bonzini

13 ms·

241.
by bonzini 9mo ago
LLMs are probabilistic black boxes, trying to inject determinism in their natural language processing (as opposed to e.g. forcing a grammar for the output) may very well screw them over completely.
242.
by bonzini 9mo ago
I think it's plausible that different languages would prefer different tokenizations. For example in Spanish the plural of carro is carros, in Italian it's carro. Maybe the LLM would prefer carr+o in Italian and a single token in
243.
by bonzini 10mo ago
(Former) pawns can be on the 1st or 8th rank if they've been promoted. You can place an unpromoted pawn on the 1st or 8th rank to encode en passant, too. Also castling can be encoded as extra squares available to the rooks only (&quot
244.
by bonzini 10mo ago
It also can encode chess960 positions. With the article's encoding, uncastled rooks can only be decoded if their starting position is known, which it isn't in chess960.
245.
by bonzini 10mo ago
Challenging the IP bans in Italy is stupidly hard. Your VM gets an IP address that was used a few months ago for soccer piracy? Too bad, you won't be able to access it from Italy.
246.
by bonzini 10mo ago
Doh of course. But the 26 bytes + 2 bits irritated me so I didn't think about it.
247.
by bonzini 10mo ago
Bishops only need 5 bits instead of 6 (they can't move to a square of different color), shaving 2 bits and thus reaching exactly 26 bytes. BTW 495 can be computed as a binomial coefficient C(8+5-1,5-1), the number of combinations of 8
248.
by bonzini 10mo ago
NNUE is for deep searches, as far as I understand this just says what move to do based on the state?
249.
by bonzini 10mo ago
The actual innovation in QEMU was that the architecture-dependent part was much smaller than a full JIT compiler, because it used the C compiler to build small blocks and parsed ELF relocations to be able to move them into the translated co
250.
by bonzini 10mo ago
> Everywhere. Show the PRs.
251.
by bonzini 10mo ago
That's your distro, not systemd. systemd only does what the configuration files tell it to do.
252.
by bonzini 10mo ago
I have checked three smaller machines. On one (coreelec) systemd has 7M of resident data, of which 5.5M are shared libraries; by comparison, the numbers for sshd are respectively 6M and 3.5M. On an OpenWRT machine without systemd (it's
253.
by bonzini 10mo ago
That was almost 15 years ago and the support is evidently not as useful. Also it's entirely contained within a program that creates systemd .service files. It's super easy to extract it in a separate project. I bet someone will do
254.
by bonzini 10mo ago
Technically that's part of podman, not systemd. But it's the same architecture that was used to support sysvinit scripts. (In fact, nothing prevents anyone from extracting and repackaging the sysvinit generator, now that I think o
255.
by bonzini 10mo ago
> Why wouldn't you want unit files instead of much larger init shell scripts which duplicate logic across every service? Indeed, that criticism makes no sense at all. > It also enabled a ton of event driven actions which laptops&
256.
by bonzini 10mo ago
As I wrote in another comment, just check out /run/systemd/system. You'll find the wrapper units that systemd creates for your sysvinit scripts.
257.
by bonzini 10mo ago
Thanks!
258.
by bonzini 10mo ago
It is, but my point is that these entries are more historical oddities than a sign of particularly high population density of Spanish cities. There are similarly dense suburbs in most cities/countries, they just happen not to have a lo
259.
by bonzini 10mo ago
Not only it's easy, the exact contents of the systemd unit can already be found in /run/systemd/system.
260.
by bonzini 10mo ago
> (#2 and #3 are both in Spain at https://en.wikipedia.org/wiki/List_of_European_Union_cities_... in "density per km2" for example, 50% of top 10 on that list is in Spain!) That list does not make much se
261.
by bonzini 10mo ago
Sanctions in the 1950s?
262.
by bonzini 10mo ago
Three-four years is a lot of time for these companies to face the harsh reality.
263.
by bonzini 11mo ago
Only to later add switches that chose between x87 and SSE? :)
264.
by bonzini 11mo ago
> Forget the bells-and-whistles and answer this: does the use of Rust generate a result that is more performant and more efficient than the best result using C? These are the performance results for an NVMe driver written in Rust: https
265.
by bonzini 11mo ago
That was way way blown out of proportion. https://lwn.net/Articles/991062/ has a much less somber quote from just three months later: > Ted Ts'o said that the Rust developers have been trying to avoid scar
266.
by bonzini 11mo ago
There are devices that do not have access to memory, and you can write a safe description of such a device's registers. The only thing that is inherently unsafe is building DMA descriptors.
267.
by bonzini 11mo ago
You can absolutely write drivers with zero unsafe Rust. The bridge from Rust to C is where unsafe code lies.
268.
by bonzini 11mo ago
Converting to RPN is, roughly speaking, the easiest way to generate code for either register or stack architectures. Once you have a parse tree, visiting it in post order (left tree, right tree, operation) produces the RPN.
269.
by bonzini 11mo ago
Yep. It can refactor very well but that's it. For complex code bases it cannot even build boilerplate that makes sense; at most it saves some typing.
270.
by bonzini 11mo ago
I did find some benefit in lowering the cost of exploratory work, but that's it—certainly worth 20€/month, but not the price of any of the "ultimate" plans. For example today I had to write a simple state machine (for a

More ›