Anisse

@Aissen@treehouse.systems
689 Followers
415 Following
2.8K Posts

Linux dev. Android user. Security curious. Gopher hobbyist. Beginner rustacean. Casual hardcore gamer. — French native, english toots — Personal account — He/him.

Posts in CC-BY-ND for a month, then CC-BY-NC-ND, contact me for other arrangements.

Profile picture is a line drawing caricature by french artist Frank Tizzoni of myself; I have a french-style flat cap and a light beard.

websitehttps://anisse.astier.eu
Kernel Recipes 2022 livebloghttps://kernel-recipes.org/en/2022/category/live-blog/
Kernel Recipes 2023 livebloghttps://kernel-recipes.org/en/2023/category/live-blog/
Message me on Signalhttps://signal.me/#eu/y7Yv18foRXDoXscEQlKOEY5G899AwTr8Mlp0wO0eU3hN3pSM3JtEWJpwudNLxfXz
@cadey or the network of apps using residential network resources of wifi/charged smartphones could also start using the compute resources as well… At scale PoW against well funded unethical parties will be though.
I am reminded of this article: Beating the kCTF PoW with AVX512IFMA for $51k ; and now I wonder if Anubis can survive success: highly-specialized SIMD (or hw) implementations will either make the Proof of Work cost too low for scrapers or too high for mobile/slow computers users.
WDYT @cadey ? What is the end game?
Beating the kCTF PoW with AVX512IFMA for $51k

PoW is gone 🦀🦀

anematode
Repeating my occasional and yet desperate plea to all my nerdy friends - DO NOT PLAY WITH COMPILERS. It is a trap. You will write it forever, and it will never do what you want, nobody will ever use it, it will not be fulfilling, and you will waste your precious life. Please, for the love of the gods, find something more satisfying to do than scream at a parser and SSA/ASTs.

Nothing to see here: NVIDIA IR compiler gives better FP8 performance if the kernel name contains NVIDIA's `cutlass`…

The most charitable explanation is that there are undocumented optimizations paths… that only NVIDIA can use.

https://github.com/triton-lang/triton/pull/7298/commits/a5e23d8e7e64b8a11af3edc1705407d91084b01d

[Gluon][Tutorial] Persistent attention by Mogball · Pull Request #7298 · triton-lang/triton

Rewrite the attention kernel to be persistent. This gives better performance at low-contexts. However, fp16 at large context has suffered a bit due to a ptxas instruction scheduling issue in the so...

GitHub

Hey, #FediHire me! 🥳
I'm looking for a decent #job, preferably remote.
Any help is appreciated.

I bring a ton of skills and more than a decade of experience as a #software #engineer, have been working mostly on #web #applications and some #mobile over the years, and done lots of #systems programming and #firmware in my spare time projects. 🧑‍💻⚙️

My main languages are Rust and JavaScript/TypeScript.
I'm based in Germany.

More details and CV are at https://dan.orangecms.org/.

Please #boost, thanks!

Hi, I'm Dan!

I know, I know. "Blocking replies is impossible". "You can't stop people from disagreeing with you on the internet".

I've heard it all, and don't really care.

Here's my proposal: much like how I can set the post visibility here, I should be able to tell my server to drop any replies to that post.

Not "don't send notifications about replies". Drop them. Don't host them.

And compliant servers would honor that request in the UI.

Now, don't get me wrong, I don't want to host harassment and share it around. I get vile nightmare fuel you don't want to see.

Defederating is a good solution. But I feel strongly that Mastodon (as the putative leader pushing the protocol) could do two things.

One would be to make it easier to share naughty-lists full of bad actors in real time. This would help stop issues before they start.

The other is blocking replies.

@Aissen @gynvael You don't need to know the software: in the end, it's just a handful of parsing strategies, not a bug that is only available between 2 specific versions of a specific software.

So, just use a generator, use the files, and see the results.

Some files targeting gmail are more advanced than classic ambiguous or polyglot files.

I love this research on schizophrenic files (and polyglots in general), but I feel that the scenario in the quoted post is maybe too convoluted to happen in real life.

It needs lots of recon to know which zip software each person in the chain will be using; and at this point, spear-phishing and others become lower cost than crafting custom files with the associated QA.

cc @Ange @gynvael

https://infosec.exchange/users/gynvael/statuses/114783517659300005
×
@vbabka what did you expect? Something like this? 🙃
@kernellogger actually yes, some new funny name ;) good one!