Current State Mac (June/2020)

[ErwinSchrodinger]

The biggest problem with apple is it's run like a cult instead of a tech company. Instead of trying to please any of their customers, they tell their followers what they will enjoy.

Effective marketing. Lots of companies do the exact same thing, with varying degrees of success. That's why choosey moms choose JIF. And none of that really has nothing to do with SE support of the Mac client, which is the argument that I think most Mac users are trying to make. SE is solely to blame for how well their software supports platforms it's released on. Will a natively developed version of FFXIV perform better on Mac OS than a natively developed version on Windows? Probably not, and yeah Apple is at least partially to blame for that. But I don't think most Mac users would be upset if it performed worse than Windows/PS4, as long as it just performed better than how it currently performs in a WINE wrapper.

To the OP: I play on the Mac Client with an eGPU, and while it doesn't run as nice as it does on Windows, I get enough performance that I'm satisfied with it. I think the only way we'll see improvements is if the 3rd party WINE provider somehow better supports Metal in Mac OS. Maybe if SE decides to rebuild FFXIV in the future, they'll select a platform that has more tools available for making cross platform games where they don't have to rely on a WINE wrapper. Fingers crossed.

[justinhuang]

Effective marketing. Lots of companies do the exact same thing, with varying degrees of success. That's why choosey moms choose JIF. And none of that really has nothing to do with SE support of the Mac client, which is the argument that I think most Mac users are trying to make. SE is solely to blame for how well their software supports platforms it's released on. Will a natively developed version of FFXIV perform better on Mac OS than a natively developed version on Windows? Probably not, and yeah Apple is at least partially to blame for that. But I don't think most Mac users would be upset if it performed worse than Windows/PS4, as long as it just performed better than how it currently performs in a WINE wrapper.

To the OP: I play on the Mac Client with an eGPU, and while it doesn't run as nice as it does on Windows, I get enough performance that I'm satisfied with it. I think the only way we'll see improvements is if the 3rd party WINE provider somehow better supports Metal in Mac OS. Maybe if SE decides to rebuild FFXIV in the future, they'll select a platform that has more tools available for making cross platform games where they don't have to rely on a WINE wrapper. Fingers crossed.

I do hope there are better optimizations for Mac too.

[MsQi]

Effective marketing. Lots of companies do the exact same thing, with varying degrees of success. That's why choosey moms choose JIF. And none of that really has nothing to do with SE support of the Mac client, which is the argument that I think most Mac users are trying to make.

Oh, no no no, it goes much deeper than that. Have you been to a Macworld Expo? Apple's capriciousness makes it hard for third parties to support them consistently.

A fully native client would probably rival that of comparable hardware running Windows. However if a change in architecture is coming I wouldn't count on it.

[Packetdancer]

To the OP: I play on the Mac Client with an eGPU, and while it doesn't run as nice as it does on Windows, I get enough performance that I'm satisfied with it. I think the only way we'll see improvements is if the 3rd party WINE provider somehow better supports Metal in Mac OS. Maybe if SE decides to rebuild FFXIV in the future, they'll select a platform that has more tools available for making cross platform games where they don't have to rely on a WINE wrapper. Fingers crossed.

As of the version that supports Catalina, I'm pretty sure the macOS version does support Metal instead of being atop OpenGL. It's just taking a somewhat roundabout method to get there. (DirectX goes through DXVK to be translated into Vulkan API calls, the resulting Vulkan calls go through MoltenVK to be translated into Metal API calls, and then Metal is actually executed.) It makes sense, inasmuch as WINE now uses DXVK so that any DirectX software can take advantage of Vulkan, so building the macOS version atop the existing MoltenVK translation layer rather than writing an entire new DirectX-to-Metal translation makes obvious sense.

The shift to that tower of translation layers is seemingly why the client performs (slightly) better on machines with discrete graphics than it used to, but can no longer run properly on anything using integrated Intel Iris Pro graphics; in a (very cursory) examination of the Crossover bottle in action, MoltenVK atop Metal on the Iris Pro GPU seems only able to emulate Vulkan 1.0, while DXVK seems to require Vulkan 1.1 to function properly.

[ErwinSchrodinger]

As of the version that supports Catalina, I'm pretty sure the macOS version does support Metal instead of being atop OpenGL. It's just taking a somewhat roundabout method to get there. (DirectX goes through DXVK to be translated into Vulkan API calls, the resulting Vulkan calls go through MoltenVK to be translated into Metal API calls, and then Metal is actually executed.) It makes sense, inasmuch as WINE now uses DXVK so that any DirectX software can take advantage of Vulkan, so building the macOS version atop the existing MoltenVK translation layer rather than writing an entire new DirectX-to-Metal translation makes obvious sense.

The shift to that tower of translation layers is seemingly why the client performs (slightly) better on machines with discrete graphics than it used to, but can no longer run properly on anything using integrated Intel Iris Pro graphics; in a (very cursory) examination of the Crossover bottle in action, MoltenVK atop Metal on the Iris Pro GPU seems only able to emulate Vulkan 1.0, while DXVK seems to require Vulkan 1.1 to function properly.

Seems plausible. I guess all of those translation steps result in a very poorly optimized experience.

I don't know what engine SE uses for FFXIV (proprietary: Luminous?), but several popular ones support Vulkan, and by extension would support MoltenVK, right? I'm sure I'm trivializing the amount of work needed, but wouldn't it be relatively easy to port a game if they were using an engine that supports multiple API's out of the box? Would seem to make sense in bringing it to more platforms and opening it up to a larger audiences imo.