Mac Studio M1 Ultra GPU will be beaten by AMD and Nvidia in just 6-7 months.

[darngooddesign]

Why? Apple have told everyone they're moving to Arm. There's zero point in creating a new Intel Mac Pro at this point.

Zero point.

It wouldn't be a new MacPro, but rather an MPX with the new CPUs. Otherwise its a really ****** way to treat your customers.

 

[TiggrToo]

It wouldn't be a new MacPro, but rather an MPX with the new CPUs. Otherwise its a really ***** way to treat your customers.

Why on earth would they even do that?

All that would do is further extend the time they'd be required to support Intel. In the meantime Apple would required to produce motherboards etc to support it.

Again. Absolutely zero reason Apple would do that. It would cost them money and confuse the market.

It's an ARM world at Apple, now. Get used to it.

 

[arvinsim]

....and then 6-7 months after that Apple will beat them...and then 6-7 months after that AMD and Nvidia will win again...and then 6-7 months after THAT Apple will win again, and THEN 6-7 months after THAT...

Companies one-upping each other? Just the way I like it!

 

[mr_roboto]

I'm not talking about gaming and OpenGL is still used for graphics. Fortran is still used for code development, for that matter. Not everyone finds the latest shiny thing worth switching to just because its new nor is it always better for the application at hand.

It's simpler just to ditch the platform that is the most agressive at dumping backwards compatibility. Cheaper too. MacOS doesn't bring enough to the table to make that worth it.

I wasn't talking about gaming either.

It's not about ooh shiny new thing, it's more like the old thing is awful and the new thing addresses what's bad in the old thing. The GL API design is about 40 years old now, counting the decade when it was SGI's proprietary Iris GL, and it hasn't aged well. For at least 20 years it's been very out of alignment with what hardware is actually doing. Thanks to this impedance mismatch, GL drivers are very tricky to write if you want both performance and correctness, and you're always leaving a lot of performance on the table. This difficulty is magnified if you want to properly support all the dusty corners of the GL API spec, and there are many of them.

 

[quarkysg]

It wouldn't be a new MacPro, but rather an MPX with the new CPUs.

You wouldn't like the performance you get from the latest and greatest CPU hobbled by the PCIe bus. This is like driving a Formula 1 car in a congested road.

 

[exoticSpice]

Why? Apple have told everyone they're moving to Arm.

Apple updated the 27" iMac in 2020 after the they told users they are switching

 

[MayaUser]

Apple is dragging themselves down.

yea yea , apple is doomed ...you windows/android users are saying this for years...so your credibility is lost over those years

 

[satcomer]

yea yea , apple is doomed ...you windows/android users are saying this for years...so your credibility is lost over those years

Yea it sets like the bad old day in 90s before Steve Came back! These guess sound actually as those guys did! The arguments are almost the same today as back then!

 

[macOS Lynx]

Traditional GPU manufacturers don’t do new architectures that often. There (generally) is a two year cycle for them.

You missed my point completely ??‍♂️

Companies one-upping each other? Just the way I like it!

It's like people are against competition these days! I want AMD and Nvidia to end up beating Apple because it means Apple will then work hard to reclaim their crown. We've seen what happens when the company on top becomes complacent (*cough* Intel for most of the latter half of the 2010s *cough*). Competition drives innovation.

 

[diamond.g]

You missed my point completely ??‍♂️

It's like people are against competition these days! I want AMD and Nvidia to end up beating Apple because it means Apple will then work hard to reclaim their crown. We've seen what happens when the company on top becomes complacent (*cough* Intel for most of the latter half of the 2010s *cough*). Competition drives innovation.

Oh no I understood your point. I was just saying that realistically AMD/Nvidia won’t have an answer for Apples advances in GPU every year.

 

[jjcs]

I wasn't talking about gaming either.

It's not about ooh shiny new thing, it's more like the old thing is awful and the new thing addresses what's bad in the old thing. The GL API design is about 40 years old now, counting the decade when it was SGI's proprietary Iris GL, and it hasn't aged well. For at least 20 years it's been very out of alignment with what hardware is actually doing. Thanks to this impedance mismatch, GL drivers are very tricky to write if you want both performance and correctness, and you're always leaving a lot of performance on the table. This difficulty is magnified if you want to properly support all the dusty corners of the GL API spec, and there are many of them.

The tools I'm speaking of do not generally require shaders or other advanced features, as they didn't exist when they were written. We'd just prefer to continue to use them without a forced rewrite that costs money no one wants to spend. OpenGL is a simple APi to write to. It's not going away in the engineering world for quite some time. Of course, that isn't a market Apple cares about. Shame they don't.

 

[jjcs]

It wouldn't be a new MacPro, but rather an MPX with the new CPUs. Otherwise its a really ****** way to treat your customers.

Have you met Apple? Doesn't seem to matter to the customers, though.

 

[ahurst]

The tools I'm speaking of do not generally require shaders or other advanced features, as they didn't exist when they were written. We'd just prefer to continue to use them without a forced rewrite that costs money no one wants to spend. OpenGL is a simple APi to write to. It's not going away in the engineering world for quite some time.

Same for a lot of scientific applications and tools. Luckily, Apple’s OpenGL-on-Metal layer seems to work perfectly fine on Apple Silicon and modern macOS, so I’m guessing OpenGL apps are going to continue to work on macOS for a while yet. The only caveat is that we’re not going to get any features beyond GL 4.1 (though I know there’s a developer doing a fair bit of work to reimplement the missing GL 4.6 extensions in Metal).

What *I’d* like to see as a show of goodwill by Apple would be to open-source their GL-to-Metal layer like they’ve open-sourced XNU and Webkit: that way, if macOS itself ever drops support it’ll still be trivial to compile and run OpenGL apps. I’m sure the community could help out a lot ironing out bugs and adding missing extensions, too.

 

[arvinsim]

Same for a lot of scientific applications and tools. Luckily, Apple’s OpenGL-on-Metal layer seems to work perfectly fine on Apple Silicon and modern macOS, so I’m guessing OpenGL apps are going to continue to work on macOS for a while yet. The only caveat is that we’re not going to get any features beyond GL 4.1 (though I know there’s a developer doing a fair bit of work to reimplement the missing GL 4.6 extensions in Metal).

What *I’d* like to see as a show of goodwill by Apple would be to open-source their GL-to-Metal layer like they’ve open-sourced XNU and Webkit: that way, if macOS itself ever drops support it’ll still be trivial to compile and run OpenGL apps. I’m sure the community could help out a lot ironing out bugs and adding missing extensions, too.

But then that means that Metal adoption would be slower.

 

[jjcs]

Same for a lot of scientific applications and tools. Luckily, Apple’s OpenGL-on-Metal layer seems to work perfectly fine on Apple Silicon and modern macOS, so I’m guessing OpenGL apps are going to continue to work on macOS for a while yet. The only caveat is that we’re not going to get any features beyond GL 4.1 (though I know there’s a developer doing a fair bit of work to reimplement the missing GL 4.6 extensions in Metal).

What *I’d* like to see as a show of goodwill by Apple would be to open-source their GL-to-Metal layer like they’ve open-sourced XNU and Webkit: that way, if macOS itself ever drops support it’ll still be trivial to compile and run OpenGL apps. I’m sure the community could help out a lot ironing out bugs and adding missing extensions, too.

It would be nice if they would quantify what "deprecated" means in this instance. It's the goodwill aspect I'm not expecting to see from them for pretty much anything like this. Their way or the highway.

 

[jjcs]

But then that means that Metal adoption would be slower.

As far as the older applications in question, Metal adoption is going to be "never".

 

[jujoje]

Same for a lot of scientific applications and tools. Luckily, Apple’s OpenGL-on-Metal layer seems to work perfectly fine on Apple Silicon and modern macOS, so I’m guessing OpenGL apps are going to continue to work on macOS for a while yet. The only caveat is that we’re not going to get any features beyond GL 4.1 (though I know there’s a developer doing a fair bit of work to reimplement the missing GL 4.6 extensions in Metal).

It does seem that Apples OpenGL on Metal layer works better that it's regular OpenGL, at least comparing my iMac Pro to my MBA, so feel confident that that is the case.

Which developer is reimplement the GL 4.6 extensions in Metal? Kinda curious and it would be great to be able to get a vaguely modern GL feature set.

 

[ahurst]

Which developer is reimplement the GL 4.6 extensions in Metal? Kinda curious and it would be great to be able to get a vaguely modern GL feature set.

This is the repository, no clue how well it works since I haven’t tested myself: https://github.com/openglonmetal/MGL

 

[jujoje]

This is the repository, no clue how well it works since I haven’t tested myself: https://github.com/openglonmetal/MGL

Awesome, that for that

 

[jjcs]

This is the repository, no clue how well it works since I haven’t tested myself: https://github.com/openglonmetal/MGL

That's cool.