←back to thread

Apple's Shifting Differentiation

(stratechery.com)
292 points kaboro | 1 comments | 11 Nov 20 13:21 UTC | HN request time: 0.209s | source
Show context
jeswin ◴[11 Nov 20 15:56 UTC] No.25059378[source]▶
People should stop cherry picking benchmarks (like SPECint2006) when comparing CPUs. And many seem to be misreading Anandtech's recent comparison as - Wow look M1 is almost as fast as a 5950X. Except that it's simply impossible.

Wait for the actual benchmarks to come out. Especially real world workloads. I'll be very surprised if desktop Ryzens (or even the Intel CPUs) don't run circles around the M1 in a laptop.

replies(2): >>25059553 #>>25059557 #
1. bufo ◴[11 Nov 20 16:15 UTC] No.25059553[source]▶
People have been saying that for years and keep getting it wrong. There is no evidence whatsoever that current x86-64 chips will do better. The polar opposite of SPECint2006 is stuff like Speedometer 2.0, where Apple chips still crush it https://twitter.com/dhh/status/1318644849437003780.

The only previous valid point was that previous iPhones/iPad couldn't sustain these high clock rates for long since they were fanless, fortunately adding a fan is not very hard.

Sure the 5950X will probably win the multi-core benchmarks since it has 2x cores but it does not look promising for single core speed.