Navi RDNA vs GCN 1.0: Last-Gen vs Next-Gen GPU Tech Head-To-Head!

Discussion in 'Video Cards' started by PontiacGTX, Sep 11, 2019 at 6:56 PM.

  1. crazycrave

    crazycrave Limp Gawd

    Messages:
    436
    Joined:
    Mar 31, 2016
    That's interesting stuff your working with and being AMD since HD 4890 XT I managed to not sale two of my cards R 9 - 280 /290x both Tri X models which I have used the 290x with my Ryzen 5 3600 as to test DX 12 and I noticed right off how different Hawaii looks in Image Quality vs Polaris and Navi as to me Hawaii looks better .. but anyways there are a lot people still running Hawaii and older cards as I have requests for more 3600 / 290x videos on you tube as I need to rip RX 5700 out for that to happen or build my 3rd 3600 platform for older cards or maybe 3700x this time .

    I enjoyed the video .
     
  2. Stoly

    Stoly [H]ardness Supreme

    Messages:
    6,327
    Joined:
    Jul 26, 2005
  3. sirmonkey1985

    sirmonkey1985 [H]ard|DCer of the Month - July 2010

    Messages:
    21,449
    Joined:
    Sep 13, 2008
    interesting way of looking at and comparing the architectures.. thanks for posting this.
     
  4. cyklondx

    cyklondx Limp Gawd

    Messages:
    254
    Joined:
    Mar 19, 2018
    I ran image quality comparison between 2 monitors (same models), and couple of older games - like farcry, quake 4, doom3 (possibly one or 2 more)
    Radeon x1950 Pro 512M AGP vs Fury-X

    Not to mention Fury-x was not displaying some effects at all, it looked quite meh, and low quality poly, textures vs what I saw on lower resolution with x1950 Pro card.
    1024x768 or 1280x768 were giving me same look in jaggies etc vs 1080p 1440p on fury-x

    In terms of image quality, i think most of it is dependent on your internal settings (as they do differ by defaults on some cards) i.e. my old 290x had defined to high quality vs performance, so you may not necessarily getting what you get.

    watching the vid now.
     
    N4CR likes this.
  5. N4CR

    N4CR 2[H]4U

    Messages:
    3,834
    Joined:
    Oct 17, 2011
    I'd really love to see an outlet tackle this IQ testing again. Have heard a few say there is no difference these days but it would be good to see evidence-based testing. I don't have my 290X for now so can't test between Vega and 290X (which is considered to have great IQ) or I'd volunteer for duty!
     
  6. Ready4Dis

    Ready4Dis Limp Gawd

    Messages:
    501
    Joined:
    Nov 4, 2015
    Ive got a 280x, fury and 560/570 still, but no Vega or anything newer.
     
  7. cyklondx

    cyklondx Limp Gawd

    Messages:
    254
    Joined:
    Mar 19, 2018
    when you make screenshots on both x1950 or fury-x it looks the same (farcry)

    its only how its being displayed (scaling) on the monitor.

    both @ 1080p (both using same config file.)

    here's x1950 (windows XP sp3)


    fury-x (windows 10)


    // Ready4Dis
    This should be easily reproducible scene from farcry. I advise getting it on other cards. (I can also get it with radeon vii)


    in other games like sw republic commando well you can see far more changes in how the graphics are being rendered.



    when you try to compare the ss, you can see that 1950 has better water over distance, while fury-x renders things more smoothly. || Thats why i think it has to do something with scaling, and how it understands colors - and how it shows up on monitor vs whats actually being rendered. (both cards were set with Full RGB), but x1950 had better colors, and fury-x looked washed out vs old radeon.
    (on 980ti it looks even more washed out > even with full-rgb, but screenshot wise its like the same;)

    i have sold most of my old gpu's, but i recommend testing following gpu's to compare:
    nv900
    ati7970
    nv480
    ati6950|6970
    nv285
    ati5870
    nv7600 (GS or GT since they were G73 same as PS3 used) | x1950


    I picked those gpu's are those were interesting gpu's - as the way code is executed by those cards should be at greatest difference between each other.




    I think we would need to go old school, and get a synchronized camera with monitor and film the results as we see it.
     
    Last edited: Sep 13, 2019 at 11:14 AM
  8. Ready4Dis

    Ready4Dis Limp Gawd

    Messages:
    501
    Joined:
    Nov 4, 2015
    I don't have farcry, and it would take me a significant time to download since all I have available is satellite internet.
     
  9. cyklondx

    cyklondx Limp Gawd

    Messages:
    254
    Joined:
    Mar 19, 2018
    what do you have that can be ran on older gpu's and newer systems?
     
  10. Ready4Dis

    Ready4Dis Limp Gawd

    Messages:
    501
    Joined:
    Nov 4, 2015
    No clue, rocket league :). I'll have to check as I put a new SSD and wiped just about everything. My son has a few games installed on his desktop I could test with, just have to see which ones.

    Edit:. Just so you know I tend to do more media encoding and software development than gaming, hence the lack of games.
     
  11. cyklondx

    cyklondx Limp Gawd

    Messages:
    254
    Joined:
    Mar 19, 2018
    that would be outside of my realms, and not good test.
     
  12. Ready4Dis

    Ready4Dis Limp Gawd

    Messages:
    501
    Joined:
    Nov 4, 2015
    Yeah, I know it's not a great test which is why I put my smiley. I will try to find out what other games I have available for some testing. I think my GTX 470 is to far gone (corrosion on the circuit board) to be able to test.
     
  13. crazycrave

    crazycrave Limp Gawd

    Messages:
    436
    Joined:
    Mar 31, 2016

    I noticed playing Strange Bigrade DX 12 under 1080p Ultra as Hawaii looked different in how it's front end handles DX12 as It was the first to support it on AMD 's side before color compression tech came out .. so it does not use it in DX 12 renders .
     
  14. GoodBoy

    GoodBoy [H]ard|Gawd

    Messages:
    1,452
    Joined:
    Nov 29, 2004
    Did the tester use same input type on both monitors? (Haven't watched video yet, at work).

    I know with nVidia, the HDMI output defaults to lower bit color setting (or used to on older drivers) on the HDMI output only, but you can increase it in the control panel and set that to restore the color appearance.

    Maybe there are some output difference, even if set to full RGB, between say an old DVI port and some newer port like displayport. The newer types (I believe) have greater bandwidth.

    You might also try swapping the displays after the first test to make sure its not the display.

    Edit: Could this also possibly be a side effect of the color compression on the newer arch? Thought those compression techniques were lossless...