fulafel

10 ms·

181.
by fulafel 6mo ago
"TPU 8t and TPU 8i deliver up to two times better performance-per-watt over the previous generation" sounds impressive especially as the previous generation is so recent (2025). Interesting that there's separate inference and
182.
by fulafel 6mo ago
Could they safer and/or higher dose acetaminophen pills if they included NAC?
183.
by fulafel 6mo ago
In the comparison table, kuri-fetch is described as "no Chrome needed".
184.
by fulafel 6mo ago
There's no mode spelled the same ("High Performance") - and I don't think Linuxes universally do this: > Processors are always locked at the highest performance state (including "turbo" frequencies). Unless
185.
by fulafel 6mo ago
For others unfamiliar with Windows, according to https://learn.microsoft.com/en-us/windows-server/administrat... "High Performance" entails: > Processors are always locked at the highest performance
186.
by fulafel 6mo ago
Macs or other competing systems don't have on-die memory. (Except for the caches, which everybody has)
187.
by fulafel 6mo ago
Battery capacity of smartphones seems to double every ~8 years. The design space is adding more battery capacity, reducing battery life, or using less power.
188.
by fulafel 6mo ago
Putting the battery outside the water insulation zone might work for that, it's a sealed pouch anyway.
189.
by fulafel 6mo ago
In deed. We are really late in ramping down fossils usage and emissions, and the death toll is higher than the other bad things in the news headlines.
190.
by fulafel 6mo ago
Yes, but this was to appease Windows, probably older versions and/or 32 bit versions of it.
191.
by fulafel 6mo ago
> Heights were fairly new, and exciting, and the consequences of falling were not well understood yet. Indeed. Until Herbert Fall wrote his treatise about it in 1902 the entire concept was really absent from our collective imagination.
192.
by fulafel 6mo ago
To the first question: blame Windows I guess. But even on older chips, GPU code could access memory allocated on the CPU side so this didn't cap the amount of data your GPGPU code could crunch.
193.
by fulafel 6mo ago
> Apple Silicon changes the physics. The CPU and GPU share the same physical memory (Apple's Unified Memory Architecture) ... no bus! Beware the reality distortion field: This is of course how it's worked on most x86 machines f
194.
by fulafel 6mo ago
I'm not a Mars colonisation advocate, but sounds like exposure to that may be manageable: "Perchlorate is toxic to people only in the sense that it can disrupt the production of thyroid hormone, an important growth hormone needed
195.
by fulafel 6mo ago
A lot of the frameworks do it. There's RC in GNOME/GTK, C++ stdlib, and built into Objective-C. And of course it's easy to think of lots of apps that heavily use those or another form of GC.
196.
by fulafel 6mo ago
On the other hand RPi doesn't support suspend. So which wins depends if your application is always-on.
197.
by fulafel 6mo ago
Aisle said they pointed it at the function, not the file. So, the nr of LLM turns would be something like nr of functions * nr of possible hints * nr of repos. Could indeed be a useful exercise to benchmark the cost. This would still be mor
198.
by fulafel 6mo ago
It was architecturally a 16-bit CPU so a flat 32-bit address space would be a non sequitur. If you wanted flat 32-bit addressing, there was a contemporary chip that could do it with virtual memory: Motorola 68010 + the optional external MMU
199.
by fulafel 6mo ago
Several operating systems on 286 (eg Xenix, Coherent, OS/2) used its MMU for multitasking and memory protection. See https://en.wikipedia.org/wiki/Intel_80286#Protected_mode
200.
by fulafel 6mo ago
RFC 7368 is a 2014 "informational" (no ietf standing) doc so it's not a source for current IETF advice. Also it was part of the since closed "homenet" working group initiative trying to define some new stuff that di
201.
by fulafel 6mo ago
> mDNS operates at the link-local scope (link-local addresses) This is not the case for the addresses returned. See eg https://www.rfc-editor.org/rfc/rfc6762 6.2. Responding to Address Queries When a Multicast
202.
by fulafel 6mo ago
You also get this if you use mDNS, but without the ULA hassle and you get to use DNS names instead of hardcoding IP addresses.
203.
by fulafel 6mo ago
If you want to have an airgapped network, sure. For most people it doesn't make sense. You'll just get the worst of of both worlds.
204.
by fulafel 6mo ago
People have, of course, been looking. Linux has been the #1 corpus for the methods for ages.
205.
by fulafel 6mo ago
Indeed, the tragedy of the IPv4+NAT stockholm syndrome is that people view having to use ambiguous addresses as access control and can't distinguish reachability vs addressing.
206.
by fulafel 7mo ago
Mapped to modern concepts I'd say it was about iterating from a MVP. "Gabriel argued that early Unix and C, developed by Bell Labs, are examples of the worse-is-better design approach." Whereas vibe-coding is not reviewing wh
207.
by fulafel 7mo ago
People pass around stickers (or at least used to) in hacker events saying that so there has to be something to it, right? Protesting the term is, I'd wager, motivated by something like: it sounds innocuous to nontechnical people and ob
208.
by fulafel 7mo ago
You can implement the graphics part of it using WebGL. It's strictly a graphics API for drawing to the screen. But there are specific libraries for eg physics that you can use in your WebGL 2 app, or entire 3D engines (like those you m
209.
by fulafel 7mo ago
Most of free software (incl the BSD stuff) was like that. The bazaar was an attempt to characterise the new linux style way of doing it.
210.
by fulafel 7mo ago
> In such an environment the container would crash, we see the violations, delete it and dont' have to worry about it. This is the interesting part. What kind of UI or other mechanisms would help here? There's no silver bullet

More ›