Announcement

Collapse
No announcement yet.

Total War: WARHAMMER III Linux Performance Across 24 NVIDIA / AMD GPUs

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

  • baka0815
    replied
    Melcar
    Thank you very much for the write-up, that helped me a lot!

    Leave a comment:


  • Melcar
    replied
    Originally posted by baka0815 View Post
    I currently have an AMD RX580, which current gen AMD card would be "the successor" to that card? Would that be an 6700?
    What is the "RX580" of the "6xxx" cards?

    Price wise? The closest card would be the 6500XT. The 6500XT is like a slight sidegrade but can be a worse card in certain situations due to the narrow bus width and it being only a PCIe 4.0 x4 card. 1080p is fine and even some 1440p, but at this point this card chokes badly. And then you have the lack of hardware encode features on the 6500XT, which makes it a worse card than the Polaris.

    A proper performance upgrade would be a 6600/50XT (or even a 6600 if you want cheaper) as that card will be able to power most games at 1080p and 1440p (not high refresh rate), just like a RX 580 can (I run an OC RX 480 myself and pay all my games at 1440p). Maybe more than what you are willing to spend for a proper "successor" to your card, as most of those cards are $380-$400 right now. It certainly is for me, hence why I'm still on a Polaris card. The 66xx cards are PCIe 4.0 x8, so make sure your motherboard and CPU are PCIe 4.0 compliant.

    I plan to wait for RDNA3 release (end of the year I guess) and see how prices adjust. Maybe I can snag a 6650XT or even a 6750XT for cheap as I don't really care for the latest tech (and Linux always takes a few months to catch up to hardware releases), or maybe the new RDNA3 cards will have more sensible pricing (fat chance). AMD's efforts in the budget tier was lackluster this gen, to put it mildly. There is a huge price/performance gap between the 6500XT and the next tier up. They totally ignored the sub $300 market and only did low effort releases (repackaged mobile 6500XT and 6400 cards) just to have something to show.
    Last edited by Melcar; 29 June 2022, 12:30 AM.

    Leave a comment:


  • baka0815
    replied
    I currently have an AMD RX580, which current gen AMD card would be "the successor" to that card? Would that be an 6700?
    What is the "RX580" of the "6xxx" cards?

    Leave a comment:


  • BHZeto
    replied
    Amazing, this is more interesting than the game itself (sadly for the game).
    Does somebody know how these results compare to same system running TWW2?

    Leave a comment:


  • Jahimself
    replied
    Hello Michael, if you read this comment, was Smart Access Memory activated on this benchmark?

    The result for AMD cards are faster than on windows, while Nvidia are slower in symetrical ways @1080p, if we compare with the data from Kitguru, who use the same CPU for their benchmark.

    But when we increase the resolution at 1440p Nvidia card are almost equal to windows numbers, AMD are way faster in the meanwhile.

    At 4K Linux takes the lead even for Nvidia. Note for instance that the 6800 XT is at 36fps average on windows, while reaching 45fps on linux! Maybe some specificities of your setup induce those results. It might be way too time demanding, but it would be interesting to do those benchmarks on the same machine for both OS, with like 4 RX cards from AMD and 4 RTX cards from Nvidia, to see if really Linux perf are slapping Windows with this port.

    Thanks a lot once again for those data Michael!

    Leave a comment:


  • kayosiii
    replied
    The power consumption at exactly 100 watts without any variation with the rx 6600 looks very suspicious. I am willing to bet that the setup is limiting the card to that amount of power draw which may affect the other benchmark scores.

    Leave a comment:


  • hoohoo
    replied
    The graphs main text - title, gpu names - are showing up as very offwhite, nearly indistinguishable from the background. At least this is the case on my Samsung S10/Opera setup.

    Leave a comment:


  • Jabberwocky
    replied
    Originally posted by Volta View Post

    Completely not true. Linux owns Windows in CPU benchmarks and there are still more people using Windows. It seems people are afraid of change or they're just dumb.
    Yes it must be afraid to change or dumb, it can't be anything else. /s

    People want things that just work. That just isn't the case with Linux:

    1) https://www.protondb.com/explore?sort=mostBorked (over 1000)
    2) https://www.youtube.com/watch?v=Rlg4K16ujFw (Gaming on Linux is NOT Ready... - Daily Driver Challenge Finale)

    I've been a big Linux fanboi since Loki Entertainment and LinuxGamers.net ... it's been over 20 years that I've supported and donated to various projects and developers. We have come a long way in terms of gaming and general system use. If there's one thing that I have learned over the past years is that supporting hardware vendors (middleman) is more important than supporting game developers. It's wonderful to see developers port games like Total War: Warhammer III and I will continue to buy basically every Linux game that is published but I've seen studios come and go.

    If you really want people to use Linux then you need to support companies that ship Linux preinstalled System76 and Steam Deck are the best options at the moment IMO. Alternatively Dell, Lenovo, or others Asus/Slimbook/Purism(old hardware)/Pinebook(Arm) who provide official Linux support for their hardware. You need the entire system to work out of the box and work really well before you can really start to market the products.

    Leave a comment:


  • aufkrawall
    replied
    Originally posted by Volta View Post
    Completely not true. Linux owns Windows in CPU benchmarks
    Usually only when imbeciles run benchmarks who don't notice obviously botched results.

    Leave a comment:


  • birdie
    replied
    Originally posted by bug77 View Post

    I mean, Linux using far less VRAM could be an indication it's not doing the same work as it does on Windows. If, for example, there's a mipmapping issue, the Linux version would need fewer maps in VRAM.
    But it's just my guess, I couldn't say what was really going on over there.
    That makes sense, thanks for the clarification.

    Leave a comment:

Working...
X