Top all-time · #10291
Alois Kraus
Personal blog · aloiskraus.wordpress.comprogramminghardwaresystems
Alois Kraus's personal blog at aloiskraus.wordpress.com covers programming, hardware and systems. Since 2017, 7 different Hacker News readers have linked it 7 times, most often “Why Skylake CPUs Are Sometimes 50% Slower”. It is still writing: the last post was on 2026-03-16, and you can follow it by feed.
Too few sampled comments (3) to say how often HN recommends it.?
Why HN links it
HN commenters link Alois Kraus's posts as a source on how the x86 pause instruction's cycle cost varies widely between CPUs, including Skylake-SP raising it to as many as 140 cycles.
Summarised by Claude from a sample of 3 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Performance is everything. But correctness comes first.
Most-linked posts?
- Why Skylake CPUs Are Sometimes 50% Slower5 people · 5 mentions · first linked 2018-06-18
What Hacker News says?
> One problem with the x86 pause instruction is the number of cycles spent paused has varied dramatically from one CPU to another. Some CPU models ignore it. On others it is very short, a few or ten cycles, and on some it is hundreds of cycles. Here's a previous HN discussion about that (in particular Skylake CPUs greatly diverging from previous behaviour): https://news.ycombinator.com/item?id=17336853 Original link…
Comment linking /2018/06/16/why-skylakex-cpus-are-sometimes-50-slower-how-intel-has-broken-existing-code/
on “Improving Firefox Responsiveness on macOS”Submitted Why Skylake CPUs Are Sometimes 50% Slower as a story · 553 points
HN discussionIn skylake, they increased the duration of pause from "about 10 cycles" to "as many as 140 cycles". https://aloiskraus.wordpress.com/2018/06/16/why-skylakex-cpu...
Comment linking /2018/06/16/why-skylakex-cpus-are-sometimes-50-slower-how-intel-has-broken-existing-code/
on “A flexible, lightweight, spin-lock barrier”Submitted Why Skylake CPUs Are Sometimes 50% Slower (2018) as a story · 119 points
HN discussionYou could first measure the perf impact of the fence instruction and then subtract that out? But yeah I guess it may not matter much for quick and dirty calibration loop. I found somewhere ( https://aloiskraus.wordpress.com/2018/06/16/why-skylakex-cpu... ) that the pause instruction had this wild cycle difference between different CPU and it caused some grief, I had no idea. I stopped doing low level coding a while…
Comment linking /2018/06/16/why-skylakex-cpus-are-sometimes-50-slower-how-intel-has-broken-existing-code/
on “Spinning around: Please don't – Common problems with spin locks”Submitted Windows Task Manager Shows Misleading Values as a story · 4 points
HN discussionSubmitted The Definitive Serialization Performance Guide as a story · 2 points
HN discussion