ταῦ (@taustw) header
ταῦ (@taustw) profile picture

EE | Prev. @ Stanford | Tricking rocks into thinking | ex-CERN | ex-🛰️ founder

Madrid, Spain
Joined November 2022
Photo by ταῦ (@taustw)Photo by ταῦ (@taustw)Photo by ταῦ (@taustw)Photo by ταῦ (@taustw)Photo by ταῦ (@taustw)Photo by ταῦ (@taustw)Photo by ταῦ (@taustw)Photo by ταῦ (@taustw)Photo by ταῦ (@taustw)Photo by ταῦ (@taustw)
ταῦ retweeted
I really do think EE is next in line after SWE Followed by Mech E I am not saying it's going to be fully automated without human oversight but it's probably going to be heavily assisted
ChatGPTproで入念に設計計算を詰めて、後はCodexでシミュレーションと基板設計をやってもらって、半日でここまで来た👀私はKiCADは全く触れてないけど、意外にも悪くない部品配置をしている👀
37
27
6
611
32,094
ταῦ retweeted
Data centers are our generation's version of nuclear power in that they seem to be supported by anybody with any brains and are exclusively hated by the most self-satisfied, uncurious retards alive.
362
2,247
172
25,688
655,770
Every time I think European voters cannot be more in denial about pensions, they prove me wrong: they are even more in denial than in my worst nightmare. The average French person believes there are 6.4 workers for every retiree. The real number is about 1.8.
Really interesting (and worrying) polling on French pensions The electorate estimates that for every pensioner we have 6 workers. In reality it's 1.8/1... Even in 1960s we weren't close to 6/1. Melenchon's supporters think its 13/1.
45
626
38
4,472
181,235
ταῦ retweeted
scoreboard
4
4
7
499
ταῦ retweeted
Wrote a new blog post on building a fully custom open source networking switch ASIC chip. talesonthewire.com/projects/…
7
21
225
5,563
I designed a radio chip with Claude in 7 days. Alone it would’ve taken months. But how do you use AI without skipping the learning? On Friday I’m asking Harald Pretl and Carsten Wulff, analog designers and professors experimenting with AI. What should I ask them?
37
45
1
513
16,423
"Embedded System Design" by Peter Marwedel An awesome, open access book link.springer.com/book/10.10…
2
69
2
759
27,622
We needed to build something that could go everywhere...
13
88
19
786
68,683
ταῦ retweeted
I miss basically everything about SF
26
2
1
336
23,895
ταῦ retweeted
Tohoku University and Toyota-supplier Aisin built a 16 nm AI chip (SoC) with 32 MB of embedded STT-MRAM that holds both a compact Linux boot image and Tiny YOLOv3 (object detection AI model) weights on-chip. No external DRAM or SSD is needed for startup or inference. It boots in 65.7 milliseconds (39× faster) and runs object detection at ~50.7 mJ/frame with big energy savings. They used weight sharing and hybrid 8/16-bit math to fit the model. The chip also includes on-chip SRAM (used for feature maps and other working data) and a DDR interface, so external memory can still be attached when larger models or more capacity are needed. TSMC manufactures the chip. Nice real-silicon demo of near-memory computing for edge AI — useful for automotive cameras, sensor fusion, and always-ready devices that need instant-on and low standby power. Not revolutionary: Others – Hailo, Ambarella, Renesas, etc. – already use large on-chip SRAM, embedded MRAM, or in-package DRAM, but it’s a good example of the gains when OS + model live in non-volatile on-chip memory. $TSM $AMBA #Hailo #Renesas #semiconductors
18
54
4
343
20,181
ταῦ retweeted
this guy figured out how to leak your exact location from any video with sound using the Doppler curves of nearby airplanes
588
1,896
648
34,981
3,722,626
Hydrogen-1.0 gives SOTA performance on FrontierScience-Research and on par to SOTA on GPQA-Diamond at an API cost similar to GLM-5.3. On our internal research scenarios, Hydrogen-1.0 scores higher than or on par with GPT-6-Astra at maximum thinking on scientific computing (76.9% vs 71.8%), forward reaction (90.5% vs 85.7%), and NMR structure elucidation (26.7% vs 30.0%). read the full blog at blog.deepprinciple.com/intro…
3
4
168
18,340
ταῦ retweeted
There are a number of computational primitives one can implement with oscillator-based computers, e.g. solving combinatorial optimization problems, and associative memory. Higher dimensional oscillators can also implement the attention, which forms the basis for modern LLMs
2
2
11
540
ταῦ retweeted
80 IQ: chips are beautiful and beautiful things have value 100 IQ: claudeslop about what layer the value will accrue in the AI era 120 IQ: chips are beautiful and beautiful things have value
1
4
43
1,774
ταῦ retweeted
I used Divyansh's approach to build an entire Nintendo into ~1GB of Qwen3.5 weights. Qwenputer, meet Nesformer: github.com/dbreunig/nesforme…
1/4 Meet Qwenputer: Qwen3.5-4B with a full RISC-V CPU inside its weights, using the model's own Gated DeltaNet layers as RAM. It runs C and MicroPython in its forward passes, no external runtime. On DeepMind math: base 55–90%, base + Python tool 95–100%, Qwenputer 97.5–100%.
29
64
26
742
43,806
A latching level shifter: with high dv/dt immunity on the output FET (M11). The circuit can be used in the output stage of a zero quiescent power dissipation (switches only on input edges) MOSFET gate driver and since it's latching, provides high dv/dt immunity.
5
52
1,107
Does it matter where you tap a current shunt? Let's simulate the 0.5 mΩ shunt moteus-x1 uses, with 10 A through it. Sense traces from the middle of the pads' inner edges read 10.1 A. 2 mm out, they read 14.1 A. A 4-terminal shunt reads right on every board I simulated. 🧵
8
15
1
209
9,025
ταῦ retweeted
Austria has been a Member State putting in doubt the whole process to get EU Inc done through some treaty interpretation. This type of comment is often considered a procedural approach to derail an upcoming law, like EU Inc. It's interesting because when contacted by some Austrians, the Ministry of Justice of Austria denied it.. now we see this.. 6/9
3
12
3
174
9,942