Apple Silicon Isn't What You Think It'll Be

it has been a couple months since wwdc and the unveiling of apple's upcoming in-house silicon max and apple itself has stated that the first of these new intel list computers are right around the corner if not really soon jury's still out on that one certainly by end of year and while you've likely seen a number of videos from your favorite youtubers theorizing about what the future may hold today i want to get down to brass tacks and actually analyze why apple wants to transition the hardware that they've engineered to make that transition happen how software will play a vital role in all of this and why the transition might not be as picture perfect as they've all led us to believe now this is not apple's first silicon transition they moved from 68k to powerpc in the 90s and then powerpc to intel in the mid-2000s and well now intel to apple silicon and in a lot of different ways one could validly argue that the x86 intel transition was by far the most important and that it almost single-handedly helped save the mac in the mid-2000s because apple and ibm's power pc chips had been great until they weren't on one hand some of the chips like the later generation g4 and g5 were desktop powerhouses but they had extreme difficulty fitting into notebook style form factors due to heat and power draw and that was a market that was booming performance per watt just wasn't matching intel or really even close and then on the other hand ibm didn't have the economies of scale that intel did and so apple's alleged unit cost made it essentially impossible to compete with pcs on price so performance was worse and price was worse with this price and performance imbalance a move was necessary and the transition to intel it didn't append the world like a lot of mac fans had originally estimated in fact intel max quickly surpassed what the highest end power pc chips were ever capable of intel chips turns out didn't actually minimize the advantages to mac os it's not like everyone wanted to use windows now and the move to intel actually willed bootcamp into existence something that would help curb worries for pc refugees who wanted to try out a mac but also needed to occasionally run windows oh and well their ads were pretty good that said not all good things last forever and the relationship with intel began to sour about five years ago according to francois piedmont i for sure butchered his name but he is an ex-intel principal engineer kind of a big deal who spilled the beans recently during a casual x-plane game stream according to him apple likely started exploring the possibility of in-house silicon in about 2015 as a result of intel's abnormally bad quality assurance with the skylake launch he said the following and basically her body that apple became the number one filer of problem in the architecture and that went really really bad uh like you know when when your customers start finding almost as much bugs as you you you found yourself you're not living into the right place it's logical to assume that this exploration was only further amplified as intel continually failed to meet its deadlines in early 2016 intel announced to the public that it would be nearing mass production of its new 10 nanometer chip process it's a big deal what it meant is that the chips would become more dense and therefore mount more powerful in the same density in the same package size and more importantly to someone like apple it would be using smaller transistors which means that the power efficiency increases less heat is drawn and less battery is pulled from the cpu this is excellent for laptops so then what's the problem well intel had a couple of delays getting 10 nanometer out the door for nearly three years that's right they just recently started shipping 10 nanometer isolate laptop cpus in july of 2020 with no date on the horizon for 10 nanometer desktop processors ouch now this is just a personal pet theory of mine but i think apple had been banking on intel releasing 10 nanometer laptop skus way earlier than they did why well the 2016 macbook pro redesign went to a thinner form factor with lesser cooling capacity that massively thermally throttled the machine and this was present through the macbook pro lineup for years basically even still in some instances is it reasonable to assume that apple had designed the laptop expecting more efficient chips that just never came maybe it might also explain why the non-retina macbook air stayed stagnant in the lineup for years waiting for something new that just never arrived now make no mistake i don't let apple off the hook because ultimately they are the ones responsible for shipping good products but i do suspect that intel's delays and failed promises may have accelerated the release of the inevitable shift to in-house silicon especially considering the fact that tim cook is a supply chain guy and he does not want his product release cycle dictated by laggard partners now moving apple silicon production in-house does in fact create a number of competitive advantages number one apple no longer has to compete with other oems for the same chips apple is not intel's number one priority by volume not even close that would go to dell and lenovo there they're leaks ahead number two they no longer have to negotiate a feature set that other oems like dell and lenovo may not be interested in for example integrated graphics which apple pushed intel to develop really heavily in the early 2010s instead of opting for more power hungry dedicated gpus is an example where apple won but in other instances they likely haven't number three shifting to apple's own in-house production gives ios apps the ability to run natively on this hardware because there aren't there's no translation layer or software modification required more on that in a minute and number four vertical integration and complete control of hardware and software gives apple the ability to do something that basically no other computer maker has been able to do in decades so let's talk about the hardware shall we what max can we expect to come first and what will their performance be relative to what apple's currently shipping well to answer that i think it's important to look at the dtk or developer transition kit you know that mac mini apple announced during the keynote that you and i can't purchase but the developers can to make sure that their apps are ready for arm excuse me apple silicon the dtk is running an a12z chip which is the same soc found in the 2020 ipad pro which is basically the same chip as the a12x found in the 2018 ipad pro the only difference between the two is that the a12z has one additional gpu core that was also present but disabled in the a12x because binning if you want to learn more info on silicon and how chip binning and all of that other processes are handled well check out my video here there's one big reason that this dtk and a12z is so important and well that's because it's really the only hardware that we have to go off of right now but here's the thing even though it's a two-year-old ipad chip which will never be made into a real consumer mac and craig federicki confirmed this much in the keynote the results are impressive especially in x86 emulation now hold on what are you talking about well in order to gracefully bridge the gap between the x86 outgoing architecture and the new incoming arm or apple silicon architecture apple has developed rosetta 2 which is a huge compatibility software layer that will allow you to run existing x86 binaries or apps on these new arm macs now realistically apple wants developers to recompile their apps to just run natively on arm and luckily they've done a really good job in the last few years pushing developers to xcode and high-level coding languages that use apple developed compilers what this means is that the transition from x86 to arm will be much easier and quicker than the move from powerpc to intel which really wasn't that bad either now on your new apple silicon mac you probably won't know or care if the app that you're using is executing native code or using the rosetta translation layer which is good that's the whole point the reason i'm talking about it at all is that by using a dtk and rosetta two apps we can get a feel for how powerful these rmacs are going to be at least initially early benchmarks posted by developers who had the dtk and violated their nda showed impressive results within about 20 percent of the native ipad pro arm benchmark even without using the four low power cores they only utilized the four high power cores that's kind of how these chips work they're eight core but four are high power four are low power now some people dissed the results but it's like this is literally a two-year-old ipad chip running inside this weirdo dev kit running a beta os inside of an emulated or translated benchmark what were you expecting what is monumental is the following even though the dtk didn't match the ipad pro running the arm benchmark natively and of course it wouldn't it still outperformed the surface pro x with the qualcomm microsoft sq1 arm processor running the same benchmark natively and additionally the gpu course found inside of the a12z inside of this weirdo dev kit actually outperformed the ryzen 4500u laptops in an opencl test that's also really impressive so let's digest this all for a minute okay the a14 soc that'll make its way into the new iphone will have a rumored 15 billion transistors on a five nanometer process and that's up from the a13s eight and a half billion transistors on the larger seven nanometer process now the rumored implications of this are huge a thirty percent increase in speed and efficiency in addition to the silicon team's annual improvements and optimizations this is just for the iphone so if we assume two additional cores for the a14x ipad variant as has been true in the past along with double the gpu core count from four to eight this would put performance near the top of the single core charts in geekbench in nearly every performance category and in multi-core scores it would nearly match the performance of the octa-core intel i9 9980h laptop cpu and extended multi-core workloads when running natively compiled arm apps and it will do it with significantly less power draw and heat generation because after all this is going to be an ipad chip now let's say they take that exact chip the a14x the ipad chip and throw it inside of a macbook pro native apps will run basically the same as ipad but even the emulated x86 apps will be expected to run around the same speed as the 2019 13 inch macbook pro with i7 85 59u processor this is extraordinarily impressive however i do think that we need to keep our expectations in check i think a lot of people believe apple's going to come in balls blaze and unveil a computer with the same compute power as like i don't know a ryzen 3700x in a laptop the size of macbook air look it's not going to happen my prediction is the following apple announces a new arm mac later this year in a form factor that we're already familiar with like the 13-inch macbook air or 13-inch macbook pro now the machine will use a chip based off of the new a14 chip and it will provide similar if not marginally better performance than the outgoing intel macbook pro so what will they advertise as the selling point if it's only a little bit faster well they'll advertise a thinner fanless design with extraordinarily good battery life and because it runs native ios apps it might even get a touchscreen oh and it's still going to be faster when running arm benchmarks it will be like 40 faster so that will be an impressive thing to say on stage look i too hope that this is all wrong and that the first apple silicon mac uses a max specific chip not derived from the a14 that just obliterates every laptop on the market and look that may happen eventually the implications of apple silicon are massive and so it could happen into the future but i expect this first year to be a little slow expect a low tdp relatively low performance laptop to come first and then lower and imax to come and then finally we'll get high performance machines like the mac pro very last i mean look imagine an arm mac pro with the ability to add all sorts of crazy expansion slots like independent cpu cores or fpgas to help in specific workloads the options are limitless once apple is in charge of their own architecture designs i think stuff like that afterburner card available in mac pro that's utilized for prores workflows that's just the beginning of this all and it is not inconceivable to me that it will be possible to build a mac pro in the future tailored exactly to your specific workload on bare metal rather than having to buy a generic cpu and trying to make stuff work in software boy that would be neat wouldn't it then again maybe they won't because apple has demonstrated time and time again especially in the mobile space that while their cpus are special sure they're really optimized in software and it's the optimization of hardware to run code where apple's products truly begin to shine this transition is going to be an exciting one and you can bet that we're going to be covering it here on snazzy labs later this year so get subscribed if you haven't already and enable the bell notifications so you don't miss any new videos we have some really exciting ones coming soon let me know what you think in the comments down below is the new apple silicon mac gonna be crazy insane or is it going to be relatively modest if you enjoyed this video leave a like if you didn't while that other button seems to work okay too get subscribed for more awesome videos like this one but most importantly and as always stay snazzy

Channel: Snazzy Labs

Views: 672,696

Keywords: snazzy labs, quinn nelson, snazzyq, apple silicon, arm mac, arm, silicon, apple silicone, apple mac, imac, new mac, macbook pro, macbook air, arm macbook, apple silicon macbook, a14, a14x, windows arm, mac os arm, mac os big sur, big sur, windows 10, nvidia, qualcomm, intel, 10nm, 7nm, 5nm, iphone 12, soc, gpu, cpu, computer, instruction set, new apple mac, mac, mac pro, fpga, asic, processor, amd, keynote

Id: 6VrJAkIaTfQ

Length: 13min 59sec (839 seconds)

Published: Mon Sep 14 2020