Another AVX vectorization workshop. Price: $799 or €749 More info bellow...

More info: johnnysswlab.com/avx-neon-vecto… Register at: [email protected]

Do modern allocators (like jemalloc, tcmalloc, mimalloc or simple libc malloc) have metadata for each chunk of memory, or do they use other mechanisms to figure out the properties of each chunk (e.g. if address is between 0x740000 and 0x750000, the size of chunk is 32 bytes)?

My favorite combo: sw and lotr

Today I found the new edition of "Performance analysis and tuning on modern CPUs" in my post box. I enjoyed the first edition and I am looking forward to what the new edition brings. Thank you Denis Denis Bakhvalov !

Last chance to register for AVX vectorization workshop! More info: johnnysswlab.com/avx-neon-vecto… Register: [email protected]

An onsite NEON workshop will take place as a part of NDC Conferences in Kongsberg, Norway, on September 22nd and 23rd. If Norway is too far away, the same workshop is available online from the comfort of your home or office. More info: johnnysswlab.com/avx-neon-vecto…

Onsite NEON workshop will be held in Norway. I would be very happy to see some of my twitter companions there and meet live :)

Does it matter if we are compiling with optimizations off (O0) or optimizations on (O3) if the problem is memory bound? Let’s find out… #optimizations #performance #instructionlevelparallelism #ilp #compiler #gcc #memorybound johnnysswlab.com/an-optimizing-…

Tutorial: Building the Simplest Possible Linux System This tutorial walks you through building and booting the simplest possible Linux system, first under QEMU and then on real hardware.

A debugging method where you put an error at the source to check if the error propagates at the destination is the most underestimated debugging technique ever!

Serbian language has this verb "drlya" which means "to try out stuff until it works" and embedded engineers really use it a lot.

A new post. I asked ChatGPT to rewrite it in conversational and provocative tone, it came out awesome, but I didn't dare publishing it like that. Read the rewrite: chatgpt.com/s/t_68683f811d…

Johnny's Software Lab is coming to #CppCon #avx #performance #workshop #datacenters #embedded Find out more: cppcon.org/class-2025-avx/

C++ is a box full of surprises. Just found out there is std::midpoint(a,b) that calculates (a+b)/2 but without overflow.

#AVX #Vectorization Workshop is at #CppCon. Register now! cppcon.org/class-2025-avx/