barelyOK
127 posts
Installed Fedora on my 16 year old laptop. Ran a c++ code with openMP to multiply two random 512 x 512 matrix. It took around 1.2 secs to execute.
This laptop has Intel core 2 duo cpu. Previously I ran Linux Mint with xfce on it. It was comparatively smoother compared to Fedora.
English
@VazeKshitij Clearly. Even 2% improvement in efficiency is significant in this sector.
English
The automotive sector is quite funny to me. The reason being the sheer size of the thing, and the many ways that it impacts us all, without the conscious realization of most.
For context, India’s automotive sector is one of the largest and most strategically important industries in the country, serving as a major driver of manufacturing, exports, employment, and technological development. Today, the industry is valued at roughly $250 billion and contributes around 7% to India’s GDP, while accounting for nearly half of the nation’s manufacturing GDP. From passenger vehicles and commercial trucks to two-wheelers, tractors, EVs, and auto-components, the scale of the ecosystem is enormous.
The sector also serves as one of India’s largest employment generators, supporting nearly 37 million jobs directly and indirectly across manufacturing plants, supply chains, logistics, software, electronics, servicing, and ancillary industries. India has emerged as one of the world’s largest automobile producers, while also becoming a major export hub for vehicles and automotive components.
Now, what’s particularly fascinating to me is that any minuscule innovation, if executed perfectly, can have ramifications that extend far beyond what any of us can truly comprehend.
Imagine if there existed a way to significantly improve the fuel efficiency of engines and powertrains. Even a seemingly tiny gain - say 2% or 3% - sounds trivial when viewed at the scale of a single vehicle. But when multiplied across tens of millions of vehicles operating daily, the consequences become staggering.
Suddenly, fuel imports reduce. Logistics costs fall. Public transport becomes cheaper to operate. Supply chains become more efficient. The cost of moving goods drops. Inflationary pressure eases ever so slightly. Entire businesses become marginally more profitable.
And that’s the strange thing about industries operating at this scale - tiny engineering optimizations stop being “small.” They begin to ripple outward into economics, infrastructure, policy, and eventually, everyday life itself.
The same applies to manufacturing efficiency, battery technology, thermal management, materials engineering, semiconductor reliability, predictive maintenance, or even something as mundane as improving the lifespan of a component by a few percentage points. At automotive scale, microscopic improvements compound into macroscopic consequences.
Modern civilization, in many ways, quietly rests upon the assumption that locomotion will remain cheap, reliable, and scalable. And the automotive industry sits right at the center of that assumption.
And in the same vein, say hi to @rhygentech .
More, shall follow soon :-)
English
our paper on EVE has been accepted to ACL 2026 industry track!
EVE is an end to end open source llm framework for earth observation intelligence, covering domain specific data corpus creation, custom LLM training, RAG based retrieval, and new EO benchmarks, with all code, data, and models publicly released.
paper and code here - eve.philab.esa.int/a-domain-speci…
English
@kirillk_web3 SWE-bench pro scores are useless at this point. Also Kimi's API performance is not that good.
English
> be Kimi
> nobody outside China pays attention
> everyone paying $200/month for Claude
> Kimi already there. 8x cheaper.
> drop K2.6
> 300 sub-agents. 4,000 steps simultaneously.
> 12 hours continuous execution. zero human oversight.
> beats Claude Opus 4.6 on SWE-Bench Pro
> open weights. free.
> Vercel: "50%+ improvement on our benchmark"
> while everyone was paying for closed models
> Kimi was quietly becoming the infrastructure
> different game.
Kirill@kirillk_web3
English
barelyOK retweetledi
