Top all-time · #7898
Andi Kleen
Personal blog · halobates.de/blogsystemsprogramminghardware
Andi Kleen's personal blog at halobates.de covers systems, programming and hardware. Since 2012, 10 different Hacker News readers have linked it 10 times, most often “Mental models for modern program tuning [pdf]”. It is still writing: the last post was on 2026-09-19, and you can follow it by feed.
Too few sampled comments (5) to say how often HN recommends it.?
Why HN links it
HN commenters link Andi Kleen's blog posts on ECC memory error patterns in access logs, batching locks in concurrency, and lock elision in Linux. The links are cited as sources, without judging the writing.
Summarised by Claude from a sample of 5 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Most-linked posts?
- Mental models for modern program tuning [pdf]2 people · 2 mentions · first linked 2016-06-06
- Literature survey for practical TSX lock elision applications2 people · 2 mentions · first linked 2016-05-30
- The weekend error anomaly2 people · 2 mentions · first linked 2012-04-21
What Hacker News says?
I maintain a tool that reports errors on Linux systems with ECC memory, and also a website describing the tool ( http://mcelog.org ). I wrote about patterns in the access logs correlated to time some time ago in my blog: http://halobates.de/blog/p/181
Comment linking /blog/p/181
on “For a brief period, the Windows kernel tried to deal with gamma rays”Submitted Quitting an Intel x86 Hypervisor as a story · 121 points
HN discussionBoth static and dynamic histograms can be pretty important, actually. Dynamic ones for performance, static ones -- usually for correctness (imagine developing a tool just like QEMU, which needs to emulate each instruction type). Performance counters by themselves aren't granular enough for an exact histogram. But you can use them (especially the LBR[1] and the fancy new PT[2]) to reconstruct an approximate control-f…
Comment linking /pt-tracing-summit15.pdf
on “Does a compiler use all x86 instructions? (2010)”Submitted Mental models for modern program tuning [pdf] as a story · 3 points
HN discussion>essentially no cost unless the lock is contested. Atomics are not free. And fetching a cache line can be very expensive, even without contention. However lock less algorithms are often worse because they have even more atomics and even more cache line transfers. And they're really hard to get right. (iirc there was a statistic somewhere about the average number of corrections in lock less papers -- it wasn't pretty…
Comment linking /applicative-mental-models.pdf
on “An Introduction to Lockless Algorithms”Submitted Literature survey for practical TSX lock elision applications as a story · 3 points
HN discussionIntel employees added it to glibc. I think Intel wanted to show everyone "don't be scared, silly, you're already using it!" https://lwn.net/Articles/534761/ http://halobates.de/adding-lock-elision-to-linux.pdf
Comment linking /adding-lock-elision-to-linux.pdf
on “Intel to disable TSX by default on more CPUs with new microcode”Submitted The weekend error anomaly as a story · 2 points
HN discussionOriginal quote from the author: «batch your locks. Don’t make critical sections too small. Having the smallest locks is not cool anymore.» http://halobates.de/blog/p/231 In the whitepaper: «The longer lock region has potentially longer block times, but minimizes the communication overhead of transferring the lock too frequently. This is a simple case with only two threads. With more cores in each socket, the queuein…
Comment linking /blog/p/231
on “Concurrency: "having the smallest locks is not cool anymore"”Submitted Literature survey for practical TSX lock elision applications as a story · 1 points
HN discussion
