@AgileJebrim @valigo Sure, but even in real time systems you have to prototype your algorithms somewhere. In radar signal processing that often is python, because why not.
English
Michael F.
1.4K posts
Michael F.
@Michi07f
Mechatronik / Artificial Intelligence JKU Linz
Присоединился Mart 2010
619 Подписки86 Подписчики
If you still don't hate recruiters enough, or if you are still not convinced that hr is the most incompetent, harmful, and entitled industry in tech, just watch this video.
But TLDR is "I have 20 seconds to find 5 keywords in your CV, but nobody taught me how to use Ctrl+F"
English
@AgileJebrim @valigo Python and Docker is just basic computer literacy for any engineer.
English
@ErbunnNinja @yacineMTB @oscpmentor You have tried Ubuntu and then gave up, haven't you?
Plasma is pretty great. It even comes with a volume mixer, try that on macOS.
English
@yacineMTB @oscpmentor Yea, it’s literally Linux with the best ui.
You won’t find better in Linux, not without hours of tweaking and retweaking.
Linux is great, but macOS is better for anyone who values their time.
English
@PatrickToulme I want this so bad for sass. I just want to emit a mufu.sin without a preceding fmul.ftz 1/(2π).
English
Launching pyptx — a Python DSL for writing NVIDIA PTX kernels.
One PTX instruction = one Python call. Write pure PTX in Python.
Direct Hopper + Blackwell support: wgmma, TMA, tcgen05, mbarriers. JAX + PyTorch integration.
Includes GEMM, grouped GEMM, RMSNorm, SwiGLU, and a PTX→Python transpiler
pip install pyptx[torch]
pip install pyptx[jax]
github.com/patrick-toulme…
English
@F_Vaggi @cHHillee @tenderizzation I assume batch invariant means it only works for a certain batch size/layout. Don't know though.
English
@cHHillee @tenderizzation I assumed it meant that the gradients of the ith element are independent of the other elements in the batch but I’ve not heard that nomenclature before.
English
Quick question: how many people are coding for Qualcomm's Hexagon? I was leafing through the V79 HVX Programmer Reference Manual and damn, there are some fun instructions in there.
"Proprietary ISA from one vendor, who cares" vs "Cool as hell ISA on tons of devices":🤔
English
@jsuarez @PoratEitan CUDA C is not really a thing. NVCC has always been a C++ compiler.
English
@josethevrtech What even is apple pricing these days? They are arguably the best value in many categories (macbook neo, mac mini).
English
If VALVE gives out and supports SteamOS for free on custom and homemade PCs, then in my opinion it would be fair for Valve to charge “Apple pricing” for its own hardware. Since there would be no difference other than “made by VALVE”.
The 4Chan Archives@blacknredtext
The important question
English
@mikeryko1 @i_am_fabs Ja, stimmt. Bei 1by ist es wirklich nicht mehr viel Arbeit.
Deutsch
@Michi07f @i_am_fabs Wenn man schon 1x12s GRX fährt, muss man "nur" Schaltwerk und STIs tauschen oder? Gibt die ja jetzt als full wireless.
Deutsch
Man kann bei Shimano GRX Schaltungen easy Di2 nachrüsten?????? WARUM HAT MIR DAS NIEMAND GESAGT.
Deutsch
@L_33_H Angry birds and that design? Are they still stuck watching the 2010 ipad reveal keynote?
English
Is it reasonable to consider that since the HBM3 memory of a H100 has a bandwidth of ~3Tb/s and the chip can do ~900TFlops, a rule of thumb is that every bfloat16 should be reused ~600 times?
English
@See_Elegance @Andrei_Tarkhov I guess you have to ask the clankers better questions.
English
@Andrei_Tarkhov Couldn't you just publish your 100-pager as a blog post with the same results? Blogs are also more likely to show up in search results, I don't remember last time chatgpt cited someone's thesis.
English
A novel argument to do a PhD in 2026 is to expand the training set of AI models by a unique 100-pager. Before, only a few experts in the world would read it — now, every single model can do so & benefit from reusing it in unexpected contexts. It all happened so fast…
English
$200/month is enough to buy an H100 GPU for 6 hours every workday
English
@motihari1903go Wie macht man täglich 15.000 Schritte, ist das nicht einfach extrem zeitintensiv, wenn's nicht während der Arbeit automatisch passiert?
Deutsch
Bin recht fit. Ex-Leistungssportler, täglich ca 15.000 Schritte, 6x/Woche Krafttraining, dazu Ballsport, Wandern...
Aber: Ich kenne 70-jährige, da muss ich die Beine in die Hand nehmen, wenn wir gemeinsam einen Berg raufgehen.
Das ist ein Lebensziel: Fit und aktiv altern.
Deutsch
@kalomaze People did similar stuff before, it got patched and the world moved on.
en.wikipedia.org/wiki/Jailbreak…
English
the claude mythos thing where it apparently found a way to get full kernel access via execution of normal javascript on an ordinary web page. dear God
English
@lisyarus You can rotate it relative to the rest of the spacecraft. Problems may arise if the moment of inertia of your camera is about as large as the moment of inertia of the spacecraft itself.
English
It just occurred to me: you can't rotate a camera on a spacecraft, because that will alter the angular momentum. You can if you have two cameras, though, if their rotation cancels out 😎
(I guess in reality this is solved by RCS/SAS/etc)
English
@CanuckCreator @ccatlett1984 That's pretty impressive. Have you removed all comments from the gcode just to make sure it doesn't say "sliced from benchy.stl by orcaslicer x.y on DDMMYYYY"?
English
@ccatlett1984 so we did also run it through Claude Opus 4.6 and it WAS able to determine it was a benchy
After 58s of computer on their servers
i fear how long an RPI or equivalent would take to sort that out....
English
When asked, ChatGPT says it can detect a boat in gcode, and when given the gcode for a sliced benchy, it uhh, thinks its NOT a boat. and when asked what the gcode is, it thinks its a wedge
and this is full running on OpenAI servers ChatGPT, wtf is a rpi based detection gonna do?
Nero3D the Canuck Creator@CanuckCreator
Govt folks wanting printers to rat on you for printing gun stuff. Bruh spaghetti detection barely works as is. You think printers are smart enough to know what a lower receiver is.
English