the economic environment most high end unix-powered workstations plus the beloved apps are positioned in.
    when clients pay upward to $1200.- an hour and it s not all ending up
    in our (FFI artists) pockets you cannot afford to rely on anything but a
    box that will take allot of beating and still run when all other mammals have
    dissapeared from the surface of this earth.
    I never read postings with complaints about the price of a spirit datacine,
    and why is it more expensive then let s say a Bosch or Rank.

    this is going to sound very repetetive but i feel the urge to say it:
    so you not really pay for software, you pay for the history of R+D, bug fixes and blood and sweat that lead to flame 9.2.
    i would not compare combustion to flame and ask why is flame so much
    more expensive until i start my own studio and have to get jobs out the
    door that barely cover my overhead.
    so it really depends from what position you argue from.

    small independend outfits will of course tend to defend combustion and
    after effects and bash IFFS.
    i hope in the near future to see a smoother blend and more options
    between the Inferno/Flame world and the true native desktop apps like
    shake and combustion.

    i think autodesk/apple are already paying attention to the demand for a more flexible choice of tools and therefore listening to ALL of you.
    i think very soon everybody will get the solution he/she seeks for the
    price they wish to pay.

    Wolfgang Maschin

    flame* | shake | softimage

    “It seems you have been living two lives, Agent P. In one life, you are Peter Smith, assistant cook at a Jack in the Box in Barstow….in the other…you go by the chat alias “Randerson”…spreading propaganda, lying, and being a generally immature pest…

    #209430
    Anonymous
    Inactive
    patdawg wrote:
    chrise wrote:
    >Graph, you did NOT see any of those PC based systems running realtime uncompressed 4K. All of them were playing proxies.

    I’m pretty sure that the Baselight system does playback realtime
    4K – but it does use 8 pcs to do this:
    A quote from the filmlight website:

    In contrast to other parallel processing platforms, the multi-node Baselight architecture provides true parallel operation of all hardware components. Each node processes a strip of the input frame and a proprietary video engine then combines the processed strips to form a single frame. This method of parallel processing removes the latency inherent with other approaches and allows the full power of each node, including its graphics processor, to be used all of the time.

    Chris

    Wow, and it’s cheaper than flame too! NOT!!! The thing starts at around $80,000 for plain old Baselight on one CPU…to get realtime 4K you have to get Baselight Eight…which is about eight times the cost…just for the system, not including storage. At that price flame is a hell of alot cheaper, and does a hell of alot more. If you just wanted grading you would save assloads of money if you got Lustre.

    I think Andy quoted some number for the Baselight8 later in this thread so I will not waste time talking cash rather than to say the numbers you’re quoting are wrong.

    As far as buying a Lustre and saving lots of money, that’s actually quite inaccurate. To begin with Lustre isn’t even in the remote area of Baselight in terms of performance. For example most of the grading is being done in 1k proxy mode as opposed to full 2k for small canvas shows. Forget 4k in real time. The ( is greading mulitple streams of 4k in real time (no proxies needed). The 8 nodes of the Baselight8 are (now) dual dual-core AMD and capapable not only of a tremedously higher throughput playback wise but also of preocessing compared to the Lustre. To combat this edge Autodesk is moving forward with thier incinerator rendering engine which is essentially a pre-emptinve burn rendering farm attached via inifiband. While this will help keep the rendering times down on the end it is a very different animal to the online/active processing scheme you find in a baselight8 or 4. The full number-crunching capability of all 8 nodes is always at your finger tips to perform whatever task is at hand.

    So they are very differnet solutions. A fully configured Lustre with Incinerator is more or less the same price as a Baselight8 so the price discussion goes out the window.

    If you want to save assloads of money you buy the right tool not the hype. For exmaple buying a Flame to grade a feature film is a joke. Try this little test. Go in to the Colorwarper with 4k 10log footage. Push play and try to adjust the gamma. Likewise you wouldn’t want to do sfx on a baselight. For that matter try doing that in Lustre.

    I find it ironic how much mis-information there is out there.

    As far as the discussion at hand, given that Disrcreet has just announced that Inferno is shipping in japan using PC hardware and a similar Incinirator architecture and the end of life of Tezro it would appear that the cluster concept is winning ground and that commodity hardware will replace the old big iron.

    Best,
    Chris

    #209433
    Anonymous
    Inactive

    The Burn is capable to act as the Incinerator in the Inferno for Linux, not only a render software, but like a cluster?

Viewing 15 posts - 16 through 30 (of 32 total)
  • You must be logged in to reply to this topic.
Copy link
Powered by Social Snap