No announcement yet.

convertible/tabletpc to use also with Resolve basic editing and converting 4K

  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    Stelvis, yes, I agree, absolutely speaking.

    But maybe you missed my point that is relative speaking:
    I would compare for example the video card in a Surface Book with the card in a workstation laptop to a card in an enclosure TB3 box.
    It's a different scenario as normal desktop comparisons because in a laptop world it's not sure we can have the same PCIe3.0 16x bandwidth that we have in desktop.

    But my point was this:
    is it better a Nvidia 970 on a 16x Pcie 3.0 or the best existent monster Quadro on a 4x Pcie3.0 with equal memory on board?
    We know for sure that Resolve use all available CUDA or OpenCL's cores not like games.

    BUT is the CPU/system able to use ALL the pcie lanes? I doubt it, (but I may be wrong).
    IF it's not capable of using well all the lanes, the Quadro could be more powerful of one 970 (put in 16x3.0) EVEN if put on a 4x PCie3.0 lane (that is 8x Pcie 2.0).
    I mean, bandwidth and GPU power are 2 different things. Nobody doubt that a powerful board is a lot better, BUT if put in lower lanes?
    Obviously this is nonsense in a normal environment: you will choose a better card AND put in a 16x... Here the problem arises because we are comparing those eventually using an external TB3 (that is misleading at first, because they claim a pcie 16x BUT it's only the physical connector, not the real bandwidth lanes that are only 4x!).

    This video is dated but it shows well the bandwidth problem using 3dMark:
    it's ONLY a 0,6% of difference between Pcie2.0 8x and 16x!!

    To compare pcie 3.0, x8 or x16 is ininfluent (yes, on games...):

    And Pcie3.0 4x (TB3) is equal to Pcie2.0 8x, so making maths one should conclude: 3.0 16x quite= 3.0 8x --> 3.0 8x = 2.0 16x --> 2.0 16x has only a 0,6% bandwidth improvement on 2.0 8x --> 2.0 8x = 3.0 4x = TB3
    So TB3 quite= pcie3.0 16x
    IF this is true an external TB3 could be compared to fixed GPU laptops. Otherwise not.

    This is the theory, I would love to be proved wrong with the practice of someone using Resolve with these different systems...
    To me it should very useful to have a Blackmagic utility to know which GPU power we can have In Resolve at the same way they give us the disk speed test. Not only for the GPU cores, but also for different pcie lanes with the same GPU. OR if 3dMark suffices, because simulates how Resolve will use the card (not like games), it could be useful find out some comparisons with different lanes connection of the same GPU in Pcie 3.0.
    Last edited by frank10; 11-24-2016, 10:43 AM.


    • #17
      If what I said up is true, there is another important consequence interesting:
      if a video card inside a desktop 3.0 16x is the same of 3.0 4x in external box TB3, then one could buy a great video card and put in the external enclosure and then use it both @studio connected with TB3 on desktop pc and when going @outside connected with TB3 to laptop or ultrabook. Of course, both must have TB3 port enabled to use external GPU.

      You will have:
      1) the same great performance in both pc and laptop
      2) you buy only 1 video card instead of 1 + pricey laptop with fixed GPU (anyway inferior of TB3 box)
      3) you save money on laptop without a pricey GPU (anyway with good CPU and RAM)

      Not bad, IF this theory is true!


      • #18
        HP and Lenovo makes crappy laptops. I'd scratch them off my list unless you're looking at the HP Top of the line z-book or the high end Lenovo Thinkpad. The Surface Book, all decked out model, is a nice option even without the usbc or tb3. Those ports are only needed if you are future proofing your purchase. But if you need it today, it's a good option.


        • #19
          Guys remove my fucking account right now you fucking bitch


          • #20
            Guys remove my fucking account right now you fucking bitch