TR Forums

SuperSpy wrote:

The one part where attempting to compare the actual performing hardware (Vortex vs Kaby/Coffee Lake for instance) would be to entertain questions like "Could Apple ship something like a Macbook Air with a home-grown ARM CPU?"

Very happy to dig into this, but we simply can't do it yet. For as much code as they share, iOS is NOT macOS. They're just not comparable as they stand now, and neither are the applications that run on each right now. I'd love to see how the A12X performs on macOS in something like an Air/Pro body. I'd love it. But we don't have that here. That's just how it is for now.

EU seems to have a different definition of what they think "real world" usage is. For example, I'm currently working on a media encoding platform that involves uploading multi-gigabyte files and processing them using various tools. Since the files are so large it would be nice to be able to verify what the user is uploading before they actually upload it, but the capabilities we have for doing that within the browser are still fairly limited.

There is one project called emscripten which is an LLVM to javascript compiler, and someone has fed ffmpeg to it, so I could conceivably use that. I suppose you could call that a "real world" application also, but even gzipped it's like a 6MB file, and who knows what performance would be like if it even worked without a bunch of weird bugs. I know it's super hip and trendy to think mobile devices are going to take over the world, but the fact that it's possible doesn't automatically make it easily usable or even a good idea.

DancinJack wrote:

SuperSpy wrote:

The one part where attempting to compare the actual performing hardware (Vortex vs Kaby/Coffee Lake for instance) would be to entertain questions like "Could Apple ship something like a Macbook Air with a home-grown ARM CPU?"

Very happy to dig into this, but we simply can't do it yet. For as much code as they share, iOS is NOT macOS. They're just not comparable as they stand now, and neither are the applications that run on each right now. I'd love to see how the A12X performs on macOS in something like an Air/Pro body. I'd love it. But we don't have that here. That's just how it is for now.

You think so? I'd actually be quite shocked if internally Apple wasn't a compiler switch/#ifdef away from having a full ARM64 install of MacOS and all the included first-party apps. Much like they did in the past with POWER/x86.

That may be true, but the public builds that people are producing these benchmarks on are not that. Not to mention the apps that are in production currently run very different code and have very different capabilities. We're just not comparing apples to apples right now. That might change soon, but until then it's all just speculation.

SuperSpy wrote:

You think so? I'd actually be quite shocked if internally Apple wasn't a compiler switch/#ifdef away from having a full ARM64 install of MacOS and all the included first-party apps. Much like they did in the past with POWER/x86.

Well, it's a compiler switch and a solid emulator ...

which is even harder when the chips you might move to are not actually better.

Redocbew wrote:

EU seems to have a different definition of what they think "real world" usage is

Everybody has a different opinion of what real world usage is. Mine is different from yours. Mine changes depending on the workflow and location. I'm not comparing the iPad Pro to my 12 core ESXi server or to my VR gaming rig.

I've got a 1TB iPad Pro on order. I'm going to be comparing it to my T480.

SuperSpy wrote:

DancinJack wrote:

SuperSpy wrote:

The one part where attempting to compare the actual performing hardware (Vortex vs Kaby/Coffee Lake for instance) would be to entertain questions like "Could Apple ship something like a Macbook Air with a home-grown ARM CPU?"

Very happy to dig into this, but we simply can't do it yet. For as much code as they share, iOS is NOT macOS. They're just not comparable as they stand now, and neither are the applications that run on each right now. I'd love to see how the A12X performs on macOS in something like an Air/Pro body. I'd love it. But we don't have that here. That's just how it is for now.

You think so? I'd actually be quite shocked if internally Apple wasn't a compiler switch/#ifdef away from having a full ARM64 install of MacOS and all the included first-party apps. Much like they did in the past with POWER/x86.

Bingo.

End User wrote:

Bingo.

Not at all, DancinJack's point is orthogonal to SuperSpy's.

I don't doubt, for a moment, that everything Apple has in MacOS has been compiled for their ARM for a long, long time. Years and years.

Stating that they have done this, which I (and others) have been stating is obviously true FOR YEARS, is non sequitor to the point that iOS != MacOS.

It has nothing to do with it.

DancinJack was simply saying there are serious issues with doing a comparison between iOS and MacOS.

Stating that MacOS (and all layered software from Apple) is built on ARM, INTERNALLY, I mean, what? HUH?

That's ...literally... not relevant to what he was saying. At all.

I mean, look:

DancinJack wrote:

I'd love to see how the A12X performs on macOS in something like an Air/Pro body. I'd love it. But we don't have that here. That's just how it is for now.

SuperSpy wrote:

You think so? I'd actually be quite shocked if internally Apple wasn't a compiler switch/#ifdef away from having a full ARM64 install of MacOS and all the included first-party apps. Much like they did in the past with POWER/x86.

WE ARE NOT APPLE. WE DO NOT HAVE WHATEVER APPLE HAS "INTERNALLY" BY LITERAL DEFINITION.

I know that can actually be something really, really hard to understand for those deep in the grip of mindless fanboyism.

But, COME ON. DancinJack even said he'd "LOVE" to have that, he just noted that fact that "WE" don't.


----

This nonsense has to stop. Proudly proclaiming something that NO ONE EVEN SEEMS TO DISAGREE WITH as if it vindicates ANY OTHER UTTERLY UNRELATED POINT, is infuriating.

DancinJack wrote:

until then it's all just speculation.

Speculation, circumstantial evidence, and real world platform comparisons.

End User wrote:

Everybody has a different opinion of what real world usage is. Mine is different from yours.

Sure, your "opinion" on what the "real world" is frequently differs from my "opinion" on the real world. And, you know, most other people's opinions on the real world.

Mostly because it differs from what we can actually demonstrate about it.

You don't make up for this by occasionally stating, with the implication that you're sole purveyor of all truth, things that no one was even arguing over and I'm not even sure if anyone here would even substantially dispute.

Like, if you say the earth is flat, yelling "I MAINTAIN THAT THE SUN IS REAL" at me doesn't... really... Uh.

End User wrote:

Speculation, circumstantial evidence, and real world platform comparisons.

Derfunk summed this up brilliantly earlier, but I'll rephrase:

No one disputes that Apple makes a tablet/phone that is heads and shoulders above everyone else's.

Sure, occasionally other makers have flagships that can compete and perhaps surpass them on certain, usually singular, aspects. But even that's rare.

As an overall platform, there is no comparison. Because, on the whole, they are vastly superior. It's not even close.

----

I really don't think anyone here really disputes this. It's so factually obvious that I don't know how anyone really can.

I mean, personally, want nothing to do with Apple. But that doesn't mean I can't recognize blatant reality.

Let's be blunt, who is the anti-apple zealot here? Who is the android fanboy?

Time and time again, people are saying measured things, with respectful recognition of Apple's position.

Even the dislike, which is only occasionally disparaging, is expressed via carefully qualified philosophical differences towards Apple's image and approach.

Again, I would never (and have never) paid for Apple stuff for my own use, but I've frequently called out how their approach is vastly better for other people and how they make solid products. Heck, even if it's not what I *want*, I even defend some of those philosophical differences from Apple's point of view, because I do believe it is what a -LOT- of people want, and justifiably so (or even SHOULD want, if they actually don't).

---

SO WHY DO WE KEEP DOING THIS?

what is the point?

Topinio wrote:

SuperSpy wrote:

You think so? I'd actually be quite shocked if internally Apple wasn't a compiler switch/#ifdef away from having a full ARM64 install of MacOS and all the included first-party apps. Much like they did in the past with POWER/x86.

Well, it's a compiler switch and a solid emulator ...

Correct me if I am wrong but I think Apple has a handle on getting native apps to run on multiple platforms



and via dynamic binary translation.

Topinio wrote:

which is even harder when the chips you might move to are not actually better.

Compare the A12X to the CPU in the new MacBook Air. My money on the A12X.

The iOS simulator is in no way evidence of your claim. Every iOS developer on the planet knows that packages built for the Simulator are running x86 code with some simulated (note: not emulated) system calls. It's not a virtual device; it's a window into how it SHOULD work on real hardware.

Your link to Rosetta is all about a PPC emulator for x86, which was useful in the pre-Snow Leopard days while Apple transitioned Macs to Intel processors. It's not evidence that an x86 emulator exists on macOS for ARM devices.

There's so much wrong with your post that it doesn't qualify as "wrong" at this point. As in, "you're not even wrong."

Glorious wrote:

SO WHY DO WE KEEP DOING THIS?

I like to dance.

End User wrote:

Correct me if I am wrong but I think Apple has a handle on getting native apps to run on multiple platforms

Rosetta had innumerable restrictions outside of "runs unusably slow".

There were tons of things it didn't do, and architecturally couldn't.

I mean, you keep talking about the real world...

...What do you really know about it?

End User wrote:

Compare the A12X to the CPU in the new MacBook Air. My money on the A12X.

...when running x86 code non-natively?

Again. What are you even talking about here?

Do you understand what other people are saying?

End User wrote:

I like to dance.

You like to stand by the exit, and keep talking about how the Prom King could totally beat Wolverine in a fist-fight.

OK. The Prom King is a great guy! Adored by many, envied by all. Rhodes Scholar, 4.0, 1600 GPA. All-State, perfect season, going to whatever college he wants and unless something goes horrifically wrong he's destined to be a NFL first pick. I voted for him, but it's just not my crowd. But I recognize what I just said---that, at least, is all 100% true.

Uh... OK, sure, like maybe he'd be a fighter pilot ace too, but I don't think he's ever flown a plane...?

Hey.

Why are we always talking about his mythical or hypothetical deeds? Doesn't ANYTHING ELSE ever happen at school?

Because no one wants to dance with you. You never do anything but talk about how they should be dancing with the Prom King!

You never actually dance at all, actually, just state, over and over again, that Prom King totally, absolutely, would dance with anyone who might ask. \

You know, without his feet ever once touching the floor.

derFunkenstein wrote:

Every iOS developer on the planet knows that packages built for the Simulator are running x86 code

Does anyone else in this thread know that?

derFunkenstein wrote:

Your link to Rosetta is all about a PPC emulator for x86, which was useful in the pre-Snow Leopard days while Apple transitioned Macs to Intel processors. It's not evidence that an x86 emulator exists on macOS for ARM devices.

Evidence? Obviously not. It's just an example of what Apple has already done for a previous Mac OS X hardware platform transition.

My point is that Apple has a great deal of experience under its belt when it comes to hardware transitions.

derFunkenstein wrote:

There's so much wrong with your post that it doesn't qualify as "wrong" at this point. As in, "you're not even wrong."

Pure speculation on my part.

People seem to forget that most of Apple's OS tech is already running on ARM and Intel now (and has been publicly since 2006).

The Mac OS X transition from PPC to Intel was a great success. I wouldn't be surprised if Apple were to do something similar if they were to transition macOS from Intel to ARM.

Apple may or may not:

- have macOS running on ARM hardware in their labs
- have an internal build of Xcode that has the option to create macOS apps which provide "Universal Binaries" for both Intel and ARM hardware
- have a Rosetta like dynamic translation layer to allow many Intel applications to run on ARM based Mac hardware without modification (obvious potential for a performance penalty)

Has anyone else ordered a 2018 iPad Pro?

Glorious wrote:

You never actually dance at all

When Apple releases their first macOS on ARM hardware I'm going to be dancing all night long.

End User wrote:

derFunkenstein wrote:

Every iOS developer on the planet knows that packages built for the Simulator are running x86 code

Does anyone else in this thread know that?

I do and even if I didn't and no one else but you and Funk knew that, why on earth why you try to pass it off as something else?

Not a single person has disputed that Apple may have all this stuff behind the scenes. No one. All we can do is see what they publicly put out. The iPad Pros are wonderful looking machines (if you're willing to pay that price for them).

End User wrote:

Has anyone else ordered a 2018 iPad Pro?

Not me. I couldn't imagine paying that much money for a tablet, especially when it's not my primary computing device. No way, no how.

End User wrote:

Does anyone else in this thread know that?

The point is that you completely don't understand what it means.

End User wrote:

My point is that Apple has a great deal of experience under its belt when it comes to hardware transitions.

NO ONE DISAGREES WITH YOU.

This is back to "I MAINTAIN THE SUN IS REAL". It doesn't undo the rest of your endless nonsense.

End User wrote:

The Mac OS X transition from PPC to Intel was a great success. I wouldn't be surprised if Apple were to do something similar if they were to transition macOS from Intel to ARM.

Yes, the sun is real.

What does that have to do with your previous contention that "the Moon is a ridiculous liberal myth"?

End User wrote:

Apple may or may not:

- have macOS running on ARM hardware in their labs
- have an internal build of Xcode that has the option to create macOS apps which provide "Universal Binaries" for both Intel and ARM hardware
- have a Rosetta like dynamic translation layer to allow many Intel applications to run on ARM based Mac hardware without modification (obvious potential for a performance penalty)

I believe, WITHOUT HESITATION, that this is not only true now, but that it has been true for quite some time.

I have been saying this, over and over, for a very long time as well.

To the best of my knowledge, no one here has *EVER* unconditionally disputed this. If I am wrong, and it has happened, it is extremely rare and likely not even remotely recent. And I can point to scores and scores of posts BY THE PEOPLE WHO CONTEST YOU, who either do not deny it, or flatly agree with it.

Such as my case, because I bluntly state that I believe this is true without any qualification whatsoever.

EDIT: In fact, I have every reason to believe that I've actually been saying this longer than you have.

End User wrote:

When Apple releases their first macOS on ARM hardware I'm going to be dancing all night long.

About what?

Are you Apple?

In what way is this your success, and in what way does their success even positively impact you? Can you explain this?

Every day at work I use four completely different computer architectures for general purpose computing (and I'm not even including all the non-general purpose stuff...). That is, I routinely have terminal sessions on all four and regularly have to do something on each.

Do you know how much that typically matters, at all?

idk, once a month? Possibly less?

And I develop stuff.

I'm not using multi-platform applications on a COMPLETELY non-developer, entirely consumer-centric, device.

It barely matters to me.

why on earth does it matter to you?

Glorious wrote:

End User wrote:

Does anyone else in this thread know that?

The point is that you completely don't understand what it means.

No, I think he DOES understand it and was trying to pull a fast one. That's why I took the time to reply in the first place. (I guess I'm basically reiterating what DancinJack said)

End User wrote:

Redocbew wrote:

EU seems to have a different definition of what they think "real world" usage is

Everybody has a different opinion of what real world usage is. Mine is different from yours. Mine changes depending on the workflow and location.

... which is why nobody is going to use photoshop on a tablet. I'm sure Apple will do a great job with the hardware as they often do, but it's not about the hardware. You don't really expect people to create complex clipping paths or layer masks using only a touchscreen for input, do you?

I've got a 1TB iPad Pro on order.

That's really got nothing to do with anything, but great, good for you.

Derfunkenstein wrote:

No, I think he DOES understand it and was trying to pull a fast one. That's why I took the time to reply in the first place. (I guess I'm basically reiterating what DancinJack said)

I can't say that you're wrong, but I don't have any reason to believe that he has anymore understanding that what his picture inherently implies: "hey lookit, iOS running on a MacOS!"

He certainly didn't -say- anything, right? I mean, you were solely responding to the implications that you filled-in with -YOUR- experience and knowledge, not his! Meanwhile, the picture was just in the middle of an indiscriminate shotgun blast of "ROSETTA! WIKIPEDIA EMULATION ARTICLE!"

To more fully establish what I am getting at, despite what the picture implies, he's actually explicitly claimed that he isn't a developer in the past. He certainly doesn't seem to understand it very well.

So, while I can't really prove how much (or how little) he does know, someone like him could have just installed the SDK as part of this whole fanboy bit. I don't really know much of anything about iOS/SDK, but you can do something similar by just installing the android SDK with the emulator component and then clicking any of the virtual devices. Boot, click the browser, take a screenshot. Done. You don't need to know anything about it at all, literally a 1 minute youtube video can get you that far.

---

I've been challenging him like this for a bit, and he's never met it.

If he just don't feel the need to do so, fair enough, but there's really no need to assume what he doesn't actually establish himself.

Maybe, but he's been intentionally misleading in other ways in this thread. Guess it doesn't matter, as long as everyone understands that the iOS simulator is NOT "iOS on Mac"

I just wanted to give him the benefit of the doubt. He's not stupid, he just likes "dancing." Maybe he didn't know. /shrug

Redocbew wrote:

... which is why nobody is going to use photoshop on a tablet. I'm sure Apple will do a great job with the hardware as they often do, but it's not about the hardware. You don't really expect people to create complex clipping paths or layer masks using only a touchscreen for input, do you?

I'm totally unclear on what his work is.

He travels the world on "business", but thinks that VGA ports on projectors are an aberration. That business people would never carry a USB stick---they'd just wireless connect to the projector from their phone and pull all their data down from the cloud! (btw he said this years ago)

He does "computer support" for Mac Pros (but -NEVER- talks to the people that actually use them professionally[because I just have *SOCIAL* contacts with people who do, and I know he's completely and utterly off-base whenever he offers even vague details about this]) but yet this job seems to require the absolute cutting edge in tablets, phones, and laptops.

There's endless ??!?!?! like this.

derFunkenstein wrote:

Maybe, but he's been intentionally misleading in other ways in this thread. Guess it doesn't matter, as long as everyone understands that the iOS simulator is NOT "iOS on Mac"

I believe that he's been intentionally misleading for quite some time.

It's just getting excessively aggravated now because he's really stuck on this specific schtick, and because he seems to think that he can claim anything so long as he responds to criticism by loudly proclaiming that unrelated things no one disputes are TRUE AND HE AWAITS FOR HIS DAY OF VINDICATION OVER THEM!

That's really annoying, and so people are starting to very pointedly poke holes in what he says.

DancinJack wrote:

I just wanted to give him the benefit of the doubt.

I guess I'm past that.

Photoshop on a tablet seems like a decent fit to me, but I'm not an artist and don't have any real-world experience with the Apple Pencil. Maybe Illustrator would have been better. Anything where a Wacom tablet connected to a PC was the best way to go seems like a very natural use case for a tablet, assuming it's really fast enough.

My photo editing doesn't extend past using Photoshop Elements to fix lens distortion and color in raw image files.

I actually didn't know that about the iOS emulator in Xcode, but it's not that surprising considering again the vertical integration of both OS's, the API, and the compiler. It also explains the odd warning about compatibility/accuracy which I just assumed was incomplete API coverage in the emulator or a generic '100% emulators are hard' statement.

My point about Apple porting MacOS was just in response to the argument that comparing raw hardware scores was meaningless. Normally I'd agree (the entire product matters), but that Apple could in theory pull something like that out of its back pocket makes information on absolute performance important to me.

DancinJack wrote:

No, what I thought was lame was that you changed it from the last year's iPad Pro to a competitor to the chips in Apple products.

I don't know why you're being disingenuous about this. You know why a cross platform benchmark, on a different instruction set, on hardware with wildly different specs isn't something to go by. You know that. You also know that I love the Apple products that I own and use daily. I have shilled for Apple plenty of times here when they deserve praise.

I guess if you want to discuss something like "can the iPad Pro replace everything I do with my pro laptop/desktop," then we can do that I guess? But one Geekbench number probably isn't the place to start IMO.

Wait, what's this about cross-platform benchmarks being bad?

I thought the main problem with Geekbench (at least, Geekbench 3) was that it was almost entirely constrained in L1 cache and was overly weighted to cryptography, giving an overweight benefit to phones and tablets. Geekbench 4 fixes some of it, but not all of it (cryptography is still a big portion of Geekbench, but at least its split out into its own category now)

Otherwise, what's wrong about writing a C-program, compiling it for ARM (or x86), and then running the program on both platforms to see how the two chips compare?

Modern OS kernels, be it Linux (aka Android), Windows, or even iOS, are very good about giving CPU time to threads. iOS is a fully virtual-memory platform running a modern OS scheduler and Virtual Memory system (requiring page-translations and everything), so as far as I can tell, it is fair to compare iOS against Windows or Android. Windows's scheduler only interrupts every 30milliseconds or so, and as long as there are an excess of cores, the scheduler will see that there's no other runnable task and stay out of the program's way. So its not like Windows or Linux offers penalties to programs either.

A multi-threaded benchmark would definitely be testing the OS-scheduler. But if we focus on single-threaded results, we ignore that problem entirely! And for the most part, people are focused on the single-threaded results of various benchmarks.

So I have to ask: what is fundamentally wrong about a cross-platform benchmark like Geekbench 4, or SPECint ?? I mean, aside from benchmark related issues (like SPECint doesn't really push FPUs or vector-units very hard. But otherwise it seems like a good representation of 64-bit integer-based code). We've definitely got to analyze the benchmarks to ensure that they're reasonable, but the overall methodology doesn't seem flawed to me at all.

I should note that Geekbench's closed source nature is annoying. Its harder to "trust" Geekbench, but this is more of a error in how the company reports results. As opposed to a fundamental flaw in trying to compare CPUs against each other.

The main thing that matters is whether or not the benchmarks inside of Geekbench4 or SPECint are representative of loads people are actually doing. But Geekbench4 / SPECint are specified at the C-level. The only things being tested are the OS, the compiler, and the chip. And the OS stays out of the way for the most part (aside from maybe... virtual memory... like dTLBs).

SuperSpy wrote:

My point about Apple porting MacOS was just in response to the argument that comparing raw hardware scores was meaningless. Normally I'd agree (the entire product matters), but that Apple could in theory pull something like that out of its back pocket makes information on absolute performance important to me.

It just doesn't make sense to have those kinds of arguments until it's possible. Anything else is just wild speculation. If that day comes, you'll get to evaluate it before you buy it.

And I'm sure that just like PPC -> x86, a theoretical x86_64 -> ARM64 transition will have all first party apps recompiled with universal binaries on day 1. Long-winded theorizing short, Apple isn't dumb and they won't transition without having everything perfectly aligned. Just like last time.