The final blow for real gaming on the Mac, or the beginning of a new era?

[Frankied22]

How did you become a mod?
It was running on an A12Z ARM chip, an iPad chip. Not desktop, please get educated first then complain. There will be dedicated Mac chips soon...

And don't forget that game was being played though Rosetta 2, which was translating it from x86, so it was essentially an emulated AAA game being played at 1080p on a iPad chip. I think a lot of people don't realize that and how impressive it is. While it didn't look super high quality, it was smooth and playable. Now imagine if they were playing it natively on an actual apple silicon chip they're going to put in a Mac.

 

[Joe Dohn]

What Apple could also do, would be entirely new, and also be a software solution, is to train an artificial intelligence to generate an ARM binary executable from a X86 executable.

You might be perplex, and say: "isn't it what Rosetta does?"

Not really, because Rosetta works from an Universal Binary. The AI would instead reverse engineer a X86 binary (for example, a Windows executable) and generate a corresponding ARM binary.

It turns out that within a few constraints, this is possible for a human to do – just insanely difficult, and usually not worth the effort. With a super powerful AI though, what is difficult becomes easy.

You might think this is bullcr*p, but NVIDIA recently showed an AI that can recreate PacMan – rules, graphics, everything – just by looking at the gameplay. Look at this video:

.

So, Apple following this path would definitely fit with their AI theme. This would require a lot of research, but it can be done. Will Apple think differently, or will they let NVIDIA arrive there first?

 

[leman]

What Apple could also do, would be entirely new, and also be a software solution, is to train an artificial intelligence to generate an ARM binary executable from a X86 executable.

You might be perplex, and say: "isn't it what Rosetta does?"

Not really, because Rosetta works from an Universal Binary. The AI would instead reverse engineer a X86 binary (for example, a Windows executable) and generate a corresponding ARM binary.

Frankly, I think it’s more difficult than you think. Current AI approaches work by abstracting patterns and decomposing them, what they produce is an approximation. In most cases, this approximation is good enough. With code translation, you won’t correctness 100% of the time. Also, the issue is not translating the Windows binary - its a trivial thing, you can take a Windows PE file and convert it to Mac Mach-O. The problem is all the Windows API that these executables link to. You’d need to reimplement Windows on Mac First. By the way, this is exactly what WINE is doing.

But Apple does not have to use an AI. They use an optimizing compiler. I am pretty sure that they uplift x86 code to LLVM IR and then use LLVM to generate the optimized ARM code. Unfortunately, there will always be inefficiencies because a lot of high level information is lost during compilation. For example, indirect jumps. You will have to insert additional instructions that mal original addresses to ARM code addresses (or you add some hardware translation tables to do this).

 

[Janichsan]

It was running on an A12Z ARM chip, an iPad chip. Not desktop, please get educated first then complain.

We don't actually know this.

Everyone assumes that since that's what's in the DTK and what the system info showed for the computer Craig did his demo on, but the Tomb Raider was demoed by Andreas Wendker on what could easily have been a different machine. He never talked about or showed the specs of the machine he was using.

Secondly, keep in mind that none of the WWDCs videos were actually live demonstrations. They were heavily edited and in parts, it's very obvious that the demonstrations have been recorded at another time.

It's safe to assume that Apple already has the actual chips they are going to stick in their upcoming ArMacs at least as prototypes in some machines. They could easily have been using these for the game demos instead of the iPads in a Mac mini casing.

As long as there are no actual, real life, confirmed benchmarks of existing games, we don't know how the games really perform.

 

[AndyMacAndMic]

How did you become a mod?
It was running on an A12Z ARM chip, an iPad chip. Not desktop, please get educated first then complain. There will be dedicated Mac chips soon...

Wow, someone woke up on the wrong side of the bed
Nobody knows for sure what kind of system the heavily edited pre-recorded demo's were running on, so nobody needs education (and I saw nobody complaining either ?).

 

[matrix07]

Everyone assumes that since that's what's in the DTK and what the system info showed for the computer Craig did his demo on, but the Tomb Raider was demoed by Andreas Wendker on what could easily have been a different machine. He never talked about or showed the specs of the machine he was using.

You are free to be suspicious but quite frankly you are just making an accusation that Apple cheat. True, Andreas Wendker demo on a different machine but put it in context of the whole demo if he’s using different chip/SOC than what Craig initially show us, that would mean Apple cheating and considering Apple history we should give Apple benefit of the doubt so please put your tin foil hat down a bit.

Wow, someone woke up on the wrong side of the bed
Nobody knows for sure what kind of system the heavily edited pre-recorded demo's were running on, so nobody needs education (and I saw nobody complaining either ?).

You too as well. ?

 

[Janichsan]

You are free to be suspicious but quite frankly you are just making an accusation that Apple cheat. True, Andreas Wendker demo on a different machine but put it in context of the whole demo if he’s using different chip/SOC than what Craig initially show us, that would mean Apple cheating and considering Apple history we should give Apple benefit of the doubt so please put your tin foil hat down a bit.

If I would accuse Apple of cheating, I would say they recorded the gameplay footage on a Windows PC and just played a video.

No, they simply didn't tell us all the details - for good reasons. They only proclaimed that the game was running on Apple Silicon, but never specified on which Apple Silicon. The connection between the demo and the A12Z in the DTK is something that was created in your head, not by Apple.

 

[matrix07]

No, they simply didn't tell us all the details - for good reasons.

Yes, for good reason. And that reason is Craig had already told us about the SOC used. You are asking Apple to waste precious Keynote time to repeat the same information again so YOU could be convinced. That’s not how the world works.
Maybe you’re the kind of people who still believe Area 51 hiding aliens just because US Air Force has never disputed that, I don’t know. You have the right to be suspicious. I’m just saying your suspicion is ridiculous for people who have a common sense. That’s all.

 

[diamond.g]

You are free to be suspicious but quite frankly you are just making an accusation that Apple cheat. True, Andreas Wendker demo on a different machine but put it in context of the whole demo if he’s using different chip/SOC than what Craig initially show us, that would mean Apple cheating and considering Apple history we should give Apple benefit of the doubt so please put your tin foil hat down a bit.



You too as well. ?

I know the point of the demo was to show off Rosetta 2, and to sort of show the power of their graphics solution, but they really didn't give any info other than running at 1080P. They didn't say rendering in 1080P, we are all assuming that, nor did they say what the average framerate was.
It is also somewhat surprising no one with a DTK (or access to one) has loaded the game on it to take a look themselves, especially after all the benchmarks came out.


The Xbox One S version is rendered at 1600x900 then upscaled, with the FPS capped to 30, with an implementation of VFR. It is hard to tell but I think it is using low settings for everything, or maybe medium with low texture settings.

 

[matrix07]

but they really didn't give any info other than running at 1080P. They didn't say rendering in 1080P, we are all assuming that, nor did they say what the average framerate was.

True, but I’m talking about the SOC used that @Janischsan raised the issue, which information Craig already, and explicitly, gave us.
Someone already uploaded benchmark. If there’s something that’s not right, trust me people will know, especially people at Qualcomm or Microsoft.

 

[Janichsan]

Yes, for good reason. And that reason is Craig had already told us about the SOC used.

No, the reason is that Apple does not want to reveal everything about their upcoming products that far in advance.

You are asking Apple to waste precious Keynote time to repeat the same information again so YOU could be convinced. That’s not how the world works.

Again: at no point did anyone from Apple imply all the demos are running on the same machine. I couldn't care less about how many different types of CPUs they used in the videos, I'm just cautioning you and everyone else that you are jumping to conclusions based on non-existing information.

This has nothing to do with any conspiracy theories. The whole idea of these demonstrations was to show how the upcoming, finalised ArMacs will perform. Not how the DTK performs.

 

[matrix07]

No, the reason is that Apple does not want to reveal everything about their upcoming products that far in advance.

Again: at no point did anyone from Apple imply all the demos are running on the same machine. I couldn't care less about how many different types of CPUs they used in the videos, I'm just cautioning you and everyone else that you are jumping to conclusions based on non-existing information.

This has nothing to do with any conspiracy theories. The whole idea of these demonstrations was to show how the upcoming, finalised ArMacs will perform. Not how the DTK performs.

Of course they wouldn’t want to reveal their upcoming product but there is nothing AT ALL showing there would be a reason that the machine Andreas used would be different than what Craig used but if you want to continue wearing your little hat, then wear on. ??‍♂️

On the last point. No, the whole idea of the demonstration was to show what the least Apple Silicon Apple having now, not the one in the finalized product, can perform.
I suggest you watching Gruber interviewed Craig after the Keynote. He’s talking about this plainly.

 

[diamond.g]

True, but I’m talking about the SOC used that @Janischsan raised the issue, which information Craig already, and explicitly, gave us.
Someone already uploaded benchmark. If there’s something that’s not right, trust me people will know, especially people at Qualcomm or Microsoft.

I missed this, is there a link to someone running Shadows of the Tombraider in benchmark mode on a DTK (does the Mac version still have the benchmark mode)?

 

[matrix07]

I missed this, is there a link to someone running Shadows of the Tombraider in benchmark mode on a DTK (does the Mac version still have the benchmark mode)?

The benchmark is here

Rosetta 2 Benchmarks Surface From Mac Mini With A12Z Chip

While the terms and conditions for Apple's new "Developer Transition Kit" forbid developers from running benchmarks on the modified Mac...

www.macrumors.com

?

 

[diamond.g]

The benchmark is here

Rosetta 2 Benchmarks Surface From Mac Mini With A12Z Chip

While the terms and conditions for Apple's new "Developer Transition Kit" forbid developers from running benchmarks on the modified Mac...

www.macrumors.com

?

Ah, I saw that article, didn't see anything in it about SotTR.

 

[matrix07]

Ah, I saw that article, didn't see anything in it about SotTR.

Gizmodo guessing about it roughly here

So Just How Powerful Are Apple's New Laptop Chips Gonna Be

Apple officially confirmed in its WWDC keynote yesterday that, yes, it would be using its own processors in its laptops and desktops. Not a terribly

gizmodo.com

 

[diamond.g]

Gizmodo guessing about it roughly here

So Just How Powerful Are Apple's New Laptop Chips Gonna Be

Apple officially confirmed in its WWDC keynote yesterday that, yes, it would be using its own processors in its laptops and desktops. Not a terribly

gizmodo.com

They speculate, which is why I was wondering why no one has just run the game to look.

 

[matrix07]

They speculate, which is why I was wondering why no one has just run the game to look.

Good point.

I guess all devs have better things to do.

 

[PortoMavericks]

Guys, TBDR is like really, really bad news for gaming on the Mac.

TBDR has been used to circumvent bandwidth restrictions on old, underpowered hardware (like the Xbox 360 and then the Xbox One S and Fat) and the reconstruction is always very poor on action games.

If you play a game on the iPad and the iPhone, very small screens, it looks OK, when you start pushing across screens sizes... on a 16" MacBook, things start to get ugly.

At the end of the day, the message is: we have an underpowered GPU and we'll force you (dev) to use this trick. This is the final blow on the Mac as a gaming machine like the title suggests.

 

[Zackmd1]

I know the point of the demo was to show off Rosetta 2, and to sort of show the power of their graphics solution, but they really didn't give any info other than running at 1080P. They didn't say rendering in 1080P, we are all assuming that, nor did they say what the average framerate was.
It is also somewhat surprising no one with a DTK (or access to one) has loaded the game on it to take a look themselves, especially after all the benchmarks came out.


The Xbox One S version is rendered at 1600x900 then upscaled, with the FPS capped to 30, with an implementation of VFR. It is hard to tell but I think it is using low settings for everything, or maybe medium with low texture settings.

They specifically stated 1080p and I believe (please correct me if I'm wrong) shadow of the tomb raider on Mac does not support changing render resolution. So 1080p is 1080p, no console upscaling involved here.

 

[Janichsan]

Guys, TBDR is like really, really bad news for gaming on the Mac.

...

AMD's GPUs since the Vega architecture also support TBDR, at least optionally. Nvidia uses at least tiled caching since Maxwell.

 

[PortoMavericks]

AMD's GPUs since the Vega architecture also support TBDR, at least optionally. Nvidia uses at least tiled caching since Maxwell.

Yes, AMD supported it on the Xbox 360.

TBDR is mostly used on underpowered systems.

 

[Janichsan]

Yes, AMD supported it on the Xbox 360.

Vega = desktop and laptop GPUs since 2017.

 

[PortoMavericks]

Vega = desktop and laptop GPUs since 2017.

And?

 

[dugbug]

So you're postulating that the new ARM based MacBook Airs will be powerful enough to to run Tomb Raider? I think that's going to be a stretch. The WWDC demo has Tomb Raider running on their most powerful ARM desktop (and knows what gpu). Even then, it was just at 1080p with all of the detail turned off, i.e., it did not look that great.

in x86 mode on an ipad chipset