moonchild

7 ms·

31.
by moonchild 2y ago
https://aplwiki.com/wiki/FinnAPL_idiom_library
32.
by moonchild 2y ago
total order on all accesses to a given location—loads from a single location can't be reordered w.r.t. each other either i don't remember seeing any wording relating to mixed-size accesses in the intel manual (not withstanding t
33.
by moonchild 2y ago
That's more correct, but there are two points of order. The first is that a cpu core is a massively distributed system, and there is no single point in time when an operation is executed. The second is that cpus will absolutely physi
34.
by moonchild 2y ago
Geez! Didn't henry baker warn against implementing closures this way decades ago? This is an implementation bug—the code is fine.
35.
by moonchild 2y ago
It's not documented as being undefined; it's simply not documented at all. Intel's latest uarch does partial store forwarding. There was a paper from a few years ago trying to define semantics for mixed-size accesses (not for
36.
by moonchild 2y ago
> If I were writing a program in C99, I would assume it would still be possible to communicate the same intent / restrictions to the compiler, but I'm not sure because I haven't been able to find any resources that discuss
37.
by moonchild 2y ago
> more traditional and frankly much clearer terminology about actual hardware behavior and not the frustrating abstractions embraced by the language design community Axiomatic memory models were pushed as much from the hardware world as
38.
by moonchild 2y ago
> x86 CPUs do allow themselves to reorder reads around other reads. The rule is that no memory access is allowed to cross a write operation That's not correct. Intel manual, vol 3, sec. 9.2.2: > Reads are not reordered with othe
39.
by moonchild 2y ago
Is the full dissertation available anywhere? The link on the author's website is dead.
40.
by moonchild 2y ago
checking for an overlong input is off the critical path, so latency is irrelevant. (tfa doesn't appear to use a branch—it should use a branch)
41.
by moonchild 2y ago
Ah! You are right.
42.
by moonchild 2y ago
Unicode collation is very complex. If you wanted to sort according to that, then you would need to devise an isomorphism between unicode strings and bitstrings such that lexicographic ordering in the bitstring domain agrees with unicode co
43.
by moonchild 2y ago
> repository of analyzed code yes, this is obviously what you want. there shouldn't be a notion of 'running' the compiler; it should simply be a persistent process. importantly, there are two issues, which are orthogonal
44.
by moonchild 2y ago
No, not really. A couple of features of the architecture and instruction encoding which were historically used for segmentation remain on amd64, and these features are used to implement thread-local storage on every mainstream os targeting
45.
by moonchild 2y ago
I won't make the argument against variable-length vectors here, but I will note that, although some of the extensions you mention do change the vector length, for none of them is the vector length change the most interesting aspect; yo
46.
by moonchild 3y ago
Respectfully: 1. I think censorship laws are abhorrent 2. It can be very difficult to tell what effect media has on you, and just because you do not (say) find yourself traumatised or find that your perceptions have been overtly altered doe
47.
by moonchild 3y ago
> There a few new remixes or interpolations every year. Some ask first, and some don’t. The last one to ask permission was the artist Pink, who I love Is Pink's remix publicly available anywhere?
48.
by moonchild 3y ago
capability safety inhibits supply chain attacks. memory safety is a prerequisite, but it is not sufficient
49.
by moonchild 3y ago
Video hosting is expensive—is that really enough to cover costs?
50.
by moonchild 3y ago
std::vector falls into the category of things that are easy to bounds check, st the cost, even under today's primitive compilers, is low. It's direct pointer accesses—which are common in c but not in c++ or most other languages—t
51.
by moonchild 3y ago
You can add bounds checks to c, but that costs a hell of a lot more than 1-2%. C++ has them off by default for std::vector because c++ is designed by and for the utterly insane. Other than that, I can't off the top of my head think o
52.
by moonchild 3y ago
The author is dead.
53.
by moonchild 3y ago
sorry, I didn't read the linked post closely enough—from my reading, this case is not one of the ones that was broken
54.
by moonchild 3y ago
VLIW (Itanium is a VLIW arch) is what's contentious, not multiprocessing.
55.
by moonchild 3y ago
object.field is implemented as a direct load from the object; if the object turned out to be null, then the resultant signal is caught and turned into a NullPointerException
56.
by moonchild 3y ago
> M1 etc do even better (they can speculate prefetch through a pointer Interesting. Do you have a reference or more information on this?
57.
by moonchild 3y ago
Not saying it's not worth it, just, it's strange to put them in the same sentence when one is what makes the difference between usable and completely unusable. I lived with 20-60s dirty build times for 1-line changes at symmetry,
58.
by moonchild 3y ago
'Compilers' should be incremental (and iterative), and the asymptotics are way more important than the constant factors; if the conceptual models are not adequate to effectively express very fine-grain incrementality, no amount
59.
by moonchild 3y ago
Caches, register files, and coherency traffic all definitely include error-correction.
60.
by moonchild 3y ago
> I've noticed a similar trend among folks who work on compilers, preferring to "stay in their lane" instead of embracing GPT-4 Research into ML-guided optimisation predates chatjippity.

More ›