Does anyone have benchmarks on how much better m2 is at gaming?

Specifically I’d like to see benchmarks for native games, and rosetta games like frostpunk, which seem to be cpu bound quite heavily.

I have found lots of individual benchmarks, but no direct comparison.

  • sparky@lemmy.world@lemmy.worldMEnglish
    0·
    2 years ago

    I haven’t found anything definitive. I do have both an M1 Max MacBook Pro and a brand new M2 Ultra Mac Studio so I am happy to post some results with whatever title. I know it’s not 1:1 but given that an M2 Ultra is basically a “double M2 Max”, taking half the framerate of that may be a proxy for M2 notebook performance.

    At any rate I believe Crossover + Rosetta is going to be too much of a bottleneck to see significant gains. I’ve noticed for example, Tropico 6, when running under Metal on my M1 Max MBP, gets a smooth 70-75 FPS playing on Ultra at 2560x1600. This is Rosetta running a macOS x86 and Metal binary however. Trying the Windows version under Crossover, it drops down to 30-35 FPS playing on Medium. So while playable, even if you added a lot more M2 horsepower, I suspect the translation layers would eat a large part of that gain.

    • damipereira@lemmy.worldOPEnglish
      1·
      2 years ago

      Can you try frostpunk for example? I know it felt very slow on my m1 max, specially when temperature drops (in-game, which happens every few in-game days). It runs at around 30fps. What is weird is that it runs the same on a base m1, as an m1 max, according to : https://www.applegamingwiki.com/wiki/Frostpunk.

      I think my doubt is where the real bottleneck is, is it just cpu-bound because of translation? Is it single-core or multi-core bound? Or is there a technology limitation that no matter how much more cpu you throw at it, there’s just something about translation timing that does not make it go faster.

      If m1 and m1 max perform the same, then it might come down to single-core performance, and since m2 has some single core gains, it might provide an improvement?

      • sparky@lemmy.world@lemmy.worldMEnglish
        0·
        2 years ago

        Just gave it a spin. I would call it extremely playable, getting 45 to 55 FPS running at highest settings (Very High) and I play at a pretty intensive resolution of 5120x2160, ultra wide bigger than 4K. So I’m quite happy with the performance.

        • damipereira@lemmy.worldOPEnglish
          0·
          2 years ago

          Nice, looks like there’s no artificial limitation, and it’s just a matter of horsepower. I’d love to know the difference between m2 max and m2 ultra on this, since if it’s cpu bound they should perform the same (a game won’t use that many extra cores)

          • sparky@lemmy.world@lemmy.worldMEnglish
            0·
            2 years ago

            Yeah I wouldn’t expect the difference between 12 and 24 CPU cores or whatever it is would be significant here. But Apple also doesn’t advertise the clock speeds and such anymore so it’s hard to reason about the single-core performance difference, if any at all. Maybe a browse through Geekbench results could shed some light. Unfortunately I don’t have an M2 Max machine around to test it, but I could give it a spin on my M1 Max MBP if that’s helpful to you.

            • damipereira@lemmy.worldOPEnglish
              0·
              2 years ago

              That’s the one I have! And it only runs to around 30fps, with marked slow downs when the temperature drops, it’s barely playable. I’m not going to upgrade to an m2 max macbook, but I was just hoping a possible m3 max might be good enough for the type of games I’m interested and that I would not need a gaming pc, so I wanted to see the improvement from m1 to m2.

                • damipereira@lemmy.worldOPEnglish
                  1·
                  2 years ago

                  I think I was running at native resolution, which is much less than what you’re running at.

                  Edit: Ok, I actually tried running it again, on 1080p, and it’s faster than I remember, maybe there was some optimization along the way or something happened. I still get really bad frame drops when the temperature drops. It runs at 45fps until it drops to 10fps for like 10/15 seconds while the temperature drops.

                  I actually checked cpu and gpu usage, and the cpu is not max at all, but the gpu is at 100% all the time, that might mean that a faster cpu might indeed be useful. Maybe at around 45fps it starts to become cpu bound?

                  Edit 2: on native resolution it runs at 30fps, and drops to 4fps when temperature drops.

                  Edit 3: Running at 960x600 resolution keeps it running at 50fps, even when the temperature drops. GPU usage is still at maximum.

                  Edit 4: Found the culprit. Global illumination is the setting that made the fps drop so hard when temperature drops. Can you check if it happens for you as well on m2 ultra?

    • damipereira@lemmy.worldOPEnglish
      0·
      2 years ago

      I actually watched that, but it only tests 2 native games which do not run rosetta. The performance increase there seems quite nice, but I’m wondering if rosetta titles will see less of it.

      • lucidinferno@lemmy.worldEnglish
        0·
        2 years ago

        I can’t give specifics or a side-by-side as my m1 mac mini is sold, but I was able to run Rise of the Tomb Raider at higher settings on the m2 pro mac mini than I was the m1. I’d be happy to share my frame rate and such if someone wants to request the settings they’d like me to use.

        • sparky@lemmy.world@lemmy.worldMEnglish
          1·
          2 years ago

          Ditto, I have a pretty sizable steam library mostly of strategy and simulation games, so happy to post experiences with e.g. Paradox games, Tropico, Frostpunk, etc