> Google’s new Minnesota data center comes with the world’s largest battery—and won’t raise electric bills
> The tech giant says it will fund enough new wind, solar, and long-duration storage to cover the project’s power demand and avoid shifting costs to ratepayers.
For me personally it goes I like TUI/CMD for things I use as a daily driver, I know what I want to accomplish I know exactly what movements to do and clicking GUI items is just too slow.
Then there are GUI things that I don't use that often I need some kind of map, I don't to "read the fucking manual" every 3 months when I use it and I barely remember, but if I see it in GUI my memory will kick in. It also works for most new things better than TUI/CMD because for a new tool I don't want to invest my focus, I just want to find out how to be done with my things or just learn what it is, reading manual is not fastest way, fastest way is clicking around.
Nowadays we have 3rd mode namely "chat interface" I can just chat with whatever bot is integrated in the app or just chat with the bot to use the application for me. Command line utilities work like charm in that mode I can accomplish a lot by asking vague stuff to a bot that will make stuff for me in a ways I don't even have to know how or what.
But I do believe final form of interface is not "chat interface" alone. I do believe what Karpathy already outlined, that there will be specific interfaces for systems that will use GUI elements to make it clear for the user what is going on and will be faster to clear the information to the user instead of having user to read back text.
I find it extremely hard to believe that the current administration would do that. If it has happened, it was prior administrations (which honestly were just as weak in addressing this concern). And if they still are on some list, its likely just an oversight and is a matter of time before they get the Proud Boys treatment (a presidential pardon, a settlement for emotional damage or...whatever, and a bunch of confidence that they don't need).
I was unaware of these side-loaded malicious apps until now. This is information consumers need to have.
It's very reminiscent of Sony putting rootkits on CD's. Unwanted, dangerous software is being loaded onto your computer by people you paid money to. The companies involved, including MS, should face serious blowback over this, as Sony did.
I enjoy forums, so I prioritize spam filtering to preserve signal to noise. Others may prefer to read every cialis ad. I think these two extremes and everyone in between can coexist on the same forum, but in order to do so it helps to have adequately tuned filters customizable on a personal level. In more detail if you care to: https://wiki.roshangeorge.dev/w/Overmod#Philosophy
+1. I saw the 2025 suffix and fact that it was on the original domain and hoped they released a v2 to carry us through the world of Platform Engineering, Observability 2.0, Kubernetes vs Serverless, etc.
> Every default malloc implementation worked this way about 12 years ago.
Yes, it is now 12 years later, and memory allocators are better. The memory allocators of that time were also better than the ones 12 years their prior. That's the point.
> Making lots of small allocations, even from multiple threads then blaming the allocator is a losing strategy. An allocator is only going to be able to mitigate the damage to speed and interactivity.
It's just a reality that musl is measurably worse at multiple threads allocating even in very polite conditions, because it causes lots of contention. If your program allocates in multiple threads, it is probably going to get slower with musl. If you don't want that, other memory allocators will do great even at high allocation rates with more threads.
You could write many other data structures that had equally poor behavior under multi-threaded contention by just throwing a lock around everything and calling it a day, and those bad data structures would also cause "damage to speed and interactivity" or whatever. This isn't very hard to understand.
> The compiler should be able to take something with good access patterns and make something fast, especially out of the basic C functions.
I agree, modern compilers are good. But these are extremely common specified functions, they are called everywhere all the time in every C codebase. Mature implementations that are hand optimized still pay off and also tend to be tuned for various edge cases or quirks that aren't going to come for free from the C compiler either, so it's still work even if you aren't writing assembly for everything or whatever (e.g. uarch dependent codepaths, or optimizations for short strings or whatever).
glibc's AVX2 based memcpy functions have a non-negligible performance impact in at least 1 application I maintain on the order of like 8-ish% vs musl (wall clock). It just has to memcpy/memmove a whole lot.
> Performance wise it's unlikely C string functions are actually the bottleneck in a program. Maybe for specific programs a naive memory copy function could benefit from AVX instructions.
I said "backbone", not "bottleneck". They are common functions sprinkled in everywhere throughout every application in every codepath on something like a modern Linux desktop. An inverted callstack flamegraph can show you stuff like this.
It is basically no different than compiling your application at -O1 and -O2 with GCC. Does the fact your program get 20% faster from -O2 mean that there were "bottlenecks" in it? No, there was just performance left on the table by emitting better code.
> "Have to" and "often" are debatable.
Not really. I have to spell it out apparently: actual programs written by normal human programmers do those things, all the time, they exist in and are common in the world, they depend on other code that does that and is common in the world, they run on your desktop and phone and all servers, and they benefit quite a lot from optimized components like memory allocators and string routines and -O2 making their programs faster. This is pretty easy to observe and the means of doing so should be quite obvious, so there's no real debate.
Now whether this fact holds -- whether these programs "have to" do these things or not -- in the imaginary fantasy land people have in their heads where they make up arguments to themselves about how, if every program was written how they liked it, it would be better? That I'm not so sure about, I will admit.
Opus 5 (xhigh): ~8 hours of agentic churn, ~1 hour of my time to test features and direct the agents.
As an experiment I provided ZERO technical guidance, I let it pick architecture, programming language, GPU interface, rendering interface, all of it. I only provided feedback from the perspective of a user and only things visible to a user. I.e. "Adjusting the curves spline is too slow. Make it instantaneous", I did not ask it to downsample the preview or move the operation the the GPU.
As a result I have a fully functional, super fast, stable, Linux based 2D image editor, written in Rust and D, GPU accelerated (Vulcan). Startup time is 0.458 sec. Every operation is instantaneous on a sub $1000 laptop with RTX4060.
I needed all the features that I use personally and I did not care about features I don't use or that I don't know about, so everything I needed was:
- typical format support: png, jpeg, bmp, tiff, targa
- layer management (new, delete, lock, merge, duplicate);
- layer blending (opacity, fill, and the whole panel of blending types - darken, difference, screen, etc);
- filters with preview render for live adjustments;
- drawing and selection tools complete with properties for each;
- history (undo/redo);
- complete UI - fully stylized window, window setup presets and flexible panel allocation;
- right click, alt click, shift click support for tools, matching key bindings;
- open and save a range of formats;
- text insertion (vectorized) and ability to rasterize the text;
- vector objects and ability to rasterize them;
- testing framework that can simulate mouse clicks, drawing and UI interactions;
- complete regression testing suite
I dont know Rust, I dont know D, I dont know how to use Vulcan. I know image processing and software engineering but it didnt matter at all. I only provided typical user feedback - i.e. "I cant see layer controls in the UI" or "curves adjustment is lagging when I move the spline, must be instantaneous". Developing this software required no skill, insight or know-how.
I added a few further improvements that I noticed missing so the latest version has more features with a total agentic time at 10 hours and my own time at 3 hours.
I am now using it as a replacement for Photoshop that costs $275/year in a predatory subscription where you cant cancel without a $200 penalty.
I think a lot of people out there are being manipulated about things that they don't understand.
I saw some comments from people against a proposed data center at the site of an old power plant. I pointed out that it was a mile from anything. Someone replied saying that that was still too close, and that they wouldn't want to live 50 miles from a data center.
Btw, did anyone notice they removed access to "who starred the project" recently. You can only check your own repos, not external ones. It was very nice to find related people before.
If 'it' is unnameable, there is no way to circumscribe or even describe what 'it' is. Even to show that what it refers to is an empty set, we need its description. If we use concepts like intention and extension, we can sketch out four scenarios:
extension, no intension (yes, we can point out things, which we can't describe)
extension, intension (we point out, and we describe)
no extension, intension (Yes, we can imagine and describe things vividly, but no referent in the world. Here, one can say these things exist in a Platonic world, but not the world we live in; this is where numbers, sets, ideas can exist. Neo-Platonism in Philosophy of Mathematics)
no extension, no intension (this paradox falls in this area).
It really is a testament to the power of mindshare and vendor lock-in. Git is GPL-licensed, the format is a de-facto standard, the commits and tags live in an on-disk format, merging and collaboration doesn't require a server, and yet somehow we still managed to end up with a centralized system with GitHub at the center.
I agree with you. I simply took umbrage with the puerile characterization of other human beings with different points of view as “bots”.
It’s really no different than the cringy characterization of other humans as “NPCs” often spouted by children. That sort of immature thinking exposes a superiority complex that is very off-putting.
Given that, this person would likely benefit from exposure to alternative viewpoints instead of further insulating themselves.
Any given single agent is not long-lived due to limited context. What impact does it have on the status signals the agent develop compared to the signals humans (who usually have much longer context) have developed?
> Google’s new Minnesota data center comes with the world’s largest battery—and won’t raise electric bills
> The tech giant says it will fund enough new wind, solar, and long-duration storage to cover the project’s power demand and avoid shifting costs to ratepayers.