I have never done anything which has been better for my own well-being than buying a camper, building it out into a mobile R&D lab and just going off into the mountains to grind away in peace while I'm in the states. The US has a way of sucking the soul out of people with the commute/9-5 structure and city life - figure out a way to make your entire life play out the way other people vacation.
Meet Gemma 4 12B!
A unified, encoder-free multimodal model designed to bring high-performance intelligence directly to your laptop, and released under an Apache 2.0 license.
Bridging the gap between edge efficiency and advanced reasoning. Here is what’s new with Gemma 4 12B: 👇
You really have to define “simple” here because it doesn’t get any more simple than a typical gantry setup with two motors if constrained or two opposing steppers with wheels if not constrained.
If simple equates to the least amount of electrical parts, a single motor can be used for full, low precision x-y navigation by using either a centrifugal clutch + gear reduction for angular control or a forward-backward engagement paw for control of one direction for advancing and the other for turning.
@blind_via This is what I keep telling people thinking this will be all automated soon… trace routing is maybe 1/10th of the time designing a board. Component selection and pouring through data sheets is the time consuming part 🥲
@pirofession@haydendevs the 12-bit ADC of the new model is great for resolution without scaling but I really need the 8Gsa/s because I occasionally have to test high speed signal traces. I may upgrade my other scope in my mobile lab to the new MHO934.
@mholt6 Data centers aren’t running them but NAS folks are resorting to spinny boys because SSD prices are through the roof. Supply and demand unfortunately.
🚨🇷🇺 WEST SHOCKED: RUSSIA JUST UNVEILED 'ALTAI' NEUROMORPHIC CHIP
Russia has presented its brain-mimicking ‘Altai’ neuromorphic processor — a homemade technological leap that could redefine AI warfare and confirm Western sanctions useless.
🔸 Altai delivers over 1000X better energy efficiency than traditional chips by using spiking neural networks that only fire short impulses when needed — directly copying the ultra-low-power human brain at just 20 watts.
🔸 Processes video at up to 2200 frames per second while consuming under 0.5W in a tiny 9×9 mm package — compared to NVIDIA Jetson which burns 15-60W for similar AI vision tasks.
🔸 Features 256 asynchronous cores simulating 131,072 neurons and 67 million synapses — developed by Novosibirsk’s Motive NT with Kaspersky as strategic investor and presented to PM Mishustin last year.
🔸 Transforms low-cost drones into long-endurance autonomous hunters, turns sensor networks into stealthy weeks-long intel platforms, and powers radio-electronic warfare gear that runs days on a single battery.
🔸 Currently a 28nm prototype still needing fab production — Russia may rely on Chinese facilities or adapt to coarser domestic 350nm nodes, trading some efficiency for full sovereignty.
Can the U.S. catch up to Russian chip innovation?
Competitive with who? I have two 35B models running 24/7 on research tasks at 4.3tok/s and another running at 15.6tok/s performing analysis on a set-it-and-forget-it setup which just reports to me daily on medical trends and grants. We have 10+ frontier models to choose from if we need speed or maximum intelligence. What exactly do you need your laptop to do which your home server can’t?