Posted: Mar 03, 2021 1:23 pm
Several of the items shown in that video don't surprise me.
Back in the old 1990s Archimedes days, Acorn demonstrated an emulator for the 80286 running on their version of the ARM, that actually executed faster than a native 80286 chip. You can guess how Intel reacted to that.
The ARM has always been capable of unleashing tricks of this sort onto an unsuspecting audience, and in the hands of skilled developers who know how to take advantage of its unique architecture, it can embarrass competitors with higher clock speeds while hardly drawing breath.
The unique architecture of the ARM lends itself to some interesting code optimisations, that either a skilled assembly language developer or a suitably written optimising compiler can make use of to devastating effect, if of course you exercise a little savvy deploying the requisite options. Like any CPU, it will be hampered by incompetent coding, but if you exercise diligence learning about its features, you can utilise them to make your code run at warp speed. Judicious use of both conditional execution in non-branch instructions, and proper consideration of where to bring the barrel shifter into play, lets you pull off tricks that can easily double or triple the execution speed of your code compared to code using more traditional instruction sets.
By the way, if you want to sample the fun that the ARM architecture offers, albeit at slower speeds, but without the expense of buying a brand new M1 Mac, you can always try out this ARM emulator written in JavaScript, and see its features in operation for yourself.
Back in the old 1990s Archimedes days, Acorn demonstrated an emulator for the 80286 running on their version of the ARM, that actually executed faster than a native 80286 chip. You can guess how Intel reacted to that.
The ARM has always been capable of unleashing tricks of this sort onto an unsuspecting audience, and in the hands of skilled developers who know how to take advantage of its unique architecture, it can embarrass competitors with higher clock speeds while hardly drawing breath.
The unique architecture of the ARM lends itself to some interesting code optimisations, that either a skilled assembly language developer or a suitably written optimising compiler can make use of to devastating effect, if of course you exercise a little savvy deploying the requisite options. Like any CPU, it will be hampered by incompetent coding, but if you exercise diligence learning about its features, you can utilise them to make your code run at warp speed. Judicious use of both conditional execution in non-branch instructions, and proper consideration of where to bring the barrel shifter into play, lets you pull off tricks that can easily double or triple the execution speed of your code compared to code using more traditional instruction sets.
By the way, if you want to sample the fun that the ARM architecture offers, albeit at slower speeds, but without the expense of buying a brand new M1 Mac, you can always try out this ARM emulator written in JavaScript, and see its features in operation for yourself.