Announcement

Collapse
No announcement yet.

GeForce RTX SUPER Linux Compute Performance - 18 GPU NVIDIA OpenCL Comparison

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • ermo
    replied
    Yeah, I was going to suggest RX 570/580/590, RX Vega 56, RX Vega 64 and Radeon VII too.

    I'm getting into Blender, so OpenCL performance is beginning to matter to me.

    Leave a comment:


  • Michael
    replied
    Originally posted by Qaridarium
    really not a single amd card? .... we do not even need many AMDs just the relevant ones to "buy"

    means RX590 and 5700XT and Vega64/Radeon7 ... most people of the AMD customers will buy one of these 4.

    we do not even need benchmarks of old cards like RX480/580 not even vega56 .
    Navi isn't supported by ROCm as the main cause for not testing AMD cards this round, but will come in a separate follow-up article.

    Leave a comment:


  • tildearrow
    replied
    Originally posted by Michael View Post

    Not fresh ones, when time allows I'll do some non-Navi ones.
    Non-Navi? I am surprised that ROCm still doesn't work under Navi...

    Leave a comment:


  • tildearrow
    replied
    Originally posted by Venemo View Post
    Would've been interesting to also add a couple of AMD cards to the comparison.
    I agree. NVIDIA-only benchmarks don't look interesting to me if it doesn't have AMD added to it.

    Leave a comment:


  • Venemo
    replied
    Would've been interesting to also add a couple of AMD cards to the comparison.

    Leave a comment:


  • Michael
    replied
    Originally posted by kieffer View Post
    Don't you have a Volta card to highlight the difference between the "compute" card and the "gaming" cards ?
    I've never had any Volta card.

    Leave a comment:


  • kieffer
    replied
    Don't you have a Volta card to highlight the difference between the "compute" card and the "gaming" cards ?

    Leave a comment:


  • Michael
    replied
    Originally posted by merced View Post
    Looking at these results, how should I interpret performance per watt for deep learning applications? I would love to see a PlaidML FPS per watt chart.

    OpenBenchmarking.org, Phoronix Test Suite, Linux benchmarking, automated benchmarking, benchmarking results, benchmarking repository, open source benchmarking, benchmarking test profiles

    Leave a comment:


  • merced
    replied
    Looking at these results, how should I interpret performance per watt for deep learning applications? I would love to see a PlaidML FPS per watt chart.

    Leave a comment:


  • Michael
    replied
    Originally posted by loganj View Post
    do you have these tests or at least a part of them for amd cards?
    Not fresh ones, when time allows I'll do some non-Navi ones.

    Leave a comment:

Working...
X