Mac New Game Porting Toolkit is Wine

[dada_dave]

A little more information from one of the devs of the porting kit

Gokhan Avkarogullari (@gavkar@mastodon.gamedev.place)

@yiningkarlli @aras @romainguy@androiddev.social we built a dxil to metallib converter and directx11 and directx12 to Metal runtime translator. Non graphics APIs are translated by Wine. But we don’t use any tech from moltenVK or DXVK or spirv-cross etc. The shader converter can be shipped by...

mastodon.gamedev.place

And a message about how Apple sees macs and gaming

Gokhan Avkarogullari (@gavkar@mastodon.gamedev.place)

Our Metal Ecosystem team took the technology and have made them available to the developers in a way that is easy use and integrate . Macs are perfect game machines. Great GPUs, CPUs, memory bandwidth, storage, audio, video, display and they are quiet, fast, with long battery life.

mastodon.gamedev.place

So it does seem like developers can ship at least some translations with their games - specifically the DX to Metal translations they’ve created.

 

[leman]

I also feel their attitude makes the mistake of thinking things should go from 0-100 immediately.

Yeah, I agree. I don't understand that attitude at all. You either adopt technology controlled by others, but they you abandon your ability to innovate, or you build up a product that others want to use. That won't happen instantly obviously.

So it does seem like developers can ship at least some translations with their games - specifically the DX to Metal translations they’ve created.

The way it works is that the shader translator takes the DX shaders and produces a Metal shader library. This is essentially a DX-to-Metal-IR compiler. Very different from the usual paradigm of converting shaders to Metal shading language and then editing and tweaking those. This approach allows you to use the same exact shader codebase for both Windows and macOS, which is a great move on Apple's side. You can always implement some Mac-specific shaders (e.g. using tile shading to optimise your deferred lighting approach) and link them into the pipeline, which gives you a lot of flexibility.

 

[Jimmyjames]

The way it works is that the shader translator takes the DX shaders and produces a Metal shader library. This is essentially a DX-to-Metal-IR compiler. Very different from the usual paradigm of converting shaders to Metal shading language and then editing and tweaking those. This approach allows you to use the same exact shader codebase for both Windows and macOS, which is a great move on Apple's side. You can always implement some Mac-specific shaders (e.g. using tile shading to optimise your deferred lighting approach) and link them into the pipeline, which gives you a lot of flexibility.

Do you think this approach will lead to good performance, or does it leave some on the table?

 

[leman]

Do you think this approach will lead to good performance, or does it leave some on the table?

Hand-optimized stuff probably always will be better, especially if you have a complex pipeline that makes certain architectural assumptions. But this should be good enough

 

[Nycturne]

They are not working with Crossover (according to Crossover spokesperson at least), and I don't think that this will negatively affect Crossovers business. If anything, it will draw more attention since it has been now demonstrated that DX12 emulation on macOS is feasible.

Crossover has two sets of customers, though. One is direct to end users, sure. The other is developers looking to bring games to the Mac without doing a full port (FF XIV and EVE Online for example). I can’t imagine that Crossover is going to get a lot of developer business going forward.

If you were going this route, would you rather it be the platform owner or a third party maintaining the compatibility layer? Especially knowing that Apple is including functionality that makes it possible to do some tuning for Metal without requiring I port the whole app first?

Apple is not interested in these tools to be used to actually run games. I suppose this is why they avoided upstreaming the patches as well as keeping their DX12 implementation layer private. This is positioned strictly as a developer tool, lacks ergonomic players need, and restricted as such by the license agreement. A new Proton is the last thing Apple wants, since this would kill any potential for Mac gaming altogether.

Note my comments were about developers, so I’m well aware of what this is aimed at.

My reference to Proton is more in the sense that Valve wants Steam to be successful on Linux. They built Proton as a way to enable that and projects like the Steam Deck. Folks have gotten it to work outside of Steam because it is open source, but it is very much about bridging a gap and getting more games (and thus gamers) on Linux to demonstrate that it is worth spending effort on a more native port. Seems to me Apple has the same goal here. The rest is nit picking.

 

[Jimmyjames]

Sometimes I have a moments positivity regarding Mac gaming, and then I read a comment like this, from a long time Mac game dev, working for one of the only Mac games porting houses, which has just given up on Linux ports, and I wonder what is wrong with this community

.

Literally talking themselves out of a job

 

[leman]

Crossover has two sets of customers, though. One is direct to end users, sure. The other is developers looking to bring games to the Mac without doing a full port (FF XIV and EVE Online for example). I can’t imagine that Crossover is going to get a lot of developer business going forward.

Shipping a game as a wine bottle is still significantly less work than doing a port. I don't know how many business customers Crossover had until now, but if their solution is robust enough they should still be able to attract some interest.

If you were going this route, would you rather it be the platform owner or a third party maintaining the compatibility layer? Especially knowing that Apple is including functionality that makes it possible to do some tuning for Metal without requiring I port the whole app first?

Are you asking me personally? My problem is, I don't compromise. I hate crappy products and I am unreasonably obsessed with quality. I would never use a half-baked solution like Crossover to run my application. Then again, I am not the right person to ask since I never actually shipped a product. Gamedev is just a hobby I dabble in my spare time.

My reference to Proton is more in the sense that Valve wants Steam to be successful on Linux. They built Proton as a way to enable that and projects like the Steam Deck. Folks have gotten it to work outside of Steam because it is open source, but it is very much about bridging a gap and getting more games (and thus gamers) on Linux to demonstrate that it is worth spending effort on a more native port. Seems to me Apple has the same goal here. The rest is nit picking.

I disagree about Proton. As I understand it, the point of Proton is full emulation so that Steam games can be run unmodified. This is not about Linux, this is about Steam as a platform. Proton is absolutely disastrous for Linux gaming as it removes any kind of incentive to develop native Linux games. Why bother if you can just make and optimise your game for Windows/DX12 and have the Proton devs deal with the gritty detail? Even more, Proton kind of makes Vulkan pointless as a gamedev target. Why bother to deal with Vulkan if you get Vulkan support for free anyway (not to mention that DX12 has far superior development tools). I understand that Linux users are happy about Proton because at least now they can play games, but this effectively yields any initiative to Microsoft. Linux gaming becomes Windows and Microsoft API emulation.

Apple really doesn't want to go that route as they want to keep initiative. This allows them to innovate and differentiate their platform. Of course, this is also a much more difficult and risky path. But again, let's not forget that Apple plays a long game here. Their solutions will improve. I mean, already now people can run Cyberpunk on base M1 at Ultra settings 1400p at 15FPS, and that's with the emulation overhead and no optimisation whatsoever. As a native Metal version it could probably reach 40-50 fps on high (with upscaling). And that's a 15W chip. The psychological effect of this should not be underestimated.

 

[KingOfPain]

Andrew Tsai has made another video with footage of games that previously didn‘t run on macOS:

 

[Nycturne]

Shipping a game as a wine bottle is still significantly less work than doing a port. I don't know how many business customers Crossover had until now, but if their solution is robust enough they should still be able to attract some interest.

But that's just it, Apple is saying "Here's a way to ship a game as a wine bottle, but if you put extra effort, the D3D->Metal layer will suck a lot less."

It's the same thing as Crossover, only with more fleshed out DX12 support, tools for going a few steps further, and less likely to break randomly when Apple changes something in the OS. Why would I pay for Crossover?

PS: It does look like Codeweavers is claiming their code is being used. They seem excited to push the idea that they are happy to consult on your game if you are working with the kit, but if Apple is serious, they are likely going to have folks filling that role in-house to work with partner developers. https://www.codeweavers.com/blog/mj...ting-toolkit-powered-by-crossover-source-code

I disagree about Proton. As I understand it, the point of Proton is full emulation so that Steam games can be run unmodified. This is not about Linux, this is about Steam as a platform. Proton is absolutely disastrous for Linux gaming as it removes any kind of incentive to develop native Linux games. Why bother if you can just make and optimise your game for Windows/DX12 and have the Proton devs deal with the gritty detail? Even more, Proton kind of makes Vulkan pointless as a gamedev target. Why bother to deal with Vulkan if you get Vulkan support for free anyway (not to mention that DX12 has far superior development tools). I understand that Linux users are happy about Proton because at least now they can play games, but this effectively yields any initiative to Microsoft. Linux gaming becomes Windows and Microsoft API emulation.

The same could be said of this. You can put in effort to align better with Metal's pipeline, but you are still generally beholden to D3D's initiative with this approach because at the end of the day, you aren't making a Mac app, you are running a wine bottle.

 

[Jimmyjames]

But that's just it, Apple is saying "Here's a way to ship a game as a wine bottle, but if you put extra effort, the D3D->Metal layer will suck a lot less."

Hmmm is that correct? That isn't my impression at all. I understood Apple is saying, "here's a way to see if your game can run at all, and what performance (unchanged) will be like, so you can quickly judge how much work is needed".

It's the same thing as Crossover, only with more fleshed out DX12 support, tools for going a few steps further, and less likely to break randomly when Apple changes something in the OS. Why would I pay for Crossover?

PS: It does look like Codeweavers is claiming their code is being used. They seem excited to push the idea that they are happy to consult on your game if you are working with the kit, but if Apple is serious, they are likely going to have folks filling that role in-house to work with partner developers. https://www.codeweavers.com/blog/mj...ting-toolkit-powered-by-crossover-source-code

Apple is using their code for the non-gaming work. All gaming related translation is due to their own work.

Gokhan Avkarogullari (@gavkar@mastodon.gamedev.place)

@yiningkarlli @aras @romainguy@androiddev.social we built a dxil to metallib converter and directx11 and directx12 to Metal runtime translator. Non graphics APIs are translated by Wine. But we don’t use any tech from moltenVK or DXVK or spirv-cross etc. The shader converter can be shipped by...

mastodon.gamedev.place

 

[leman]

But that's just it, Apple is saying "Here's a way to ship a game as a wine bottle, but if you put extra effort, the D3D->Metal layer will suck a lot less."

Not at all. They give you the option to run your game on your development machine in a wine bottle, but shipping this Bottle is explicitly prohibited by the license agreement.

It's the same thing as Crossover, only with more fleshed out DX12 support, tools for going a few steps further, and less likely to break randomly when Apple changes something in the OS. Why would I pay for Crossover?

Crossover is a user-targeted application, game porting kit is a developer tool. Already installing it is kind of a pain in the ass. Sure, one can wrap it (someone already did - and I wouldn’t be surprised if Apple legal went after them for breach of license).

The same could be said of this. You can put in effort to align better with Metal's pipeline, but you are still generally beholden to D3D's initiative with this approach because at the end of the day, you aren't making a Mac app, you are running a wine bottle.

No, you are shipping a 100% native Mac app that uses Metal directly, it’s just that your shaders are written in Microsoft's HLSL. And yes, you are still beholden to DX12 base technology, but since you are actually forced to rewrite your rendering code for Metal you are more likely to use Apples tech (especially since it’s generally much easier to use).

 

[Nycturne]

Not at all. They give you the option to run your game on your development machine in a wine bottle, but shipping this Bottle is explicitly prohibited by the license agreement.

This is what I get for not reading the read me first. Yeah, it's there for analysis purposes.

Crossover is a user-targeted application, game porting kit is a developer tool. Already installing it is kind of a pain in the ass. Sure, one can wrap it (someone already did - and I wouldn’t be surprised if Apple legal went after them for breach of license).

Crossover is also sold to developers. Knowing the above, they still have a target market, although still somewhat diminished (I hope).

 

[Colstan]

Here's a concise seven minute clip from a two hour interview, featuring Apple's VP of Hardware Engineering John Ternus and VP of Worldwide Marketing Greg Joswiak, concerning Apple's efforts in getting new games on the Mac. Instead of having to wade through two hours, a nice person on Reddit was kind enough to clip it for us.

The good news is that Apple is fully committed to this endeavor, introduced new tools for developers which should make it easier and quicker to port games to the Mac, and considers the entire line of Apple Silicon Macs to be capable gaming machines. Joswiak: "We used to only have a limited amount of Macs that we made that could really do high-end gaming, and the difference is now that every Mac that we sell is capable of doing high-end gaming." Clearly, this is a not so subtle jab at the poor performance of Intel's integrated graphics, which shipped in the majority of Intel Macs as the primary GPU. Joz also specifically notes that the gaming performance is "because of what we've done with Apple Silicon, because of what we've done with unified memory, our GPUs". I think that statement, once again, demonstrates that Apple has no intention of using third-party GPUs.

The bad news is that the interview was done by John Gruber, who isn't a gamer, had no idea what questions to ask, apparently didn't prepare at all, and didn't understand the significance of the entire gaming announcement. I understand why Apple uses him, he's always got the gimp suit ready to go, but this was clearly not an area that Gruber is equipped to cover. Half of the video is Gruber struggling to find a coherent sentence.

 

[Jimmyjames]

Here's a concise seven minute clip from a two hour interview, featuring Apple's VP of Hardware Engineering John Ternus and VP of Worldwide Marketing Greg Joswiak, concerning Apple's efforts in getting new games on the Mac. Instead of having to wade through two hours, a nice person on Reddit was kind enough to clip it for us.

The good news is that Apple is fully committed to this endeavor, introduced new tools for developers which should make it easier and quicker to port games to the Mac, and considers the entire line of Apple Silicon Macs to be capable gaming machines. Joswiak: "We used to only have a limited amount of Macs that we made that could really do high-end gaming, and the difference is now that every Mac that we sell is capable of doing high-end gaming." Clearly, this is a not so subtle jab at the poor performance of Intel's integrated graphics, which shipped in the majority of Intel Macs as the primary GPU. Joz also specifically notes that the gaming performance is "because of what we've done with Apple Silicon, because of what we've done with unified memory, our GPUs". I think that statement, once again, demonstrates that Apple has no intention of using third-party GPUs.

The bad news is that the interview was done by John Gruber, who isn't a gamer, had no idea what questions to ask, apparently didn't prepare at all, and didn't understand the significance of the entire gaming announcement. I understand why Apple uses him, he's always got the gimp suit ready to go, but this was clearly not an area that Gruber is equipped to cover. Half of the video is Gruber struggling to find a coherent sentence.

Great summary and I agree with your summation.

I enjoyed listening to Ternus speak. He seems very knowledgeable and enthusiastic. The interest in gaming should be obvious for all to see, although I wouldn't be surprised to see doubters remain.

Gruber though...oof. I am trying to be as generous as possible given I know how difficult it can be to speak in public, but I found him unlistenable. Slow, unable to put a sentence together and distracting. He made the whole show uncomfortable. I would forgive it, but his entire personality is built around the idea that he is a raconteur. Hence the name of his podcast. He seems to believe he is some kind of tech Johnny Carson. He couldn't be less suited to the role. Agree also that he is unprepared to dig deep on the interesting areas, unless those areas involve typography or complaints about UI elements. It strikes me that many of the most prominent members of this community are uninterested in anything other than complaining about UI.

If I might make a slight diversion, to illustrate this point, I recently had a discussion with a famous podcaster and former Mac OS X reviewer. He seemed very annoyed by Ternus stating that they weren't thinking about discreet gpus. He clipped the appropriate segment, and I responded that Apple believes a soc can compete with a discreet gpu, and if you look at the nascent scores for the M2 Ultra, it seems competitive with everything but the 4090. He replied that while Apple may be able to do that, they hadn't yet. As proof of that, he tooted this to me

John Siracusa (@siracusa@mastodon.social)

Attached: 1 image @jimmyjamesuk Maybe if you do most of your gaming offscreen.

mastodon.social

Yes, he thought the performance delta of the ON SCREEN tests in gfxbench was proof of how far Apple's gpus lagged the best. Seriously. the self appointed gamer who has been playing for years and yearns for the absolute best performance has absolutely no idea how Apple Silicon performs. We are being woefully underserved by the "top" members of the community.

Rant over.

 

[dada_dave]

Great summary and I agree with your summation.

I enjoyed listening to Ternus speak. He seems very knowledgeable and enthusiastic. The interest in gaming should be obvious for all to see, although I wouldn't be surprised to see doubters remain.

Gruber though...oof. I am trying to be as generous as possible given I know how difficult it can be to speak in public, but I found him unlistenable. Slow, unable to put a sentence together and distracting. He made the whole show uncomfortable. I would forgive it, but his entire personality is built around the idea that he is a raconteur. Hence the name of his podcast. He seems to believe he is some kind of tech Johnny Carson. He couldn't be less suited to the role. Agree also that he is unprepared to dig deep on the interesting areas, unless those areas involve typography or complaints about UI elements. It strikes me that many of the most prominent members of this community are uninterested in anything other than complaining about UI.

If I might make a slight diversion, to illustrate this point, I recently had a discussion with a famous podcaster and former Mac OS X reviewer. He seemed very annoyed by Ternus stating that they weren't thinking about discreet gpus. He clipped the appropriate segment, and I responded that Apple believes a soc can compete with a discreet gpu, and if you look at the nascent scores for the M2 Ultra, it seems competitive with everything but the 4090. He replied that while Apple may be able to do that, they hadn't yet. As proof of that, he tooted this to me

John Siracusa (@siracusa@mastodon.social)

Attached: 1 image @jimmyjamesuk Maybe if you do most of your gaming offscreen.

mastodon.social

Yes, he thought the performance delta of the ON SCREEN tests in gfxbench was proof of how far Apple's gpus lagged the best. Seriously. the self appointed gamer who has been playing for years and yearns for the absolute best performance has absolutely no idea how Apple Silicon performs. We are being woefully underserved by the "top" members of the community.

Rant over.

Ugh that’s painful.

 

[Colstan]

Gruber though...oof. I am trying to be as generous as possible given I know how difficult it can be to speak in public, but I found him unlistenable.
Agree also that he is unprepared to dig deep on the interesting areas, unless those areas involve typography or complaints about UI elements. It strikes me that many of the most prominent members of this community are uninterested in anything other than complaining about UI.

I stopped visiting Gruber's blog when it started going through an identify crisis. I don't know whether it's supposed to be an Apple centric technology site, or a political opinion journal. I don't find Gruber's tech thoughts to be insightful, just wordy and written as a poor attempt to sound erudite. I don't care about his political opinions. Period.

If I might make a slight diversion, to illustrate this point, I recently had a discussion with a famous podcaster and former Mac OS X reviewer.

I was going to cut Siracusa some slack. We have to remember that he is currently in a prolonged, protracted, deep period of mourning. Without so much as consulting him, Apple released the Apple Silicon Mac Pro in a state which he finds untenable. If you listen to the post-WWDC edition of ATP, everything after the Mac Pro discussion was colored by his abject horror at the state of his most precious device.

I've listened to ATP long enough to know what Siracusa expects. What I can't explain is why he desires it. He purchases Mac Pros, not because he needs them, but because he wants them. That is the case with the most recent 2019 model, and the model he used for an entire decade before that. He also purchased a Pro Display XDR, again, not because he needs it, but because he wants it. He's a full-time professional podcaster that dabbles with Xcode on the side. Nothing he does requires that level of power in a computer or clarity in a monitor.

Now, I have no problem with him spending his disposable income on a fancy computer and display to match. I don't begrudge someone for buying what they want. Life is short, if he has the resources, then he should purchase what makes him happy, as long as he isn't giving up the family farm in the process.

However, what this clearly makes him is not part of the target market for the Mac Pro. It's a professional device meant for a professional niche customer. Siracusa is an enthusiast, much like the ne'er-do-wells over in the Mac Pro forum on MacRumors. He thinks that because he owns the product, therefore he is an expert, and hence knows what's best for the professional market, far better than Apple. His disappointment is legion, joining the other dozen people who purchased a Mac Pro for similar enthusiast reasons.

The only taxing program he runs on his 2019 Mac Pro is a Windows-only game with Boot Camp. I don't remember which game, because I don't care. Nothing he does with macOS requires that amount of grunt. When his co-host, Casey Liss (the only not-crazy member of ATP) inquired why he doesn't just build a Windows PC to go alongside a reasonably appropriate Mac, Siracusa responds that "I don't want a Windows PC for irrational silly reasons". Even he admits that his logic lies somewhere between flawed and non-existent.

The reason that he wants third-party graphics cards for the Apple Silicon Mac Pro is "in case I want to use them for gaming", even though he openly states that he only plays one game, a Windows title in Boot Camp. Boot Camp functionality is unavailable with Apple Silicon Macs, therefore expandable graphics cards would do him no good. Just wanting to have that capability, for the sake of having expandability, is the stuff of fairy tales.

He has also made it clear that the Mac is not a gaming platform. When Resident Evil Village was announced, he dunked all over it. Similarly, with the substantial WWDC 2023 gaming announcements, he had an even worse reaction, because not only does he downplay the ability for Macs to be capable gaming machines, but he was in deep distress over the state of his beloved Mac Pro.

In summation, Siracusa owns a Mac Pro and XDR which he has no use for, refuses to build a PC to play his only Windows game because of self-admitted irrational reasons, hates the new Mac Pro because it won't utilize third-party graphics cards, just in case he wants to use them for gaming, which is a capability that he thinks Apple Silicon isn't suited for. Did I miss anything?

I've been trying to understand the thought processes of the likes of Gruber, Siracusa, Rene Ritchie, Marco Arment, and the collective hive mind of the MacRumors forum denizens. On the one hand, we have the sycophantic toadies who believe that "Apple can do no wrong". On the other side of the equation, there are the "Apple is doomed and Tim Cook needs to be fired" cadre. These two cohorts are opposite sides of the same shekel.

I'm convinced that these folks have developed a parasocial relationship with Apple, except instead of public individuals, it's an entire corporation. The Apple boot lickers have intertwined their personal identity with the fruit company. The angry chest-beating malcontents have felt betrayed by Apple, over some imagined slight, because they didn't provide them with the product that they personally imagined during a fever dream. Either way, it's an unhealthy parasitic relationship, one which is highly lucrative for some of the afflicted.

Then there's the rest of us who just want a damn computer. A normal person understands that you can be a fan of a company's products while still being critical when Apple falls short. The fruit company is composed of regular human beings, who in all likelihood want to make the best possible devices that they can, but also at times make mistakes, just like everyone else. They shouldn't be deified, nor vilified, but evaluated on what products they release for their customers. Not judged on whatever fanciful wish is cast and then getting upset when the genie doesn't appear.

I used to respect the opinions of these "Apple influencers", before they drank their own kool-aid. Now, I may listen to them for entertainment, or simply ignore them entirely. Regardless, I don't take them seriously, not in the slightest, when all logic and reason evaporated many eons ago.

 

[Jimmyjames]

The only taxing program he runs on his 2019 Mac Pro is a Windows-only game with Boot Camp. I don't remember which game, because I don't care. Nothing he does with macOS requires that amount of grunt. When his co-host, Casey Liss (the only not-crazy member of ATP) inquired why he doesn't just build a Windows PC to go alongside a reasonably appropriate Mac, Siracusa responds that "I don't want a Windows PC for irrational silly reasons". Even he admits that his logic lies somewhere between flawed and non-existent.

I think he plays Destiny? I wonder if it would run well in the Game Porting Toolkit?

The reason that he wants third-party graphics cards for the Apple Silicon Mac Pro is "in case I want to use them for gaming", even though he openly states that he only plays one game, a Windows title in Boot Camp. Boot Camp functionality is unavailable with Apple Silicon Macs, therefore expandable graphics cards would do him no good. Just wanting to have that capability, for the sake of having expandability, is the stuff of fairy tales.

He has also made it clear that the Mac is not a gaming platform. When Resident Evil Village was announced, he dunked all over it. Similarly, with the substantial WWDC 2023 gaming announcements, he had an even worse reaction, because not only does he downplay the ability for Macs to be capable gaming machines, but he was in deep distress over the state of his beloved Mac Pro.

Absolutely. I recall when REV was announced along with NMS and I genuinely thought he would be happy as a person with an interest in the platform. I tweeted him and asked if he would consider mentioning it as a means of promoting the Mac as a gaming platform. To put it mildly, he refused. "Everyone has already played it" etc. The usual refrain.

In summation, Siracusa owns a Mac Pro and XDR which he has no use for, refuses to build a PC to play his only Windows game because of self-admitted irrational reasons, hates the new Mac Pro because it won't utilize third-party graphics cards, just in case he wants to use them for gaming, which is a capability that he thinks Apple Silicon isn't suited for. Did I miss anything?

That about sums it up. What is particularly galling is that he spends much of his time discussing and debating these topics, and yet seems not to know much about it, all while being against it, as you describe!

I used to respect the opinions of these "Apple influencers", before they drank their own kool-aid. Now, I may listen to them for entertainment, or simply ignore them entirely. Regardless, I don't take them seriously, not in the slightest, when all logic and reason evaporated many eons ago.

A wise move I think.

 

[Colstan]

One more small tidbit from the Gruber interview with Apple executives. When asked about third-party graphics cards not being available on the Mac, John Ternus reponded:

"Fundamentally, we've built our architecture around this shared memory model and that optimization, and so it's not entirely clear to me how you'd bring in another GPU and do so in a way that is optimized for our systems. It hasn't been a direction that we wanted to pursue."

The last thread I read over at MacRumors about this were people demanding eGPU support. It think that will go as well as it did for those who demand Boot Camp.

 

[Jimmyjames]

One more small tidbit from the Gruber interview with Apple executives. When asked about third-party graphics cards not being available on the Mac, John Ternus reponded:

"Fundamentally, we've built our architecture around this shared memory model and that optimization, and so it's not entirely clear to me how you'd bring in another GPU and do so in a way that is optimized for our systems. It hasn't been a direction that we wanted to pursue."

The last thread I read over at MacRumors about this were people demanding eGPU support. It think that will go as well as it did for those who demand Boot Camp.

I’m quite happy they said this. Firstly because it will end the constant speculation I hope! And secondly because I believe this means they can compete without a discreet gpu.

Macrumors threads are hilarious. I just saw someone post that Apple need a discreet gpu and they will never compete with a mobile chip…a few posts under the posted benchmarks showing them doing exactly that.

I am now very eager to see what the M3 brings. My one remaining concern is that the M3 will be based in the A16.

 

[dada_dave]

I am now very eager to see what the M3 brings. My one remaining concern is that the M3 will be based in the A16.

I don’t think that’s very likely given the timing and the fact that they’d have to port the A16 cores to N3 as opposed to just using for the N3 designed A17 cores.