> Meanwhile in just about every other area that uses fp computation it's been the defacto standard for decades.
Not that strongly for more parallel things, quite similar to the situation with atomics on cuDNN. cuBLAS for example has a similar issue with multi-stream handling, though this can be overcome with a proper workspace allocation: https://docs.nvidia.com/cuda/cublas/index.html?highlight=Rep....
Still better than cuDNN where some operations just don't have a reproducible version though. The other fields are at least trying. DL doesn't seem to be.
Fil-C works because you recompile the whole C userspace. Unsafe Rust doesn't do that... and for many practical purposes you probably want to touch the non-safe-version of the C userspace.
Still, it's all LLVM, so perhaps unsafe Rust for Fil-space can be a thing, a useful one for catching (what would be) UBs even [Fil-C defines everything, so no UBs, but I'm assuming you want to eventually run it outside of Fil-space].
Now I actually wonder if Fil-C has an escape hatch somewhere for syscalls that it does not understand etc. Well it doesn't do inline assembly, so I shouldn't expect much... I wonder how far one needs to extend the asm clobber syntax for it to remotely come close to working.
HTTPS is there, so you go down to that level only if you want to distrust any element of the public key infrastructure. Which, to be fair, there are plenty of reasons if you are paranoid -- they do tell you who's doing what in a shady way as they revoke, so there's a huge list of transgressions.
AM404 really is the metabolite that keeps giving. The central nervous system system effects were fun too, acting on cannabinoid receptors and/or TRPV1 channels -- so either stony or spicy.
From my limited experience of metamizole it feels a bit stronger than paracetamol/acetaminophen. Neat little drug if your genetics can take it.
Tangential: China technically banned metamizole due to the agranulocytosis scare, but somehow small clinics always have fresh stocks of this stuff. And their stocks don't look like my metamizole for horses! It's pressed out of the usual magnesium stearate instead of whatever rock-hard thing they use for animal drugs in China.
Not a professional, but still use it like that. They also have a new smartblast thing, which works much faster (really, really like Google!) but only on highly similar proteins.
on one hand i can see how it's going to break stuff (and I see they've changed it! []). on the other hand... mind talking a bit more about how it did break things?
There's some work in looking at alternative ways of administering the BCG vaccine (iv injection, inhalation). The IV numbers are pretty promising -- if the monkey numbers hold. I hope that someone, somewhere pushes it into human trials soon.
There's also the nebulous but very nice-sounding "non-specific effect" with it. Who knows.
We do know that for the lucky ones on which SSRIs do work, it seems to reverse some of the brain structure changes associated with depression -- some abnormally small parts tend to grow thicker. The reversal can be seen at one week after the start of medication use, before one can even tell whether an antidepressant has started working (a commonly-cited duration would be 2 weeks).
Lenore’s approach uses rejection sampling, which translates to needing an uncertain amount of input — a uint128 may or may not be enough.
But there is one related algorithm that doesn’t: https://github.com/apple/swift/pull/39143. This method only requires (output word size + 64) bits, which is really convenient for us since we probably don’t have 2^64 buckets. Same ballpark as (hash >> 64) * n, just with bias correction.
With this method the output value will be very not-scrambled relative to the high bits of the hash. Now whether that’s a problem is… someone else’s question. One can always do some shifts and XORs.
(1) changing the RSA decrypt function in OpenSSH is all the code hidden in crc64 does: that's the only known behavior, but we don't know what the changed function does besides letting some authentication through, nor do we know if there are other things it does
(2) there's no malicious machine in your LAN exploiting the RSA decrypt to log onto your sshd: nobody has seen one yet, but it doesn't mean there's no such thing.
If you are not using a distro that does dpkg or rpm, or if your machine is not x86-64, you're free from the "code hidden in crc64", the one that targets sshd, CVE-2024-3094. Are there unknown backdoors? Who knows. Do we count the landlock sabotage as a backdoor?
It's hard to deal with unknowns. Assume the worst, maybe, but what even is the worst?
The retaliation is quite real—CF keeps your entire e-mail address and name in there, so you are essentially doxxing yourself. Pretty sure 8chan posted a lot of the reports they got back in the day.
Not that strongly for more parallel things, quite similar to the situation with atomics on cuDNN. cuBLAS for example has a similar issue with multi-stream handling, though this can be overcome with a proper workspace allocation: https://docs.nvidia.com/cuda/cublas/index.html?highlight=Rep....
Still better than cuDNN where some operations just don't have a reproducible version though. The other fields are at least trying. DL doesn't seem to be.
On that note Intel added reproducible BLAS to oneMKL on CPU and GPU last year. https://www.intel.com/content/www/us/en/developer/archive/tr...