eklitzke

5 ms·

1.
by eklitzke 3d ago
Encrypting filenames is an interesting idea, but also doesn't really seem that useful in practice to me. Most projects or directories probably don't contain filenames that are sensitive. If you do have sensitive filenames, storing
2.
by eklitzke 3d ago
You only "need" to do this if you are handling arbitrary untrusted filenames. It's pretty common for people to write shell scripts that are only used for handling files they create that don't have unusual characters, and
3.
by eklitzke 7d ago
It's widely believed so because Stockfish finds all of the right (or better moves) in the games from the original paper.
4.
by eklitzke 7d ago
AFAIK there isn't data about fatality rate for rideshare services (I'm sure Uber has this data but I don't think they make it public). Without this data I wouldn't automatically jump to the conclusion that rideshare serv
5.
by eklitzke 4mo ago
Most of the training cost is not in the final training run, it's in all of the R&D (including salaries, equity, etc.) that it takes to get to the final training run. The actual cost of all of the TPUs (or GPUs), power, networking,
6.
by eklitzke 5mo ago
A well run public transit system should obviously be cheaper at scale than robotaxis, but the incentives for Waymo (or Uber, or Lyft, etc.) are very different than the city's incentives. It's very possible that in practice private
7.
by eklitzke 5mo ago
Yeah, none of this makes sense to me. Allocating memory for stack space is not expensive (and the default isn't even 1MB??) because you're just creating a VMA and probably faulting in one or two pages. They also say: >The syste
8.
by eklitzke 5mo ago
I can't speak to what every team at Google does, but there are machines with Nvidia GPUs in Borg. However Google charges orgs internally for cpu/memory/gpu/tpu usage and TPUs are *way* more efficient in terms of FLOPS&#x
9.
by eklitzke 5mo ago
This is true but the relative overhead of this is highly dependent on the protobuf structure in one's schema. For example, fixed integer fields don't need to be decoded (including repeated fixed ints), and the main idea of the &qu
10.
by eklitzke 9mo ago
Pretty much all of the history of HN front pages, posts, and comments are surely in the Gemini training corpus. Therefore it seems totally plausible that Gemini would understand HN inside jokes or sentiment outside of what's literally
11.
Cosmological Lithium Problem(en.wikipedia.org)
4 points by eklitzke 10mo ago|0 comments
12.
x86 architecture 1 byte opcodes(sandpile.org)
90 points by eklitzke 11mo ago|31 comments
13.
by eklitzke 11mo ago
I think it's important to point out the distinction between what POSIX mandates and what actual libc implementations, notably glibc, do. Nearly all non-reentrant POSIX functions are actually only non-reentrant if you are using a 1980s
14.
by eklitzke 11mo ago
NUMA has a huge amount of overhead (e.g. in terms of intercore latency), and NUMA server CPUs cost a lot more than single socket boards. If you look at the servers at Google or Facebook they will have some NUMA servers for certain workloads
15.
by eklitzke 1y ago
A few reasons, I think. The first is that getaddrinfo is specified by POSIX, and the POSIX evolve very conservatively and at a glacial pace. The second reason is that specifying a timeout breaks symmetry with a lot of other functions in Uni
16.
by eklitzke 1y ago
I agree. If you actually know what you're doing you can use perf and/or ftrace to get highly detailed processor metrics over short periods of time, and you can see the effects of things like CPU stalls from cache misses, CPU stall
17.
by eklitzke 1y ago
Writing drivers is easy, getting vendors to write *correct* drivers is difficult. At work right now we are working with a Chinese OEM with a custom Wifi board with a chipset with firmware and drivers supplied by the vendor. It's actual
18.
by eklitzke 1y ago
From what I could tell from the article and the linked videos the innovation here is that it essentially lets you serve the shuttlecock while it's facing the wrong direction. Normally even if the shuttlecock has spin when it crosses th
19.
by eklitzke 1y ago
Yeah if you were really trying to make things fast you'd have the compute and NFS server in the same rack connected this way. But you aren't going to get this from any cloud providers. For read-only data (the original model is abo
20.
by eklitzke 1y ago
NFS can be super fast, in a past life I had to work a lot with a large distributed system of NetApp Filers (hundreds of filers located around the globe) and they have a lot of fancy logic for doing doing locale-aware caching and clustering.
21.
by eklitzke 2y ago
This is vastly oversimplifying the problem, the difference between IPv4 and IPv6 is not just the format of the address. Different protocols have different features, which is why the sockaddr_in and sockaddr_in6 types don't just differ
22.
Agenoria(github.com)
1 points by eklitzke 2y ago|0 comments
23.
by eklitzke 2y ago
I don't understand why they say that reference counting is "slow". Slow compared to what? Atomic increments/decrements to integers are one of the fastest operations you can do on modern x86 and ARM hardware, and except i
24.
House committee approves bill requiring new cars to have AM radio(theverge.com)
59 points by eklitzke 2y ago|53 comments
25.
by eklitzke 2y ago
Just to expand on this, since it wasn't originally clear to me from reading your post, the contact privacy feature is about using SGX enclaves for the purpose of populating your known contacts on Signal. When you log into Signal for th
26.
Yes, there are more driverless Waymos in S.F(sfchronicle.com)
29 points by eklitzke 2y ago|55 comments
27.
ChromeOS to Migrate to Android-Based Tech Stack(chromeos.dev)
50 points by eklitzke 2y ago|9 comments
28.
GM-owned Cruise reached $8M settlement with dragged pedestrian(fortune.com)
1 points by eklitzke 2y ago|0 comments
29.
High-Functioning Anxiety Isn't a Medical Diagnosis. It's a Hashtag(nytimes.com)
1 points by eklitzke 2y ago|0 comments
30.
Self-driving cars are underhyped(slowboring.com)
4 points by eklitzke 2y ago|0 comments

More ›