As some here may be aware, Apple is in the process of migrating its Mac lineup from Intel and x86/CISC to its own custom silicon and ARMv9/RISC. I had the opportunity late last year to acquire one of the new M1 Macs. This is a fan less MacBook Air with 8GB of RAM and a 256 GB SSD. So, what is it like?
First things first, this thing is ridiculously fast and responsive. The benchmarks out there showing M1 blowing everything but the most expensive gaming class CPUs out of the water (and with those it is standing tow to toe) are turning out to be right. Applications literally blink open and closed. Everything just screams - and this is a FANLESS notebook. And it doesn’t even get super warm or throttle down. As to compatibility between Intel and M1 compiled software it comes with a translator built in (Rosetta2) and Apple even took advantage of the SOC architecture here (more on that later) to build in a block where typically hard to translate x86 code gets a special co-processor.
So, how is it doing this? Not only does it blow all x86 CPUs away it runs rings around all the ARM SOCs. The answer here is Apple’s in house designed microarchitecture (Which has strange resemblances to Intel’s Core2 architecture... ). In simple terms, Apple’s in house microarchitecture (called A series on iPhones and lower end iPads and M on Macs and the iPad Pro) processes at least double the number of instructions per clock cycle than anyone else. The Pipeline is 8 wide which is insanely huge - the closest is AMDs top end Ryzen at 4 wide. This enables the apple SOCs to do a lot more work at a much lower clock speed so they are extremely efficient.
And M1 is also doing something new in that this is a laptop using an SOC and even better - one will a full Unified Memory Model and with the RAM actually on die which reduces the RAM lag to virtually zero. On the downside you can’t user upgrade the RAM, on the upside the RAM speed is ridiculously high and even the SSD while a separate module is positioned right next to the SOC to again reduce distance and increase performance. The SOC also contains an 8 core GPU and a series of other blocks with things like Machine Learning, hardware based security and other modules.
Right now M1 has 8 CPU Cores (4 Performance and 4 Efficiency - the mixed core types is another first for a Personal Computer), 8 GPU cores and 16 Machine Learning cores plus other blocks. In a few weeks the next iteration is expected to change that to 8 Performance Cores and 16 GPU Cores while the rest stay stable. This will be interesting to follow as it represents in some ways the hardware future for Personal Computing.
First things first, this thing is ridiculously fast and responsive. The benchmarks out there showing M1 blowing everything but the most expensive gaming class CPUs out of the water (and with those it is standing tow to toe) are turning out to be right. Applications literally blink open and closed. Everything just screams - and this is a FANLESS notebook. And it doesn’t even get super warm or throttle down. As to compatibility between Intel and M1 compiled software it comes with a translator built in (Rosetta2) and Apple even took advantage of the SOC architecture here (more on that later) to build in a block where typically hard to translate x86 code gets a special co-processor.
So, how is it doing this? Not only does it blow all x86 CPUs away it runs rings around all the ARM SOCs. The answer here is Apple’s in house designed microarchitecture (Which has strange resemblances to Intel’s Core2 architecture... ). In simple terms, Apple’s in house microarchitecture (called A series on iPhones and lower end iPads and M on Macs and the iPad Pro) processes at least double the number of instructions per clock cycle than anyone else. The Pipeline is 8 wide which is insanely huge - the closest is AMDs top end Ryzen at 4 wide. This enables the apple SOCs to do a lot more work at a much lower clock speed so they are extremely efficient.
And M1 is also doing something new in that this is a laptop using an SOC and even better - one will a full Unified Memory Model and with the RAM actually on die which reduces the RAM lag to virtually zero. On the downside you can’t user upgrade the RAM, on the upside the RAM speed is ridiculously high and even the SSD while a separate module is positioned right next to the SOC to again reduce distance and increase performance. The SOC also contains an 8 core GPU and a series of other blocks with things like Machine Learning, hardware based security and other modules.
Right now M1 has 8 CPU Cores (4 Performance and 4 Efficiency - the mixed core types is another first for a Personal Computer), 8 GPU cores and 16 Machine Learning cores plus other blocks. In a few weeks the next iteration is expected to change that to 8 Performance Cores and 16 GPU Cores while the rest stay stable. This will be interesting to follow as it represents in some ways the hardware future for Personal Computing.
Last edited: