François Farquet

Here is an awesome dashboard showing how GraalVM JIT performance evolved over the last year. With details of all individual performance changes: graal.d3s.mff.cuni.cz/see/overview This is work by the performance experts of Charles University in Prague in Petr Tuma's group. 👇 1/10

Please help spread the word about our @graalvm 25 release 🚀🏆!

@axelfontaine @Devoxx Do not worry, @GraalVM stays around and will continue to make programs run faster and more efficient.

@marclou Nice! You would be a strong freediver!

I had my lowest resting heart rate last night. AC blasting at 18 degree, a strict sleep schedule, and a proper wind down routine.

We just merged the current status of the upcoming JDWP support for @GraalVM Native Image! 🥳 This will soon provide developers with the same debugging experience they are used to in Java, but for native images! Stay tuned for more details. github.com/oracle/graal/p…

Nice! Now compatible with latest JDK again

@TintinFournier C’est article n’a aucun sens. Merci pour la perle.

A deux doigts d'inventer les balades en montagne. blick.ch/fr/well/du-spo…

.@GraalVM-based Native Kafka Broker looks amazing!🚀🔥 Charts and observations from cwiki.apache.org/confluence/dis…: — avg startup time is ~130ms; — No CPU spike is observed for GraalVM; the JVM broker had sudden spikes when the broker started and also when the load testing started. — for JVM broker the max memory usage went up to ~1GB. For GraalVM is was ~500MB with G1GC and ~250MB with serial GC. #GraalVM #Kafka #performance

@florian_tramer Right. The problem here is that the marketing department implies that recovering non-existing data is possible. To be fair, all phone photos are already greatly altered by software (combined images, low pass filters, color correction, etc). Are they still real?

Calling a diffusion tool a "zoom" is super misleading. This doesn't "zoom in". It just outputs plausible pixels that may not match reality (even in this demo it mistakes "Asis" for "Asia"...) Wait till people "enhance" blurry license plates or faces and get back hallucinated crap

@alina_yurenko @jbaruch @gradle Looks nice! Almost as nice as the desk I’m using today. Right, @alina_yurenko ?

am I in Hollywood :))) @jbaruch @gradle

@Sharat_Chander @jcon_conference @alina_yurenko @alina_yurenko: ✌️💁🏼‍♀️ @fniephaus: “I don’t know this girl”

Day 1 I have a (back seat) #Java driver at ⁦@jcon_conference! Hi, ⁦@alina_yurenko⁩! 💜🙏🏽

It's finally out! Teaser: Linearly scalable and sound, yet precise, static analysis for Java. Spoiler: Analyzing @springframework PetClinic 5x faster! Link to PDF: codrutstancu.com/pldi24main-p25… #PLDI2024

@alina_yurenko Yeah. My dad was just visionary in its way of using Google Search.

it's funny how chagpt brings us back to asking questions and maintaining a dialogue in complete sentences, rather than throwing keywords at google :))

@MathieuAvanzi Parce qu’en Valais, on regarde la télé le soir 📺, mais on prend le télè la journée ! 🚠

SUISSE ROMANDE : Comment prononcez-vous le mot #télé ? En zone claire : tèlè (avec des voyelles bien ouvertes). Seul le Valais fait de la résistance !

@francoisfleuret Now we all want to know what moof is! Is the goal to have it say ‘spoon’? (But I guess it will now since X is likely in the training set)

"These LLMs are nothing like the human brain. They need zillions of examples while the human brain can deal with a few."

This is nous-capybara-34b.Q4_K_M.gguf , I am amazed. That's so perfect that I wonder if my "standard tests" did not end up in the training set. @NousResearch ???

What would you like to see in GraalVM? Let us know by sharing your feedback in our community survey!📝 It takes less than 5 minutes and we will use your feedback to plan future releases. Share now: graalvm.typeform.com/survey

@vinceflibustier Même expérience à travers le monde en tant que suisse!

Alors, en fait, à mon sens, la planète entière déteste les Français. Quand tu vas au Canada, au Mexique, aux USA, dans les pays de l'est, on te regarde super mal jusqu'à ce que tu dises "I'm Belgian" et là, grand sourire.

@Sharat_Chander @Oracle_NL Utrecht office is beautiful!

Hello ⁦@Oracle_NL⁩!

Recently we visited @ICepfl again to meet with students and talk about the research projects we do at @OracleLabs Switzerland. Thank you all for coming! You can find information about our projects and open positions at labs.oracle.com/pls/apex/r/lab… #internship #research

@TintinFournier 5/8 sont avocats ou notaires. Belle diversité là également :)

Pendant ce temps là ...

@TintinFournier @LuisierD @suissepad @plrvs @mariannemaret Avec 1000-2000 voix d’écarts, peut-être que ces éléments auraient pu jouer. Mais pas avec 18’000 d’écarts avec un candidat qui régresse par rapport à 2019 face à une sortante en grande progression et qui le bat dans tout le canton. C’est un peu masochiste…

@LuisierD @suissepad @plrvs @mariannemaret Probablement. Je le pense aussi. Mais on peut aussi imaginer que les UDC pourraient voter le ticket Rieder/Nantermod et que certains à gauche voudront "briser le monopole PDC/Centre" en s'abstenant, voire en votant Nantermod.

Je mets ça là pour le @plrvs Résultats du Conseil des États par région. @mariannemaret première dans le Bas-Valais et le Valais central. 2ème dans le Haut-Valais. Peut-être le moment pour reconnaître qu‘un 2ème tout est inutile.