jpecar

I'm baffled by the fact that NUMA is still an issue in 2026. My impression is that this was all solved back in dotcom era already on those big SUNs. At least in HPC we solved this already in mid 2000s. Why is supposedly modern world still wasting time on this? Kernel these days exposes just about everything you would ever want to know about a system topology and every runtime should be making use of that information. If it does not, I cannot consider it ready for this century.

Cool (literally ;)

Now they just need to find something that will work on Venus.

To add on this, what I see gaining traction are "workflow managers", tools that let people specify flow of data through various tools. These can figure out how to parallelize things on their own so users are not burdened with this task.

So from what I see actual programming language doesn't matter as much as how the work is organized. Anything helping people simplify this task is of immediate benefit to the science.

All these fancy HPC languages are all nice and dandy, but the hard reality I see on our cluster is that most of the work is done in Python, R and even Perl and awk. MPI barely reached us and people still prefer huge single machines to proper distributed computing. Yeah, bioinformatics is from another planet.

DB of known proteins is not where the money can be made, designing new proteins is. This is why AlphaFold3 (that can aid in this) is now wrapped in layers of legalese preventing you to actually use it in the way you want. At least that's what my lifescience users tell me. Big Pharma is now paying Big Money to DeepMind to make use of AF3 ...

Servers? I thought they left even racks behind, they're now selling these "AI factories".

Afaik Intel's first foray into this territory was their i960mx which ended up in F-22.

In performance per watt? One ;)

Wow, google suddenly looks much better. How do we make this change permanent? (not a joke)

VR situation was much worse. On the hardware where 2020 was OKish, 2024 was unplayable.

Yeah ... by now 2020 reached a beta quality, still plenty of bugs left to squash but most of them are minor. 2024 on the other side ... is just transitioning from tech demo to alpha quality. It needs a few more years to reach beta and then by 2035 or so it might actually become a reliable product.

Been using lvm cache since about 2017 in high performance storage environment. Didn't like it much as it was too much of a blask box back then and didn't have any knobs to tune to adjust its behavior to the workload and we had plenty that didn't fit its logic. Ended up going full flash.

Hedgehog's dilema. Interacting today with random average human being leads in 99% to such a painful disappointing conclusion that I got PTSD from it. Just being within a line of sight of another human being makes me nervous and looking for a place to hide.

Reminds me of ACME phoon ... I used to have this in my .bashrc ;) https://acme.com/software/phoon/

And if you want an anime series that you can classify like this, check out "Now and then, here and there".

And how does that lead to colorful flashing pattern that "collapses" some 30 seconds after it appears?

Yes, this makes magic on real hardware and I see it do nothing on emulators. So something really funny must be going on here ...

I still don't get what

POKE 23641,194

RUN

does on zx spectrum. And you get different patterns for different values, say 197. And wait for a minute or so to see something even more interesting.

Why not just call it swamp? It would better describe the python ecosystem mess ;)

Jokes aside, this feels very meta: package manager for a package manager for a package manager. Reminds me of the old RFC1925: "you can always add another layer of abstraction". That RFC also says "perfection has been reached not when there is nothing left to add, but when there is nothing left to take away".

And as a hpc admin, I'm not offering my users any help with conda and let them suffer on their own. Instead I'm showing them the greener pastures with spack, easybuild and eessi whenever I can. And they're slowly migrating over.

Indeed, sound mix is rather amazing. Incredible stereo picture and depth, something you seldom hear in modern movies.

Is D runtime still crashing when host has more than 128 cpu cores? I learned this the hard way ...

Gives nice retro vibes ... mail like we were doing it back at the end of last millenium. Proper old school :)

Soooooo happy that my mail client is not even on the list here. It automatically strips out anything remotely looking like html and it sends out plain text emails only, as email was designed to be and should still be.

Anyone sending me css garbage will be directed to /dev/null. Thanks.

Very well in fact, you have to look at the whole drivetrain efficiency, not just the motor. Elaphe has a webinar on this topic on their youtube channel, look it up.

Impact was a concept car designed by AeroVironment and GM eventually developed it into EV1. There are very few infos online today on Impact, is anyone aware of some decent writeup or pics/videos of it?