twotwotwo

14 ms·

61.
by twotwotwo 1y ago
The "they do different things" bullet is worth expanding. Wikipedia, arXiv dumps, open-source code you download, etc. have code that runs and information that, whatever its flaws, is usually not guessed. It's also cheap to se
62.
by twotwotwo 1y ago
Flash endurance gets a sort of odd amount of discussion on the Internet, and this adds to the small amount of public data from running lots of Flash devices to exhaustion. (MicroSD != SSD, of course, but it's something.) Seems possible
63.
by twotwotwo 1y ago
I give credit to Andy Weir for knowing what The Martian did well--setting up a bunch of technical problems as load-bearing elements of a plot--and going and executing that same general plan, but with new particulars and bigger and with fun
64.
by twotwotwo 1y ago
I recognize it as beating a dead horse now, but JPEG XL did what was needed to be actually adopted . AVIF has not been widely adopted given the difficulty of a leap to a new format in general and the computational cost of encoding AVIF spe
65.
by twotwotwo 1y ago
At work, one of the prompt nudges that didn't work was asking it to ask for clarifications or missing info rather than charge forward with a guess. "Sometimes do X" instructions don't do well generally when the trigger c
66.
by twotwotwo 1y ago
The latest V3 strikes me as a really practical go-to among open-weights models. Lots of tasks don't need the reasoning tokens, and not having to wait for them is nice. (If something does need it you can always switch.) If you're n
67.
by twotwotwo 1y ago
Any company in this space outside of the top few should be contributing to the open-source tools (Aider, OpenHands, etc.); that is a better bet than making your own tools from scratch to compete with ones from much bigger teams. A couple fo
68.
by twotwotwo 1y ago
It's kind of weird to think that in a coding assistant, an LLM is regularly asked to produce a valid block of code top to bottom, or repeat a section of code with changes, when that's not what we do. (There are other intuitively o
69.
by twotwotwo 2y ago
FWIW, work has pushed use of Cursor and I quickly came around to a related conclusion: given a reliability vs. anything tradeoff, you more or less always have to prefer reliability. For example, even ignoring subtle head-scratcher type bugs
70.
by twotwotwo 2y ago
> in-memory and disk-backed, in-memory KV stores, which haven't really gone away even with the introduction of NVMe storage If we're talking caches and Redis, one thought about why they haven't gone away-- In practice I th
71.
by twotwotwo 2y ago
A random thought that Might Work, Who Knows(tm): first compress long-ish repetitions in the input and store the copies separately from the literals (like zstd). Then run just blocks of literals through the BWT before entropy coding. The ide
72.
by twotwotwo 2y ago
I remember, when I was just starting, the founder of the company saying that if you tell him you're offering an 'engine' that's mostly saying you don't have an app. Of course, some successfully sell 'engines&#x
73.
by twotwotwo 2y ago
> I've been doing some planning over the last couple of years on teaching a course on how to do all this Yes! It's surprisingly common to hear it can't work, or can't scale or run reliably, when all that is done. Ta
74.
by twotwotwo 2y ago
I don't like what Automattic's doing, but I don't think that means we have to throw out the whole idea of obligations that aren't legal obligations.
75.
by twotwotwo 3y ago
It's interesting this is mentioned specifically about the metadata used by this feature: fingerprinting using this feature has similarities with other cache fingerprinting (wrote a sibling comment about that). It's not actively ba
76.
by twotwotwo 3y ago
I think fingerprinting using this is mostly like the more direct ways to fingerprint with the cache, and the defenses against one are the defenses against the other. For the cross-site thing, cache partitioning is the defense. If the cache
77.
by twotwotwo 3y ago
Right, more bandwidth would really help mobile iGPUs, maybe more than it'd help many-core CPUs. I'm not sure if they want to sell super-beefy consumer iGPUs, or if AM5 is physically capable of having more channels. But it would
78.
by twotwotwo 3y ago
Like the post says near the end, this seems like at least partly the result of AMD not being the size of Intel or Apple, so not being able to have the staffing to create, validate, etc. two radically different large and small designs. So in
79.
by twotwotwo 3y ago
Other replies are also right--this is in a family of analytics databases that are unusable for single-row transactions but great at reporting over large amounts of data. The other tool with a similar focus on querying data where/how it
80.
by twotwotwo 3y ago
I do not really expect that, but it's possible to have higher-performance iGPUs than PCs have now; the bandwidth of two memory channels is a big constraint here. Apple's M3 Pro and Max appear to use 3 to 8(!) channels (and of cour
81.
by twotwotwo 3y ago
Oh, thank you! Looks like they did keep the DeskMini very much like the A300/X300, including keeping SATA. Some other stories like https://videocardz.com/newz/asrock-announces-deskmeet-deskmi... suggest the DeskMi
82.
by twotwotwo 3y ago
Not a great headline when the distinguishing feature of these chips is building a complete system without a dGPU. You can make an all-around solid computer with these, like the well-regarded new 13" Framework laptops but in a desktop
83.
by twotwotwo 3y ago
Yeah--as a user I've even seen apps where e.g. my own post immediately appears in the feed and thought "I see what you did there!" Not as disagreement, but to help map the problem space out more, I think faking it is relative
84.
by twotwotwo 3y ago
A very simple thing about this (and many systems!) is that if your whole thing is "log writes and do the real work later", you lose read-your-writes and with it the idea that your app has a big persistent memory to play in. This d
85.
by twotwotwo 3y ago
Not inconsistent with this, what seems warped to me is having these huge expectations for individuals after their first success in the first place. (Writing this thinking about someone like Romero working on something like a game; lot of wh
86.
by twotwotwo 3y ago
Yep, I very much agree with this. It's probably inevitable that languages with reflection grow some kind of language-specific serialization format because a) they can and b) there's often some use case where it looks handy to ha
87.
by twotwotwo 3y ago
I'm glad someone from AWS is here! Some general info on the sharding approach would be interesting, e.g. whether secondary indices on sharded tables are scatter-gather, or separately sharded on the index columns, or you configure it pe
88.
by twotwotwo 3y ago
I am pretty sympathetic to a CSPRNG being the default in standard libraries and so on. The relative performance has vastly changed since the first stdlib random()s; modern hardware- or SIMD-accelerated functions run at GBs/second per c
89.
by twotwotwo 3y ago
Sorry, I know you should be using a CSPRNG API for cryptography, and probably my glib phrasing there obscured that. I'm saying the cost of running actual cryptographic primitives has dropped a ton over time: on computers from decades a
90.
by twotwotwo 3y ago
Sorry, to clarify, I know CSPRNG APIs are exposed and what you use for safe randomness. What I'm saying is that outside rare, extreme situations--maybe some HPC Monte Carlo simulation where you need gobs of randomness--you might as wel

More ›