Announcement

Collapse
No announcement yet.

20-Way AMD / NVIDIA Linux Gaming Benchmarks For The 2018 Holidays

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

  • Michael
    replied
    Originally posted by yurikoles View Post
    Michael, I think this explanation must go article.

    BTW, you ignore 1080p in latest tests, so it may be still valuable for those who stay on FullHD, which is most popular resolution according to Steam (64.29%).
    It all depends what I am testing in terms of games and hardware. With many games when benchmarking at 1080p on cards like the GTX 1080 and newer or even RX Vega, they are simply CPU bound at 1080p.

    Leave a comment:


  • yurikoles
    replied
    Michael, I think this explanation must go article.

    BTW, you ignore 1080p in latest tests, so it may be still valuable for those who stay on FullHD, which is most popular resolution according to Steam (64.29%).

    Leave a comment:


  • Michael
    replied
    Originally posted by yurikoles View Post
    Michael you ignored my comment. The RX 560 is lost in perf/$ chart, despite fact it was in all test results.
    Not ignored, just didn't see all comments (especially with my time really being a mess the past few days cycling frequently between hospital to visit wife and working...) But the reason I didn't include the RX 560 is that since in many of the tests it performs rather slow (below 60 FPS). So including it in those perf-per-dollar graphs would likely show it being out in front for value, which it is technically, but not too viable really for most Linux games given its rather slow performance. So for the dollar graphs basically included the modern mainstream/mid-range cards and better.

    Leave a comment:


  • yurikoles
    replied
    Michael you ignored my comment. The RX 560 is lost in perf/$ chart, despite fact it was in all test results.

    Leave a comment:


  • yurikoles
    replied
    Michael it seems like RX 560 is lost in perf/$ charts.

    Leave a comment:


  • Michael
    replied
    Originally posted by mitcoes View Post
    Is it not time to benchmark COMBOS?
    Freesync + AMD vs G-sync + Nvidia
    And their combo price range and performance per dollar in combos
    At least until there will be crossed drivers, if they ever be

    As Freesync monitors are cheaper ...
    would not be AMD GPUs in advantage when you are going to purchase a combo for gaming
    in 1080p as 4k is still not for gaming?
    I have a few FreeSync 4K displays and will talk more about them when the DRM bits are merged. I have no G-SYNC displays nor plans/budget to get any anytime soon.

    Leave a comment:


  • nuetzel
    replied
    Originally posted by bofh80

    In case you missed some of the random information that came after. 4.19 fixed the clock gate issue for me. i had to remove the featuremask as it was causing graphical issues. But the wattman output now seems to suggest a different mask to what I was using.
    Hello bofh80 and gurv,

    please have a look @ amd-staging-drm-next
    commit #29bf84e646c45b3370b640d44c0ce454ecb845c3
    drm/amd/pp: handle negative values when reading OD

    Leave a comment:


  • mitcoes
    replied
    Is it not time to benchmark COMBOS?
    Freesync + AMD vs G-sync + Nvidia
    And their combo price range and performance per dollar in combos
    At least until there will be crossed drivers, if they ever be

    As Freesync monitors are cheaper ...
    would not be AMD GPUs in advantage when you are going to purchase a combo for gaming
    in 1080p as 4k is still not for gaming?

    Leave a comment:


  • gurv
    replied
    Originally posted by bofh80
    Is there any proper write up somewhere for getting amdgpu with a rx 570 (or similar model) up and running on fedora 29 (28 at worst) ?

    I've followed that and verified by grub boot entries. amdgpu.ppfeaturemask=0xffffffff amdgpu.dc=1 amdgpu.dpm=1
    For some reason my card is stuck on the 300mhz entry. I've used watman-gtk (hence ppfeaturemask) to verify. I haven't tried to apply any wattman settings as I don't want misconfigure anything further. I'm assuming on default values clock gating should work out of the box.
    I've had the exact same problem on my RX 470.
    When using the feature mask boot parameter clocks are stuck at 300MHz on 4.18

    Thanks to a good fellow on these very forums, the workaround seems to be :
    echo "1" | sudo tee /sys/class/drm/card0/device/pp_sclk_od
    echo "1" | sudo tee /sys/class/drm/card0/device/pp_mclk_od
    And voilĂ , your card should be running normally.

    You can than overclock through p-states:


    FYI I've been able to push my RX 470 to 1350MHz/1750MHz at stock voltages.
    Didn't try to go higher yet.

    Leave a comment:


  • Jabberwocky
    replied
    I feel spoiled for having so many cards with open source drivers are hitting > 60FPS on 1080p. Loving it!

    Leave a comment:

Working...
X