Well M1 has the advantage of going into an entirely known and optimized-for system.
Intel needs to go into multiple kinds of motherboards and work with billions of combinations of other hardware.
For Apple building M1 is a much more integrated approach which is why they can produce better performance. If they built a similarly general purpose chip as intel they would likely not produce something as good as M1 or current Intel.
M1 is less than 5% of than the total chip market, intel are obviously not doomed, intel/amd/arm is powering like the other 95% including billions of servers etc. that M1 will never compete in.
No consumer buying intel is suddenly going to not buy intel because it isn’t on par with M1, unless they were buying a mac anyway.
Phones and laptops are replacing good old fashioned desktops. The demand for efficiency and battery life is only going to grow. Even in server rooms, efficiency matters, they are a huge power draw.
I think Apple is in a great position to ride this wave as they are already surfing it. Even down to the languages they use avoiding GC which reduces memory requirements which in turn produces better battery life.
It certainly does necessitate the need. Apps and use cases keep advancing, which introduces more complexity, which requires faster CPUs, which requires better technology to fit that into the exit power budget. This is... why OEMs keep putting faster processors into new products.
Consumer battery technology isn't advancing at pace with increased compute.
I'd beg to differ; apps and use cases are not advancing, which is why even phones from 5 years ago are perfectly capable of running modern apps. Once mobile video editing became reasonably possible, we pretty much ran out of uses for the smartphone. Everything else had been done already.
What is out there that requires faster CPUs now? AR? Does TikTok use more CPU power than I thought or something?
Intel has their fingers in a lot more pies than just consumer CPUs, so no, they're not doomed even if Apple gets and stays ahead in every metric there. AFAIUI, Apple's chips are optimized for energy use and can't compete on power, and the really impressive thing is their system design and integration. Intel also makes bare CPUs available to businesses and the general public, Apple does not. They're not really competing in their business areas.
One crisis at a time. They're currently coming back from their 10nm process failure.
Anyway, M1 is only competitive in the market for processors for Apple computers. Unless Apple makes a enormous change to their business practices, I don't see them selling their CPUs to others. Apple doesn't have a good history of corporate support that would get enterprise buyers to buy their servers, if they had servers to sell, etc. Look how hard of a time AMD has selling server CPUs, at times where they've got a clearly superior CPU (first few years of opterons, currently with Epycs); it takes more than just a superior CPU to sell server CPUs.
No? I'd even argue that Apple is the doomed one here:
- ARM is still a second-class architecture in the world of software. I work with Mac packaging every day, and M1 is still pretty poorly-supported, all things considered.
- Apple has pretty much no way to increase their performance besides horizontal scaling or increasing their silicon density. That puts them in an awkward spot, and it's probably the reason why M1 hasn't been a year-over-year upgraded chip.
- Intel competes with Apple pretty hard on 10nm already. Alder Lake has much better single-core performance, and will beat pretty much everything except the Ultra in raw performance metrics. Keep in mind, that's with silicon that is 4x (!!!) larger than what Apple is working with. Once Intel is on the same node as Apple (estimated to be by the beginning of 2024), I don't really think there will be a contest anymore.
Having used the latest-and-greatest from both camps, I don't think anyone really has bad chips, per se. Like the other commentors suggested though, Apple and Intel don't really occupy the same product space anyways. Both will likely continue to be successful, the real question is "who's going to crack first?" For my money, the answer is Apple. Your mileage may vary.
The way I see it, they're mostly in different markets -- or rather, Intel is in more markets.
Apple makes processors for their own vertical lines of computing devices.
Intel makes processors for an entire market of computing devices with diverse OEMs.
Currently, I cannot buy an Apple M processor and slap it into a variety of needs-based form factors (server, desktop tower, embedded, etc). And even if I could, Linux and Windows still aren't usable on it.
Yeah, they are done. Worked there for 15+ years in the CPU design org. The total collapse of the technical cadre in the 2010s will be studied in business schools as a case study of how to destroy a tech company from within even as they faced no competition. Intel has a dire competence problem that is impossible to address since the vast majority of management are lifers who are only interested in preserving the only system they know and not even the CEO can take on such an entrenched bureaucracy.
Any talk of Intel beating Apple/AMD in "performance" while being behind in process technology is ignorant of the engineering metrics that really matter in the industry, as well as the actual impact of process shrinks in the year 2022.
12 comments
[ 2.9 ms ] story [ 32.7 ms ] threadIntel needs to go into multiple kinds of motherboards and work with billions of combinations of other hardware.
For Apple building M1 is a much more integrated approach which is why they can produce better performance. If they built a similarly general purpose chip as intel they would likely not produce something as good as M1 or current Intel.
M1 is less than 5% of than the total chip market, intel are obviously not doomed, intel/amd/arm is powering like the other 95% including billions of servers etc. that M1 will never compete in.
No consumer buying intel is suddenly going to not buy intel because it isn’t on par with M1, unless they were buying a mac anyway.
I think Apple is in a great position to ride this wave as they are already surfing it. Even down to the languages they use avoiding GC which reduces memory requirements which in turn produces better battery life.
I can't predict the negative outcomes for intel.
Citation?
Consumer battery technology isn't advancing at pace with increased compute.
What is out there that requires faster CPUs now? AR? Does TikTok use more CPU power than I thought or something?
Anyway, M1 is only competitive in the market for processors for Apple computers. Unless Apple makes a enormous change to their business practices, I don't see them selling their CPUs to others. Apple doesn't have a good history of corporate support that would get enterprise buyers to buy their servers, if they had servers to sell, etc. Look how hard of a time AMD has selling server CPUs, at times where they've got a clearly superior CPU (first few years of opterons, currently with Epycs); it takes more than just a superior CPU to sell server CPUs.
- ARM is still a second-class architecture in the world of software. I work with Mac packaging every day, and M1 is still pretty poorly-supported, all things considered.
- Apple has pretty much no way to increase their performance besides horizontal scaling or increasing their silicon density. That puts them in an awkward spot, and it's probably the reason why M1 hasn't been a year-over-year upgraded chip.
- Intel competes with Apple pretty hard on 10nm already. Alder Lake has much better single-core performance, and will beat pretty much everything except the Ultra in raw performance metrics. Keep in mind, that's with silicon that is 4x (!!!) larger than what Apple is working with. Once Intel is on the same node as Apple (estimated to be by the beginning of 2024), I don't really think there will be a contest anymore.
Having used the latest-and-greatest from both camps, I don't think anyone really has bad chips, per se. Like the other commentors suggested though, Apple and Intel don't really occupy the same product space anyways. Both will likely continue to be successful, the real question is "who's going to crack first?" For my money, the answer is Apple. Your mileage may vary.
Apple makes processors for their own vertical lines of computing devices.
Intel makes processors for an entire market of computing devices with diverse OEMs.
Currently, I cannot buy an Apple M processor and slap it into a variety of needs-based form factors (server, desktop tower, embedded, etc). And even if I could, Linux and Windows still aren't usable on it.
Any talk of Intel beating Apple/AMD in "performance" while being behind in process technology is ignorant of the engineering metrics that really matter in the industry, as well as the actual impact of process shrinks in the year 2022.