New iPad Pro Has Comparable Performance to 2018 15" MacBook Pro in Benchmarks

[cmaier]

OK cool, name a RISC architecture.

And before you offer ARM: FJCVTZS has made the idea that ARM is still RISC completely absurd.

How about I name some RISC architectures for which I've designed actual microprocessors?

MIPS, F-RISC, PowerPC, ARM, Sparc

FJCVTZS is a RISC instruction. It reads from registers, performs a pipeline computation, and puts the result in a register. It does not require multiple trips through the pipeline, it does not read operands from or write results to memory, it does not need to be broken into micro-ops that are individually scheduled and retired. It does not have an arbitrary or variable instruction length. Its operands are found in the same location as other instructions. The instruction decoder need not be stateful to understand the instruction.

"RISC" doesn't mean "simple instructions" or "few instructions."

 

[LeahJoFoxtrot]

Can some explain why I want one of these when I don't even use the iPad I have? I've owned several iPads and I just can't find a good use case without feeling like I'm neglecting another device (like my phone)

 

[falainber]

For some details on the actual test results one might want to look at these article: ExtremeTech. The actual performance may not be as close as GeekBench results suggest.

Quote: "When you actually look at specific test results, there are areas where the Intel CPU is dramatically faster than the ARM equivalent... The general Geekbench score predicts just an 8 percent single-threaded gap between the eight-core A12X Bionic (four high-efficiency cores, four low-efficiency cores) and the four-core, eight-thread Core i7-8559U. This elides the fact that the 8559U is more than 2x faster in certain tests like SGEMM that clearly aren’t weighted very much and 1.72x faster in SFFT. Intel remains 1.2x faster in LZMA and 1.27x faster in JPEG decoding, and 1.46x faster in PDF rendering."

And then: "Results like this draw attention because they feed a narrative that Intel is losing ground to Apple. There are persistent rumors that Apple will start swapping Intel CPUs for its own silicon in 2020. From there, it’s easy to connect the dots and think that this is evidence of Intel’s own performance collapse, the end of x86, etc. Digging deeper into results often gives a more nuanced picture of what’s going on and where the limits and problems are. For example: One potential reason these results favor Apple is that Apple is still building its laptops with DDR3-2133, while its iPads use LPDDR4 at higher clocks. In theory, a laptop with DDR4-2400 instead of DDR3-2133 would perform a bit better in these tests."

So, ironically the test scores might be explained by the deficiencies of Apple hardware (in MBP) rather than Intel CPU.

 

[firewood]

Windows is still the largest bulk of the computer use around the globe.

Likely not. There are many more Android users worldwide. Some very high population countries have more mobile devices than PCs. Likely followed by Linux in data centers and wifi access points, web cams, etc. Even corporate customers often use their Windows PCs mainly as viewers into applications actually running in high powered corporate data centers, not on their slower PCs.

Apple has tons of analytics on how their Macs are used; and likely knows exactly how tiny a percentage of high end gamers and specialty power users actually boot their Macs into Window. It's not enough to keep Apple from pursuing more profitable opportunities (better battery life, more features (MLKit, et.al.) , less Intel markup).

 

[falainber]

Likely not. There are many more Android users worldwide. Some very high population countries have more mobile devices than PCs. Likely followed by Linux in data centers and wifi access points, web cams, etc. Even corporate customers often use their Windows PCs mainly as viewers into applications actually running in high powered corporate data centers, not on their slower PCs.

Apple has tons of analytics on how their Macs are used; and likely knows exactly how tiny a percentage of high end gamers and specialty power users actually boot their Macs into Window. It's not enough to keep Apple from pursuing more profitable opportunities (better battery life, more features (MLKit, et.al.) , less Intel markup).

Apple probably has analytics data but what makes you think that the percentage of high end gamers and specialty power users actually boot their Macs into Window is tiny? Maybe it's huge. Have Apple indicated that they want to switch to ARM for Macs or something?

 

[smulji]

Can't speak for the future, but I do know that 99% of college students at University of Texas at Austin use laptops, not tablets, as their go-to work/study and entertainment computers. I rarely ever see students, staff, or faculty using tablets. Most students seem to purchase high end laptops toward the beginning of their college studies with the idea of retaining them throughout their degree pursuits. A definite trend over the last two years has been an uptake in the use of pc laptops over Apple laptops. University people WANT PORTS, and they want touch screen capability in conjunction with full laptop functionality (robust keyboards and touchpads). Most students are in their 20's, so that might qualify as a "future" predictor. My gut feeling is that ultimately some hybrid mix of laptop/desktop and tablet portability will be "in the stars". With current offerings, though, the laptop is not going away anytime soon. Smart phones are the portable device of choice, laptops the workhorse for university people. As a retired techie geezer from said university, I still value one nice desktop (currently a year old iMac), three PC laptops running Linux, a vintage 2006 (1,1) MacBook Pro that I keep around in awe of its longevity, an iPhone 6S+ and LG V40 smartphones (both have headphone jacks). The iMac is my anchor, the laptops my workhorses, and the phones for glue in between the others. Oh, and my iPad Pro I use mainly for music activities, both for listening and performing. Maybe someday some virtual hybrid device / thingie will combine the functionality of all these things, but someday isn't here, yet.

" Maybe someday some virtual hybrid device / thingie will combine the functionality of all these things, but someday isn't here, yet."

You mean like a Surface Pro?

 

[Mitthrawnuruodo]

Can some explain why I want one of these when I don't even use the iPad I have? I've owned several iPads and I just can't find a good use case without feeling like I'm neglecting another device (like my phone)

I’m more or less in the same situation: I use my iPad mini almost exclusively when travelling, and use my iPhone or MacBook Air on a daily basis.

That said I think this is existing for two reasons:

1) for those (other than us, like my partner) who uses their iPad Pro for drawing, photo editing and other creative reasons (and others who use it for other heavy duty stuff like gaming, video editing, etc). They will benefit from a very capable brain in their iPads.

2) for the rest of us, that prefer (or needs) a computer, the prospect of cheaper, faster and more energy efficient CPUs that Apple can custom design to fit their ecosystem of laptops (and presumably desktops) makes me very optimistic about the future of Macs. Having experienced a couple of Apple’s architectural changes in the past without any (noticeable) problems makes me quite comfortable about Apple’s ability to make yet another smooth transition.

 

[MacBH928]

x86 emulation existed on the PPC Macs, I ran VirtualPC on one. It's nothing new. You could even run DOS on the PCTransporter board in an Apple ][.

It wasn't great but it worked; and MS has a Windows/ARM solution already, even if it doesn't run on Apple's ARM. I have no doubt someone will offer a solution to run Windows on ARM if and when Apple goes to it as their sole processor. I am also not sure how much of a difference being able to run Windows in a VM really is in a buying decision. I've run Parallels/Fusion since the old white plastic Macbooks but I'd guess only a tiny fraction of today's Mac users really need it and would switch to OPCs if that capability went way.

Emulating isn't like virtualization. People buy macs today with the comfort that Windows is there if they need it. If people stop buying macs(high prices, weak spec, lack of update, butterfly keyboard, touchbar, and now no Windows) it will have a reverse effect. The less people opt for a mac, the more that will follow in abandoning the product.

Arm supports virtualization. Nobody (statistically) cares about bootcamp anymore. Windows is far less important than it used to be. iOS compatibility is far more important.

And much old software will be ported or recompiled. Much of it has already proven portable. Even photoshop, autocad, office, etc.

When Mac came out people lamented “but it won’t run my apple ][ software!”

Life moves on.

Not sure about porting and compilation to a new form of processor, but I know there is a huge software library out there that will not work. Think of all the games on Steam and GOG, as for software that is popular like Photoshop and FireFox I am sure we will see the ports but it will take time... a lot of time.

I remember it was like few years until some software was converted to the intel system. I think QuarkExpress was one of them.

 

[chucker23n1]

So, ironically the test scores might be explained by the deficiencies of Apple hardware (in MBP) rather than Intel CPU.

Ah, but whose fault is it that Apple can’t put LPDDR4 in their laptops? Not Apple’s.

 

[torana355]

This is yet again one of those things that people on the Apple-hating bandwagon will ignore.

It's not comparable. You can't compare benchmarks on ARM to x86.
[doublepost=1541245652][/doublepost]

Geekbench scores are not meaningful compared across platforms. Apple A series Geekbench scores can be compared to other A series, perhaps Qualcomm ARM CPUs, but not to x86-64.

Do you really believe that Apple can make a 5W fanless CPU that performs that closely to a 45W CPU? If you do, I suggest cutting back on the kool-aid.

I'm glad someone is this thread has some sense. Apple fans really belive the Apple ARM chips are at Intel level

 

[jlc1978]

Emulating isn't like virtualization.

I know, thus my point about Windows/ARM still having the potential to support VMS on an ARM Mac.

People buy macs today with the comfort that Windows is there if they need it. If people stop buying macs(high prices, weak spec, lack of update, butterfly keyboard, touchbar, and now no Windows) it will have a reverse effect. The less people opt for a mac, the more that will follow in abandoning the product.

I am not sure the ability to run Windows is that much of concern for most Mac buyers; compatibility with Windows programs such as Office would be much more important IMHO. As long as the ability to seamlessly exchange fiels exists the underlying OS is of little importance; what matters then is the user experience.

as your rightly point out, building products that don't stand out as a better value than competitors will result in lost sales and potentially a downward spiral into oblivion. Good bye all this, hello oblivion.

Not sure about porting and compilation to a new form of processor, but I know there is a huge software library out there that will not work. Think of all the games on Steam and GOG, as for software that is popular like Photoshop and FireFox I am sure we will see the ports but it will take time... a lot of time.

I suspect it you'll see something similar to what happened with the previous processor changes. Popular (read profitable) programs will be ported and those that aren't disappear. It could have a major impact on OSS since if it requires complete ground up rewrites of code the interest and time may simply not be there.

I remember it was like few years until some software was converted to the intel system. I think QuarkExpress was one of them.

Some of the ports, as I recall, retained the Win GUI design and looked horrible on a Mac.

The iPad Pro is the real question mark in this scenario. If enough companies start developing ARM based versions of their Intel programs to run on powerful iPad Pros, Apple could have an existing base of programs to support a rollout of ARM based Macs that would only need to be recompiled with minor changes to run on an ARM based OS X. That scenario is different from previous processor switches since those did not have an Apple OS already running on the new processor and thus no installed base of programs.
[doublepost=1541250785][/doublepost]

It's not comparable. You can't compare benchmarks on ARM to x86.

Of course we can. This is a forum, where our technology choices and opinions define us and we have to prove our choice was better than yours. Thus, we can argue endlessly over arcane specs that have no impact on real world use while the rest of the world simply goes about doing real work with the technology.

 

[falainber]

Ah, but whose fault is it that Apple can’t put LPDDR4 in their laptops? Not Apple’s.

Real "pro" laptops use DDR4. Fake "pro" laptops use excuses.

 

[chucker23n1]

Real "pro" laptops use DDR4. Fake "pro" laptops use excuses.

They could've compared the iPad Pro against the 15-inch MacBook Pro, which has DDR4 at 2400 MHz.
[doublepost=1541265247][/doublepost]

It's not comparable. You can't compare benchmarks on ARM to x86.

If you're going to make a claim like that, try and substantiate it.

I'm glad someone is this thread has some sense. Apple fans really belive the Apple ARM chips are at Intel level

On some benchmarks, they are beyond Intel level. An iPhone XS is faster than an iMac Pro at Apple's JavaScript implementation. In part because JavaScript is entirely single-threaded (service workers aside), so the many cores of an iMac Pro aren't helping any, but still.

Make of that what you will. It doesn't mean the iMac Pro will ship with an Apple CPU any day soon. But it does mean precisely what the headline says: "New iPad Pro Has Comparable Performance to 2018 15" MacBook Pro in Benchmarks"

 

[torana355]

They could've compared the iPad Pro against the 15-inch MacBook Pro, which has DDR4 at 2400 MHz.
[doublepost=1541265247][/doublepost]

If you're going to make a claim like that, try and substantiate it.



On some benchmarks, they are beyond Intel level. An iPhone XS is faster than an iMac Pro at Apple's JavaScript implementation. In part because JavaScript is entirely single-threaded (service workers aside), so the many cores of an iMac Pro aren't helping any, but still.

Make of that what you will. It doesn't mean the iMac Pro will ship with an Apple CPU any day soon. But it does mean precisely what the headline says: "New iPad Pro Has Comparable Performance to 2018 15" MacBook Pro in Benchmarks"

You have no idea what you are talking about sorry. You can't compare ARM benchmarks to x86.

 

[cmaier]

You have no idea what you are talking about sorry. You can't compare ARM benchmarks to x86.

Why not?

If I want a way to know whether an ARM machine is faster than an x86 machine for, say, compiling code, why can’t I create a benchmark that times how long it takes to compile code on each platform? And if I’m also interested in how long it takes to perform a SPICE simulation, why can’t i time how long it takes to perform the same simulation? Or run a suite of JavaScript? Or render a 3D scene? Or ....

That’s all these benchmarks do. They measure how long it takes to produce identical outputs from identical inputs. The results tell you only how fast one is compared to the other for the tasks specifically tested. But they absolutely can be compared. Hell, half my career was comparing sparc to x86 to powerpc.

 

[torana355]

Why not?

If I want a way to know whether an ARM machine is faster than an x86 machine for, say, compiling code, why can’t I create a benchmark that times how long it takes to compile code on each platform? And if I’m also interested in how long it takes to perform a SPICE simulation, why can’t i time how long it takes to perform the same simulation? Or run a suite of JavaScript? Or render a 3D scene? Or ....

That’s all these benchmarks do. They measure how long it takes to produce identical outputs from identical inputs. The results tell you only how fast one is compared to the other for the tasks specifically tested. But they absolutely can be compared. Hell, half my career was comparing sparc to x86 to powerpc.

Yes you are right but you can't directly compare them when they are running on 2 totally different operating systems, The iPad is running on iOS and the X86 benchmarks are on a full fat operating system like OSX or Windows. Do you really think the Apple ARM chips would still outperform the Intel chips if they were both running on a full fat OS like OSX? The benchmark score of the ARM chip would fall dramatically. Every instruction set has its advantages one over the other, and they will always have different scores because every test is tailored for that specific platform.

 

[cmaier]

Yes you are right but you can't directly compare them when they are running on 2 totally different operating systems, The iPad is running on iOS and the X86 benchmarks are on a full fat operating system like OSX or Windows. They are both optimised for a specific purpose. Do you really think the the Apple ARM chips would still outperform the Intel chips if they were both running on a full fat OS like OSX? No way in hell they would, a 5W chip is not going to outperform a 45w chip for example.

Huh?

Benchmarks show only what they show. They represent a workload and you can directly compare, regardless of operating system, and get an accurate understanding of the relative performance of that workload. A benchmark does not purport to provide any information about workloads that it doesn’t test.

What difference does operating system make? Nobody runs an operating system solely to “operate.” They do it to run workloads. Video encoding, circuit simulations, code compilation, javascript interpretation, whatever.

Geek bench happens to test some complex workloads. Those workloads can be performed on both iOS and macOS.

If you are saying “x86 would run macos faster than a12x,” well maybe. The benchmark doesn’t purport to say otherwise. But it does say that certain apps would run faster on ARM.

And as a CPU designer, I’m pretty confident a clocked up and Active cooled a12x would run macos faster in a Mac than the i7 in my 2016MBP.

 

[torana355]

Huh?

Benchmarks show only what they show. They represent a workload and you can directly compare, regardless of operating system, and get an accurate understanding of the relative performance of that workload. A benchmark does not purport to provide any information about workloads that it doesn’t test.

What difference does operating system make? Nobody runs an operating system solely to “operate.” They do it to run workloads. Video encoding, circuit simulations, code compilation, javascript interpretation, whatever.

Geek bench happens to test some complex workloads. Those workloads can be performed on both iOS and macOS.

If you are saying “x86 would run macos faster than a12x,” well maybe. The benchmark doesn’t purport to say otherwise. But it does say that certain apps would run faster on ARM.

And as a CPU designer, I’m pretty confident a clocked up and Active cooled a12x would run macos faster in a Mac than the i7 in my 2016MBP.

I think you are oversimplifying things. The OS does make a difference as every test on say geek bench is tailored to that specific platform. The only real way to compare them is running the same OS imo.

On your last point, if that were really the case i'm pretty sure Apple would have made the jump already but i'm sure the gap is closing.

 

[cmaier]

I think you are oversimplifying things. The OS does make a difference as every test on say geek bench is tailored to that specific platform. The only real way to compare them is running the same OS imo.

On your last point, if that were really the case i'm pretty sure Apple would have made the jump already but i'm sure the gap is closing.

No, the geekbench workloads are identical across platforms. That is entire point of it. But don’t take it from me. Take it from GEEKBENCH’S WEB PAGE:

Designed from the ground-up for cross-platform comparisons, Geekbench 4 allows you to compare system performance across devices, processor architectures, and operating systems. Geekbench 4 supports Android, iOS, macOS, Windows, and Linux.
​

Read the first nine words a few times.

Yours are entitled to your opinion, but you are making up facts, it seems.

 

[torana355]

No, the geekbench workloads are identical across platforms. That is entire point of it. But don’t take it from me. Take it from GEEKBENCH’S WEB PAGE:

Designed from the ground-up for cross-platform comparisons, Geekbench 4 allows you to compare system performance across devices, processor architectures, and operating systems. Geekbench 4 supports Android, iOS, macOS, Windows, and Linux.
​

Read the first nine words a few times.

Yours are entitled to your opinion, but you are making up facts, it seems.

I suggest you do some more reading up on just how accurate Geekbench actually is at doing what they claim. It is by no means perfect. As a CPU designer I would think you would know this...

 

[cmaier]

I suggest you do some more reading up on just how accurate Geekbench actually is at doing what they claim. It is by no means perfect.

Straw man. I didn’t say it was perfect.

But I provided a source for my contention. I’m happy to read sources that say that you can’t use geekbench for cross platform comparisons, if you want to provide links.

 

[torana355]

Straw man. I didn’t say it was perfect.

But I provided a source for my contention. I’m happy to read sources that say that you can’t use geekbench for cross platform comparisons, if you want to provide links.

You can use it for Cross Platform as a guide , sure. My point is everyone reads the headline that the iPad is as powerful as the MacBook Pro but there is more to it then that. Benchmark scores like Geekbench aren’t a perfect way to test how fast a computer is, especially when comparing ARM chips like Apple’s A-series to Intel’s x86-based chips. Im not making up facts here this is just reality.

 

[KPOM]

It's not comparable. You can't compare benchmarks on ARM to x86.
[doublepost=1541245652][/doublepost]
I'm glad someone is this thread has some sense. Apple fans really belive the Apple ARM chips are at Intel level

Except that the writers of Geekbench claim that it is intended to be cross-platform.

 

[falainber]

Except that the writers of Geekbench claim that it is intended to be cross-platform.

"Although Geekbench is cross platform, I wouldn't recommend using this data to do anything other than compare iOS devices. I've looked at using Geekbench to compare iOS to Android in the past and I've sometimes seen odd results."
That's from mr. Anand Lal Shimpi of Anandtech fame. He is long gone from there but Anandtech still never uses Geekbench for comparing performance of CPUs (even those with the same architecture) on different platforms.

 

[Feenician]

"Although Geekbench is cross platform, I wouldn't recommend using this data to do anything other than compare iOS devices. I've looked at using Geekbench to compare iOS to Android in the past and I've sometimes seen odd results."
That's from mr. Anand Lal Shimpi of Anandtech fame. He is long gone from there but Anandtech still never uses Geekbench for comparing performance of CPUs (even those with the same architecture) on different platforms.

That comment would be about Geekbench 2 or 3. Linus Torvalds also had a big issue with these versions as they had some tests which ARM in general and A* in particular had been optimized for. I’m less sure that mobile devices being optimized for tasks that mobile devices often do is an issue but, hey, it’s Linus so I’ll defer.

Geekbench 4 is much improved and is the version referenced in this article.

https://www.realworldtech.com/forum/?threadid=159853&curpostid=159860