New hardwares are now optimized for vectors- The details are explained in part 2

It takes several hundred lines of code but now we can source the data from my SSDs faster than the copy from the page cache in memory.

Because the reading strategy is not mmap but something custom. The author also uses 6 CPU cores for reading and one core for compute. Note that data in a buffer at 13 GiB/s RAM was faster than 6 GiB/s disk array.

The mmap() call presents the process with a buffer that is a blank slate even when the data is already in memory. The buffer is populated page by page as it's accessed from the page cache. This isn't a copy, it's just the operating system mapping the cached memory into the process. This costs more than it might seem. The worst case with mmap() the counting has to pause at every 4KB page boundary while the kernel processes a fault, tracks down the page of data in the page cache, then updates the page table of the process to insert the memory into the process. Fundamentally this is a process that is limited by the memory latency, not the CPU speed or memory bandwidth. With the potential for TLB walks and searching lists that track the page cache, we’re taking potentially dozens of CPU cache misses and several microseconds of waiting on memory for every 4KB page.

On theory

On a proper modern server the CPUs will let you do IO directly to the L3 cache, bypassing memory altogether. Because PCIe bandwidth is higher than memory bandwidth, on paper we could even get more max bandwidth than we can get from memory if we carefully pin the buffers into the CPU cache.
and for the memory:
This means to take advantage of memory performance you have to actually treat it more like a disk and less like Random Access Memory.

A try is made in part 3

Microservices add CPU and network costs. The video gives some number.

Vendre et gérer des certificats SSL n'est plus rentable économiquement car Let's Encrypt fournit déjà le service gratuitement à beaucoup de sites web.

Les exigences pour une plus grande sécurité que Let's Encrypt sont rares (et cher).

A Wikipedia customized for the device location

This one is original and carefully crafted.

After the two exploits of this week, another one is published.

Le fichier n'est plus respecté comme source de vérité des données. Les applications utilisent des formats binaires.

L'IoT est cadenassé

L'IPv6 est une raison pour que l'IoT ne respecte plus le routeur.

Les applications requiert HTTPS obligatoirement, même en local sur un Intranet.

Tout passe par l'IP (Tailscale, DoH) et bypass le routeur.

Chrome devient le nouveau IE, les applications affichent "fonctionne mieux avec Google Chrome".

Le DMA renforce les positions dominantes et les monopoles.

Aujourd’hui, on doit se débrouiller avec des «Oops», «Something went wrong», ou «Vérifiez votre connexion Internet».

Tout est dit.

The principal resource is capital and only capital.

Linus is saying the quiet part out loud: AI-native projects are environments where the key ingredient is capital. Abstraction can be performed by a quiet appreciation. True automation can be disruptive.

AI-based architectures (again, assuming that the AI works reliably, which is a highly contingent proposition) translate all that stuff into capital flows. A "machine speed" project requires money to participate in, regardless of their complexity.

An optimized version of MermaidJS

Une offre « gratuite » de Microsoft pour les assos, c'est un produit d'appel qui peut s'arrêter quand ça l'arrange.

La réflexion fait suite à l'association qui a perdu 600 Go de données (voir https://next.ink/261398/arret-dune-offre-gratuite-ms-365-pour-associations-watertrek-a-perdu-600-go-de-donnees/).

Watertrek, une petite association qui travaille sur la protection de l’eau, a perdu 600 Go alors qu'elle bénéficiait d'une offre de Microsoft pour les assos que l'entreprise a stoppée.