My kernels go 2x faster than MKL for matrices that fit in L2 cache, which makes them a work in progress, since the speedup works best for prompts having fewer than 1,000 tokens.
I completely missed that user namespaces were added in 1.25. It will make homelabs much easier and safer with little effort.
Support user namespaces in pods (KEP-127)
User namespaces is a Linux-only feature that better isolates pods to prevent or mitigate several CVEs rated high/critical, including CVE-2024-21626, published in January 2024. In Kubernetes 1.30, support for user namespaces is migrating to beta and now supports pods with and without volumes, custom UID/GID ranges, and more!
“Prolly Tree” is short for “Probabilistic B-tree”. “Prolly Tree” was coined by the good folks who built Noms, who as far as we can tell invented the data structure. We here at DoltHub have immense respect for their pioneering work, without which Dolt would not exist.
I think you’re biased against Java. Amazon was started in C/C++ and Java J2EE during times when to configure a webserver required writing like 300 lines of XML just to handle cookies, browser cache and a login page. Until recently BMW had their own JRE implementation. It’s not a secret that simcards, including these in Tesla cars run JavaCard too, even government issues sim cards in EU have to run Java Card, not C++. Everything was always fine with Java until ECMA Script appeared and made people iterate on software versions faster. New programming languages and team organisation methodologies left some programming languages in the dark, but this included C# too. All are quickly catching up. If Java was so bad, it wouldn’t be here with us today, like Perl.
As someone who spends time programming, I of course find myself in conversations with people who aren’t as familiar with it. It doesn’t happen all the time, but these discussions can lead to people coming up with some pretty wild misconceptions about what programming is and what programmers do....
I feel like lately a lot of job boards have become populated with scams, training courses disguised as jobs and the few jobs that are posted are just posted for show. Indeed.com is a good example as it used to be decent some years ago and now it’s hard to find an actual listed job. Linkedin is okay so far....
A protocol for peer-to-peer data stores. The best parts? Fine-grained permissions, a keen approach to privacy, destructive edits, and a dainty bandwidth and memory footprint.
So while I’m myself struggling to fully understand what this is, it conceptually like it’s a blockchain on syncthing, where even if you subscribe to a read only share, you can locally delete what you don’t want to keep. So technically you could make bitorrent to behave like syncthing with search function for contacts you already know.
It’s about asking, “how does this algorithm behave when the number of elements is significantly large compared to when the number of elements is orders of magnitude larger?”...
Big O notation is useless for smaller sets of data. Sometimes it’s worse than useless, it’s misguiding.
I don’t agree that it’s useless or misguiding. The smaller dataset, the less important it is, but it makes massive difference how the rest of the algorithm will be working and changing context around it.
Let’s say that you need to sort 64 ints, in a code that starts our operating system. You need to sort it once per boot, and you boot less frequently than once per day, in fact you know instances of the OS that have 14 years of uptime, so it doesn’t matter at all right? Welp. Now your OS is used by a big cloud provider and they use that code to boot the kernel 13 billions times per day. The context changed, time passed by, your silly bubble sort that doesn’t matter on small numbers is still there.
Fantastic way to start a shitstorm. You people don’t even use search function logged out, because if you did, you would know they changed it in 2016. Microsoft has nothing to do with it.
No Restarts, No Disruptions: Seamless Pod Resource updates with In-Place Resizing (engineering.doit.com)
Microsoft opens a "high priority" bug ticket in ffmpeg, attempting to leech the free labour of the maintainers (trac.ffmpeg.org)
Microsoft employee:...
LLaMA Now Goes Faster on CPUs (justine.lol)
My kernels go 2x faster than MKL for matrices that fit in L2 cache, which makes them a work in progress, since the speedup works best for prompts having fewer than 1,000 tokens.
A Peek at Kubernetes v1.30 (kubernetes.io)
More powerful Go execution traces - The Go Programming Language (go.dev)
Using CRaC to reduce Java startup times on Amazon EKS | Amazon Web Services (aws.amazon.com)
JDK 21 G1/Parallel/Serial GC changes (tschatzl.github.io)
JDK 22 G1/Parallel/Serial GC changes (tschatzl.github.io)
"Prolly Tree" is short for "Probabilistic B-tree" (www.dolthub.com)
“Prolly Tree” is short for “Probabilistic B-tree”. “Prolly Tree” was coined by the good folks who built Noms, who as far as we can tell invented the data structure. We here at DoltHub have immense respect for their pioneering work, without which Dolt would not exist.
Async File IO (concurrencydeepdives.com)
Perf is not enough (motherduck.com)
Defcon: Preventing Overload with Graceful Feature Degradation (www.micahlerner.com)
deleted_by_author
What are the craziest misconceptions you’ve heard about programming from people not familiar with it?
As someone who spends time programming, I of course find myself in conversations with people who aren’t as familiar with it. It doesn’t happen all the time, but these discussions can lead to people coming up with some pretty wild misconceptions about what programming is and what programmers do....
UUID Benchmark War (ardentperf.com)
What are some actual good job boards in 2024?
I feel like lately a lot of job boards have become populated with scams, training courses disguised as jobs and the few jobs that are posted are just posted for show. Indeed.com is a good example as it used to be decent some years ago and now it’s hard to find an actual listed job. Linkedin is okay so far....
Willow Protocol (willowprotocol.org)
A protocol for peer-to-peer data stores. The best parts? Fine-grained permissions, a keen approach to privacy, destructive edits, and a dainty bandwidth and memory footprint.
JDK HTTP server handles 100,000 req/sec with 100 ms start-up time and 50 MB modular run-time image. Built with OpenJDK 21 and virtual threads (github.com)
95%-ile isn't that good (danluu.com)
Big O notation is about what matters when the numbers get big. (programming.dev)
It’s about asking, “how does this algorithm behave when the number of elements is significantly large compared to when the number of elements is orders of magnitude larger?”...
Things to say when you're losing a technical argument (web.archive.org)
GitHub: Can no longer search code without being logged in (github.com)
Edit:...