アドリアン (@adev) 's Twitter Profile
アドリアン

@adev

Engineer | 💬 Engineering, Software, Sciences, History | 💼 : @CERN, @EPFL, @BlueBrainPjt, @Woven_Toyota, @SKAO | ❤️: Happy husband and father of two

ID: 97059047

linkhttps://bsky.app/profile/adevress.bsky.social calendar_today15-12-2009 20:48:51

4,4K Tweet

206 Followers

243 Following

アドリアン (@adev) 's Twitter Profile Photo

The French abolished the death penalty a while ago but I am pretty sure they would still be in favor of using it on people that call this abomination a "croissant".

The French abolished the death penalty a while ago but I am pretty sure they would still be in favor of using  it on people that call this abomination a "croissant".
アドリアン (@adev) 's Twitter Profile Photo

This asshole is defending someone who was calling to "burn the house" of random immigrants. And he is sufficiently stupid to call it "Freedom of speech"

アドリアン (@adev) 's Twitter Profile Photo

I am always amazed to the amount of effort Google can deploy to make the average user *NOT* to use their product. Is asking for too much to ask your *own* SSO to give a basic explanation whatsoever when it fails ? This is the UX equivalent of "Fuck off".

I am always amazed to the amount of effort <a href="/Google/">Google</a> can deploy to make the average user *NOT* to use their product.

Is asking for too much to ask your *own* SSO to give a basic explanation whatsoever when it fails ?

This is the UX equivalent of "Fuck off".
アドリアン (@adev) 's Twitter Profile Photo

pretty good post on I/O optimizations. bitflux.ai/blog/memory-is… It is pretty common to see applications doing double prefetching ( DFS -> nvme, nvme -> RAM ) when it is often counter-productive for most I/I patterns.

アドリアン (@adev) 's Twitter Profile Photo

Curious of what justify the investment. Hardware people: Is there any potential application of LLM in Lithography that I fail to see (or this is just speculation) ?

アドリアン (@adev) 's Twitter Profile Photo

"sort of" OpenShmem never lift off as a programming paradigm. It's 2025 and HPC is still heavily using MPI because PGAS is such giant pain in the ass to work with. GPUs are a different beast due to SVM, but the long term trend will likely be the same.

アドリアン (@adev) 's Twitter Profile Photo

Pleased to see that Cloudflare supports Ladybird . 1- That would probably help Ladybird to better pass the usual Cloudflare bot detection system 2- In a world where Web tracking and profiling are omnipresent, we need more than ever a good and neutral Web Browser

アドリアン (@adev) 's Twitter Profile Photo

Sorry for my French, but this is fucking amazing. Beating (fairly) an optimized BLAS implementation is already not a small feat. But doing it with a brand new language, a new compiler and on a (undocumented) proprietary accelerator is pretty insane. Kudos to the Mojo team

アドリアン (@adev) 's Twitter Profile Photo

This post makes me depressed 👇 news.ycombinator.com/item?id=453588… I do not understand how anybody can still be pro-dynamic typing in programming language in 2025. We had 40y of software history showing us it leads to in maintainable piece of junk. Just stop here guys.

アドリアン (@adev) 's Twitter Profile Photo

Or just don't use W*ndows. 10y ago, using Linux Desktop was mainly militantism. Today, it is just the logical choice. Because BigTech™ has shown again and again it can not be trusted with your data.