16” M1 Max (32 core) Eve Online (Gaming) Benchmarks

[n00byn4t3r]

Metal generally enforces Vsync. I think it can be disabled on certain conditions, but I'm not even sure it's possible to disable Vsync with Metal on Apple GPUs. AFAIK, it's impossible on iPhone and iPads, but I could be wrong.

Who needs to disable it anyway? Screen tearing is just awful and with Promotion, there is really no point in disabling Vsync. It's just useful for benchmarking. But recent built-in benchmark tools have means to estimate frame rates with Vsync on.

Vsync increases input latency since it works by buffering frames (2-3 usually). In some games this can be even more severe due to less than ideal implementations.

Additionally, non-adaptive vsync will need to sync to some multiple of your displays framerate, so if you for example manage 40fps then non-adaptive vsync will force you to 30fps.

 

[Neverless]

The values seems a bit off compared to what happens in GTA played in a crossover or parallels environment. The FPS values are more like in the following range of improvements:
Between M1 8GPU and M1 Pro 14 GPU Cores 20-25%
Between M1 8GPU and M1 Pro 16 GPU Cores 35-45%
Maybe the 32 cores GPU M1 Max is a different beast than M1 Pro though.

 

[Pressure]

The values seems a bit off compared to what happens in GTA played in a crossover or parallels environment. The FPS values are more like in the following range of improvements:
Between M1 8GPU and M1 Pro 14 GPU Cores 20-25%
Between M1 8GPU and M1 Pro 16 GPU Cores 35-45%
Maybe the 32 cores GPU M1 Max is a different beast than M1 Pro though.

Emulation throws any meaningful comparison out of the window.

You should see much better scaling going from 8-core and all the way up to 32-cores.

 

[upandown]

I know there are many variables and this is only one test. And I know this M1 Max doesn’t have the same frame rates as the desktop version. But, (I’m not a gamer so this is a serious question) shouldn’t we be pretty happy with those results? The Max is playing a game at 4K and hitting 70-100fps. I’ve seen gameplay before and I thought anything above 60-70fps is quite good, no?

Let alone play at 1440p and the fps should increase to an even more acceptable level.

 

[Bandaman]

This is a great sign, but unless developers start making more games for the Mac platform, this doesn't mean much.

 

[ikir]

This is a great sign, but unless developers start making more games for the Mac platform, this doesn't mean much.

But it is a start.
macoS is super stable and efficient.
Metal is fast and modern.
Now we have hardware capable of gaming at low, mid and high-end. even M1 is fine for many casual non hardcore gamers. We just need devs considering the platform more seriously and making good ports not like Borderlands 3 trash.

 

[LoopsOfFury]

Thanks for these tests! Not bad at all. This again illustrates that these GPUs are beasty rasterizers.



This is interesting since AA is generally very cheap on TBDR hardware. It will also depend on how AA is implemented in Eve. Even more surprising that the NVIDIA GPU does not show any kind of performance difference here.

A panel with their art team back in 2015 said they were looking into replacing their MSAA implementation with TXAA, but I haven't seen anything since then.

 

[jeanlain]

Eve just received its first native Mac client a few weeks ago (previous iterations relied on WINE).

But it's an X86 application isn't it? I've heard that it uses Rosetta.

 

[LoopsOfFury]

When they announced their plans last December, they said the first release would use rosetta, with a universal binary coming later. Browsing through the forums, it looks like the launcher still uses rosetta, but the actual client is now universal.