←back to thread

283 points walterbell | 1 comments | | HN request time: 0s | source
Show context
darkamaul ◴[] No.45769289[source]
Better (or simply more) ARM processors, no matter who makes them, are a win. They tend to be far more power-efficient, and with performance-per-watt improving each generation, pushing for wider ARM adoption is a practical step toward lowering overall energy consumption.
replies(5): >>45769421 #>>45769508 #>>45769815 #>>45769973 #>>45772372 #
ahoka ◴[] No.45769508[source]
Are ARM processors inherently power efficient? I doubt.

Performance per watt is increasing due to the lithography.

Also, Devon’s paradox.

replies(5): >>45769580 #>>45770046 #>>45770800 #>>45773990 #>>45779586 #
jorvi ◴[] No.45770046[source]
They aren't inherently power efficient because of technical reasons, but because of design culture reasons.

Traditionally x86 has been built powerful and power hungry and then designers scaled the chips down whereas it's the opposite for ARM.

For whatever reason, this also makes it possible to get much bigger YoY performance gains in ARM. The Apple M4 is a mature design[0] and yet a year later the M5 is CPU +15% GPU +30% memory bandwidth +28%.

The Snapdragon Elite X series is showing a similar trajectory.

So Jim Keller ended up being wrong that ISA doesn't matter. Its just that it's the people in the ISA that matter, not the silicon.

[0] its design traces all the way back to the A12 from 2018, and in some fundamental ways even to the A10 from 2016.

replies(5): >>45770152 #>>45770164 #>>45771322 #>>45771561 #>>45780873 #
znpy ◴[] No.45771322[source]
You’re conveniently skipping the part where x86 can run software from 40 years ago but arm can drop entire instruction sets no problem (eg: jazelle).

Had been arm so weighted by backwards compatibility i doubt it would be so good as it is.

I really think intel/amd should draw a line somewhere around late 2000 and drop compatibility with stuff that slow down their processors.

replies(2): >>45771540 #>>45779448 #
le-mark ◴[] No.45771540[source]
> jazelle

That’s a blast from the past; native Java bytecode! Did anyone actually use that? Some J2ME phones maybe? Is there a more relevant example?

replies(1): >>45781279 #
1. hmry ◴[] No.45781279[source]
> Did anyone actually use that?

AFAIK you needed to pay a license fee to write programs using Jazelle instructions (so you needed to weigh whether the speedup of Jazelle was cheaper than just buying a more powerful CPU), and the instruction set itself was also secret, requiring an NDA to get any documentation (so no open source software could use it, and no open toolchains supported it).

I remember being very disappointed when I found out about that