commentzilla

mretondo said:

Apple made it perfectly clear that ARM is slower than Intel. In fact they screamed it out load if you were listening. There wasn’t a single side by side demo of FCP X running on two Macs doing a long task like rendering. Intel takes 2 minutes while ARM takes 1.5 minutes. That’s all it would’ve taken to show how fast ARM is. The reason, ARM is slower!

I don't think there was a side by side demo because the A12Z isn't going to be the shipping chip. I'm betting on a beefier design meant to wow the crowd.

That aside, my 2020 13" MBP 10th GEN i7 can't run Tomb Raider at 30 fps in 1080p. That told me everything I needed to know. Most everything they showed would choke my brand new MBP. While the graphics are impressive they'll need a beefier design to eliminate AMD/NIVIDA dGPUs.

anonconformist said:

commentzilla said:

anonconformist said:

rmusikantow said:

darthw said:

Will it be possible, eventually, for Apple to make faster SoCs than the fastest most powerful intel Xenon chips?

Yes. I just read that the new Japanese super computer, that is the fastest in the world is built using Arm chips.

can Apple eventually make their own SoCS to beat Intel Xeons? There are reasons that it could go either way:

ARM ISA is easier to decode is in its favor.
intel z86-64 ISA is more compact due to variable length instructions that reduce memory bandwidth required for a given number of of instructions that achieve the same thing.

I think there is zero chance Apple will stay with INTEL anything and they will have no issue out performing INTEL Xeons within the same power and thermal constraints.

INTEL chips bring a lot of baggage to include more complex instructions so I doubt they're more compact and while fixed length instructions may require more bandwidth.... 

"Fixed-length instructions are less complicated to handle than variable-length instructions for several reasons (not having to check whether an instruction straddles a cache line or virtual memory page boundary, for instance), and are therefore somewhat easier to optimize for speed."

Fast cache is heavier for power usage as well as space, and taking more space requires more time for the whole system to some degree, not to mention the more space a chip takes, the more expensive it is to make, as the higher number of defective chips you’ll have on any given die.

Space matters, and if they can deal with the variable length instructions in less decoding logic space/power than the caches required, it’s a win.  It’s not just about bandwidth to/from main memory, also internally, and size matters.  Resistance and capacitance increase with the size, and that increases power while decreasing speed as well: it makes the most sense to use the least total hardware regardless of the complexity of decoding basic instructions into the micro-ops, of which there are far fewer micro-ops ever in-flight at any given time than CPU caches storing ISA machine code and its data.

So far Apple appears to be winning with RISC. The A-12 is already matches at the 10th Gen i7 Quad-Core in single, multi-core, and beating it hands down in GPU performance. One of the reasons Apple is moving on.

mattinoz said:

grayfox691 said:

I find this all rather exciting because unlike in the past where we could look at the future roadmap for PowerPC and Intel we realistically have limited understanding of what apple's new silicon could look like. That goes especially for their higher-end Macs. The chip used in the developer Mac mini looks to certainly be capable for MacBook Air and other compact portables however I'm extremely interested to see what those higher-end chips could look like. 

Given the transition is scheduled to take around two years that would indicate to me that the architecture of all their Mac-based chips is more or less finalized. The new silicon should provide incredible versatility for making many processes more efficient. However, I'm curious if for the high-end iMac or Mac Pro if Apple will be motivated to silence critics worried about high-end performance. The possibilities are wide open for what kind of monster Apple could create. The Mac Pro has a lot of thermal capacity. So at the end who knows but for me at least I'm quite excited.

I guess it's going to depend on the specialist processors in the A (like ML Accelerators) and how much it makes sense to scale them up vs having small repetitive elements. Is it 2 -3 chips each to suit a part of the Mac family then bined and clocked to suit each model or do they go the other way and have 2-3 chips each to suit a roll and each machine ends up a combo of those parts?

I do wonder If Apple might be seeing it as the later. A Family of chips as kit to build each machine instead of a part for each machine, say 3 chips 2 Bookends that go in every mac (basically versions of AX and T) and the middle one that can be used in multiples. With maybe a bigger middle chip for the MacPro charging an extra $1,000 per middle chip.

Add some 3rd Party GPU choices in that mix and it will be interesting to see how Apple does do this over the next 2 years.

I guessing at most three additional chip with different clock speeds; one for low end MacBooks, iMacs and MacMini, another for high-end MacBook pros and iMacs, with another for the MacPro. Would be surprised to see the the MAcPro and iMac Pro sharing that third chip at different clock speeds. But at most 4 chips in addition to the two produced each year for iOS devices.

As for GPUs I don't expect to see 3rd party GPU options for anything other than the MacPro and eGPUs functionality for everything else. While everyone is so focused on the A-Series CPU the GPU is making the same advancements. I would also expect USB-4 which is probably why it's Thunderbolt ports are absent from the developer kit or possibly just to same money.

Apple has a lot to gain by dumping INTEL, AMD, NIVIDA and QUALCOMM from both a development and cost savings point of view. How much cheaper is a 16" MBP if Apple is only paying $50-$60 to produce the chip vs. paying INTEL and AMD $100s of dollars more. A 8-core i9 is $500 retail alone. Apple I'm sure gets a steep discount but combined with the GPU cost, Apple could stand to save $500 simply by using an in-house chip for under $100. Apple could take the margin or knock the price of the machine down, expand the base, and profit more off of software and services. I betting they want to expand the user base which will become life long customers.

rain22 said:

commentzilla said:

rain22 said:

“ but it suggests that new Apple Silicon Macs will not be struggling to keep up with the graphics on Intel Macs.”

That would be nice - but seems extremely dependent on programs being optimized. The anemic library of titles will probably shrink even further - at least until there is market saturation. 

Mac users will be stuck using dumbed down iOS software for a long time I feel. 

After all - This is the motivation isn’t it? Eventually have just 1 OS that can be modded to facilitate the device. 

iOS and macOS share the same core and were designed to be processor independent. This is why existing apps can be modified and recompiled for ARM in days. I suspect the performance will be far better than you think. That Tome Raider demo spoke volumes. A brand new 2020 13" 10th Gen i7 Quad-Core can't even do that with native code.

I hope you are right. We got burned on a bunch of G5’s during the last switch. Rosetta didn’t work half the time, crashes, software wasn’t supported, and Apple dropped support on its own suit almost overnight. Peripherals became junk as no drivers were updated and Apple put the whole onus on developers and manufacturers. We ended up tossing the G5’s for a big loss and getting the new Mac Pro’s. At least then we could upgrade our own video cards.

A lot of difference between then and now. Apple has been running iOS (macOS) on ARM for a decade now and operating has been designed to create a layer of abstraction from the hardware. This is why many apps will only need to be recompiled or can run in translation. That Tome Raider demo was a x86 mac app running without modification in translation via Rosetta 2. The old Rosetta couldn't do anything like that since apps were much closer to the hardware in those days and Metal didn't exist yet.

As for INTEL updates, I assume any app written natively for ARM can be recompiled for INTEL which means developers will easily be able to support both platforms. That's the beauty of the abstraction layer. The only question then becomes how long will Apple support new versions of macOS on INTEL since that occurs at the hardware level. I suspect it depends on the install base and their traditional obsolete/vintage status for hardware; 5 years of full support and 2 additional years minimum for security updates.

5-7 years is just about what I expect to get out of a device. Come this Fall my 7+ year old 2013 15" MBP will no longer be supported by the current operating system, which means it's down to security update status. I got my money's worth and the resale value of these machines is likely to remain high since Bootcamp is gone forever.

techconc said:

braytonak said:

While a new ARM-based MacBook is logical, I would think it would also reinforce the expectation that %desktopOS%-on-ARM = slow. Apple’s confidence in ARM would be clearer if they put it in a MacBook Air, which we already know is a capable machine. 

Either way, I would replace my 2015 and 2017 MacBooks with an ARM-based model if they ditched the butterfly keyboard in them. If this comes to fruition this year I will find it a very fascinating time, indeed. 

I really don't understand why there is this perception that ARM is on desktop would be slow.  What leads you to think that?  The single core performance of Apple's A13 chip is within 6% of the performance of Intel's fastest i9 core.  That's with a low wattage mobile chip running on a phone as compared to a much higher wattage desktop based chip from Intel.  The A14 will likely exceed what Intel can do at single core performance and they'll do it on a phone.  On the desktop, it's just a matter of Apple providing a chip with more cores.   Also, as it stands now, the 2 year old A12x based iPad pro is more powerful than a brand new Core i5 based MacBook Air.  Provided we have native applications, performance will not be an issue.

Also, Apple has already moved away from the butterfly keyboard in all of their models, so that concern is completely a non-issue.

The A13 in Geekbench has about the same overall performance (single ,multi-core and graphics) as the 10th Gen Quad-Core i5 in the 2013 Macbook Pro. I imagine the 5nm A14 is comparable to the i7 (all tests) and the i9 (single-core).

That's pretty good considering the thermal design is twice as efficient at the i5 (7w vs 15/25w).