Jump to content

pyramidgod

Registered User
  • Posts

    11
  • Joined

  • Last visited

Posts posted by pyramidgod

  1. Yea but now maybe I'm thinking it could be how my Akitio is powered. Do my stats look right for my 980 Ti in Cuda-Z?

    CUDA-Z Report

    =============

    Version: 0.10.251 64 bit CUDA-Z

    OS Version: Mac OS X 10.10.5 14F1021

    Driver Version: 10.4.2 310.41.35f01

    Driver Dll Version: 7.50

    Runtime Dll Version: 6.50

    Core Information

    ----------------

    Name: GeForce GTX 980 Ti

    Compute Capability: 5.2

    Clock Rate: 1190 MHz

    PCI Location: 0:10:0

    Multiprocessors: 22 (2816 Cores)

    Threads Per Multiproc.: 2048

    Warp Size: 32

    Regs Per Block: 65536

    Threads Per Block: 1024

    Threads Dimensions: 1024 x 1024 x 64

    Grid Dimensions: 2147483647 x 65535 x 65535

    Watchdog Enabled: Yes

    Integrated GPU: No

    Concurrent Kernels: Yes

    Compute Mode: Default

    Stream Priorities: Yes

    Memory Information

    ------------------

    Total Global: 6143.81 MiB

    Bus Width: 384 bits

    Clock Rate: 3505 MHz

    Error Correction: No

    L2 Cache Size: 48 KiB

    Shared Per Block: 48 KiB

    Pitch: 2048 MiB

    Total Constant: 64 KiB

    Texture Alignment: 512 B

    Texture 1D Size: 65536

    Texture 2D Size: 65536 x 65536

    Texture 3D Size: 4096 x 4096 x 4096

    GPU Overlap: Yes

    Map Host Memory: Yes

    Unified Addressing: Yes

    Async Engine: Yes, Bidirectional

    Performance Information

    -----------------------

    Memory Copy

    Host Pinned to Device: 1275.66 MiB/s

    Host Pageable to Device: 1250.38 MiB/s

    Device to Host Pinned: 1379.6 MiB/s

    Device to Host Pageable: 1355.51 MiB/s

    Device to Device: 108.758 GiB/s

    GPU Core Performance

    Single-precision Float: 7083.75 Gflop/s

    Double-precision Float: 233.783 Gflop/s

    64-bit Integer: 485.834 Giop/s

    32-bit Integer: 2430.57 Giop/s

    24-bit Integer: 1631.03 Giop/s

    Generated: Tue Nov 24 12:14:16 2015

  2. The performance issue is due to the 16Gbps-TB2 port being shared between the eGPU and RAID storage

    When playing 4k on premier you are sharing the 16Gbps TB2 bandwidth with your RAID Thunderbolt 2 G-Drive causing a massive decrease in eGPU performance.

    You can see this if you copy your movie to the onboard SSD storage, disconnect your G-Drive and then play the movie to the eGPU. There the full 16Gbps TB2 link will be provided to the eGPU.

    The bad news is Apple do not have a current Macbook product where any TB2 ports get a full exclusive 16Gbps per port. Eg: if you upgraded to say the 15" MBP with 2x TB2 ports, those ports are shared across the same bus so will still be sharing a 16Gbps (x4 2.0) upstream bus across each port. More details at https://forum.techinferno.com/implementation-guides-apple/8585-2014-15-macbook-pro-2x-r9_280x%4016gbps-tb2-netstor-na211tb-win8-1-osx-%5Bgoalque%5D-2.html#post117339

    Fixes for this shared bandwidth issue

    So then what's needed is to move your storage and eGPU on separate interfaces. Is there a RAID USB 3.0 product <strike>or can you host your image on network-attached storage (NAS)</strike>? @apple, giving no LAN port is ridiculous.

    Unless you require OSX you may want to look at other high performance TB2/TB3 notebooks with less hardware restrictions (ie: 100Gbps LAN, PCIe storage, optical drives) by HP/Dell/Lenovo/Acer. See candidates at https://forum.techinferno.com/diy-e-gpu-projects/4109-egpu-candidate-system-list.html#6th

    THANKS FOR THE HELP! I'm going to have to drop my footage on my SSD while I edit a project for now. I recently had my maxed out late 2013 15" retina stolen and got this maxed out 13" for now until the macbook with thunderbolt 3 comes out. I also tried it on my quadcore iMac with thunderbolt 1 and it was pretty much useless lol

  3. Im running the eGPU set up with the Akitio Thunder2, GTX 980 Ti SC 2.0, EVGA 850w Bronze and a Late 2013 Macbook Pro Retina maxed out for that year at 2.8 GHz Intel Core i7, 16gb RAM, 1TB SSD. I'm using this set up for Premiere and Davinci Resolve as CUDA.

    When my GPU isnt being used this is the performance in CUDA-Z

    Memory Copy

    Host Pinned to Device: 1275.23 MiB/s

    Host Pageable to Device: 1250.35 MiB/s

    Device to Host Pinned: 1380.52 MiB/s

    Device to Host Pageable: 1352.74 MiB/s

    Device to Device: 108.852 GiB/s

    GPU Core Performance

    Single-precision Float: 6971.89 Gflop/s

    Double-precision Float: 228.923 Gflop/s

    64-bit Integer: 369.856 Giop/s

    32-bit Integer: 2049.2 Giop/s

    24-bit Integer: 1443.37 Giop/s

    When I'm playing 4k on my timeline in premiere I'm getting this from CUDA-Z

    Memory Copy

    Host Pinned to Device: 759.846 MiB/s

    Host Pageable to Device: 775.32 MiB/s

    Device to Host Pinned: 1022.51 MiB/s

    Device to Host Pageable: 666.432 MiB/s

    Device to Device: 95.5938 GiB/s

    GPU Core Performance

    Single-precision Float: 7053.49 Gflop/s

    Double-precision Float: 230.088 Gflop/s

    64-bit Integer: 479.468 Giop/s

    32-bit Integer: 2318.85 Giop/s

    24-bit Integer: 1676.76 Giop/s

    It doesn't play 4k in real time in premiere. It also only exports 4k at 14-10fps avg or lower and 1080p at real time for around 24-30fps to my RAID Thunderbolt 2 G-Drive. My CPU says its at 350% when running these processes. Is the CPU just bottlenecking everything? I saw people running setups off lesser CPUs with less powerful cards and getting better playback? HELP PLEASE ANYONE!!

  4. Im running the eGPU set up with the Akitio Thunder2, GTX 980 Ti SC 2.0, EVGA 850w Bronze and a Late 2013 Macbook Pro Retina maxed out for that year at 2.8 GHz Intel Core i7, 16gb RAM, 1TB SSD. I'm using this set up for Premiere and Davinci Resolve as CUDA.

    When my GPU isnt being used this is the performance in CUDA-Z

    Memory Copy

    Host Pinned to Device: 1275.23 MiB/s

    Host Pageable to Device: 1250.35 MiB/s

    Device to Host Pinned: 1380.52 MiB/s

    Device to Host Pageable: 1352.74 MiB/s

    Device to Device: 108.852 GiB/s

    GPU Core Performance

    Single-precision Float: 6971.89 Gflop/s

    Double-precision Float: 228.923 Gflop/s

    64-bit Integer: 369.856 Giop/s

    32-bit Integer: 2049.2 Giop/s

    24-bit Integer: 1443.37 Giop/s

    When I'm playing 4k on my timeline in premiere I'm getting this from CUDA-Z

    Memory Copy

    Host Pinned to Device: 759.846 MiB/s

    Host Pageable to Device: 775.32 MiB/s

    Device to Host Pinned: 1022.51 MiB/s

    Device to Host Pageable: 666.432 MiB/s

    Device to Device: 95.5938 GiB/s

    GPU Core Performance

    Single-precision Float: 7053.49 Gflop/s

    Double-precision Float: 230.088 Gflop/s

    64-bit Integer: 479.468 Giop/s

    32-bit Integer: 2318.85 Giop/s

    24-bit Integer: 1676.76 Giop/s

    It doesn't play 4k in real time in premiere. It also only exports 4k at 14-10fps avg or lower and 1080p at real time for around 24-30fps to my RAID Thunderbolt 2 G-Drive. My CPU says its at 350% when running these processes. Is the CPU just bottlenecking everything? I saw people running setups off lesser CPUs with less powerful cards and getting better playback? HELP PLEASE ANYONE!!

×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use. We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.