9,514 karma · joined April 20, 2007
This is the one case (AFAIK) where input power isn't being converted to some kind of heat: it's being stored in the battery as energy for latter.
Pulp folks' Occamy is supposedly going to get a run of chips made this summer. 432 cores, each with an incredibly innovative vector processor. Really tapping those vector cores takes effort, & floating point isn't what most servers need/use, but this is still one very rapidly approaching very interesting data point, a huge step function where RISC-V could leap to the front. https://www.tomshardware.com/news/432-core-occamy-cpu-for-us... https://news.ycombinator.com/item?id=35873153
Seems like a decent success story, in the long run. GNOME did what Wayland is good at/what we couldn't do before & made a semi secure system. But GNOME over secured, made things too hard. We learned & improved things.
The page here catastrophizes with abandon. Like much zealotry surrounding Wayland. Things can be bad, times might not be smooth, but recognizing the arch, having the empathy to recognize progress & effort: that's thr capital spirit of open source.
Supposedly the F-14 with it's swivel-swing has a big 1 ton titanium box at the center. And there's a lot of those jets. But this 1.5m sphere that's 9cm thick is quite the huge single chunk of titanium.
TIL that Gabe Newall of Steam bought the Limiting Factor fairly recently.
Red Hat is surrendering their role in open source because they insist on commercial exploitation.
> The brain infrastructure that connects neurons is collectively referred to as the cytoskeleton. It includes proteins, neurotransmitters, filaments, microtubules etc. Above, we saw how electric fields generated by neurons and other parts of the cytoskeleton carry information being used for behavior. Here, we propose the Cytoelectric Coupling Hypothesis: Electric fields help organize components of the cytoskeleton in order to “tune” it to process information efficiently.
Stunning complexity.
The way sites have previews seems more nebulous, but ultimately it just helps drive traffic.
There's a lot of focus on how Intel allows too much iteration to happen, with hot bins and many samples. I can see that's indicative of a problem, can be a crutch you shouldn't be needing to lean on.
But at the same time, being able to iterate & work on the next thing actively seems like a huge potential advantage. You wouldn't want to survive off it, take it for granted, like it seems Intel has, but having to take your design & wait in line like everyone else, for your scheduled slot for chips, seems like a massive drag on innovation & improvement.
Enterprises want RHEL because the staff know & use RHEL and the enterprise can get a support package for RHEL. If Red Hat no longer is in the public eye, if it's only a paid for Linux, the demand is going to dry up, vanish.
There's only one way to retain mindshare & market share & that's to keep RHEL in front of people. This current course of action almost guarantees that RHEL fades into irrelevance.
It just is so remarkable to me that no other games dare to have the depth & accessibility that EVE does. No one else has anything remotely as open ended as EVE Online's existing APIs or data dumps. It's shocking to me that games remain so quaintly isolated in their own world, that they don't seek to open up to allow more.
I came back latter to try to understand what was going on, why I'd need an account. I could find nothing helpful to tell me what was going on. I have never seen a more nebulous relationship with a company. I could be missing lots of hints, but I tried real hard to understand what's happening here & all of it felt extremely suspicious.
hen I see an intent to ship, seeing other browser indicators/status would make that first browser's intent to ship more actionable.
Sounds like another possible convert for the Malleable Systems Collective. (And a belief I tend towards as well.) https://malleable.systems
Their top of page quote, from Lick himself,
> The user wants open software, software that can be modified, and that can participate in a progressive improvement process.
Not as many users resemble this as they used to. But they are the interesting users, from whom new value & possibilities will spring. Attend to them.
Different programs introducing their own specific programmatic layers defeats the holistic small pieces, loosely coupled premise.
Links should be links to places. If your link is an option menu, it's not a link. The friction here is terrible. There's literally an org visible & a repo visible, but someone decided I'm not capable of clicking/tapping links & made the experience way higher friction & slow.
This is good news though. It needs to expand a lot. But there's so many other big needs ROCm has, so much quality/stability improvement necessary. And CUDA has such a huge huge range of libraries that AMD has so much yet to reimplement.
Great great bot here, even as is. The web is amazing.
CUDA has been Nvidia's key enabler, makes their hardware just work. It'll be interesting to see - here & about - whether they can keep a wide customer base able to use & leverage this hardware.
The major downside is the Aladdin problem. Vast cosmic power. Itty bitty living space (& mortal lifespan to spend). These capabilities we have could be better spent, do end up entangled in so many frustrating stifling enterprises. But they're still amazing abilities. I'd wish such great adventures for anyone.
But I cannot wait to get off such a special purpose system. I would much rather be running a regular & general purpose OS. I'd rather have systemd and regular builds of packages.
Ooenwrt exists mostly because good wifi is coupled to byzantine difficult hardware that requires absurd patches & guesswork to make that hardware useful. There's been no hope for most of these systems to ever get their support up-streamed. We are seeing some positive changes very recently, with for example Qualcomm & MediaTek both seemingly not far from being usable, which is lightyears better than where we have been. Conceivably we could run whatever on these devices at that point.
Also super sad is how hard it is to avoid these frustrating quirky devices in the first place. Trying to buy wifi add-ons is enormously challenging. M.2 (or often still mini-PCIe) cards are made by very very few folks, hard to find, and often around $200. Once you acquire one, now you're faced with it's awkward size that won't fit most places, and the often weird power demands like having some 12v lugs you have to hook up.
Ideally I'd love to scatter a couple rpi and some usb wifi cards around. It's been a while, but ~5 years ago I was buying & trying every Alfa (a good name that makes usb cards for a variety of chipsets) usb card I could, and usually hosting an AP started great but would quickly fall apart. Realtek only supported like 8 client devices, for example. Other chipsets would crash & even rmmod & re modprobing drivers wouldn't bring the device back up. Some just had terrible performance. It's such a pity USB gets such poor treatment, that there aren't some capable usb cards that can be a decent AP.
Have a roommate now who I need to start talking to. I've been putting it off for months. It's hard. I've had many roommate conversations & there seems to be such a real chance that it makes things worse. Shitty people tend to get shittier.