• 0 Posts
  • 23 Comments
Joined 1 year ago
cake
Cake day: June 29th, 2023

help-circle








  • Man I just built a new rig last November and went with nvidia specifically to run some niche scientific computing software that only targets CUDA. It took a bit of effort to get it to play nice, but it at least runs pretty well. Unfortunately, now I’m trying to update to KDE6 and play games and boy howdy are there graphics glitches. I really wish HPC academics would ditch CUDA for GPU acceleration, and maybe ifort + mkl while they’re at it.


  • So many solver solutions that day, either Z3 or Gauss-Jordan lol. I got a little obsessed about doing it without solvers or (god forbid) manually solving the system and eventually found a relatively simple way to find the intersection with just lines and planes:

    1. Translate all hailstones and their velocities to a reference frame in which one stone is stationary at 0,0,0 (origin).
    2. Take another arbitrary hailstone (A) and cross its (rereferenced) velocity and position vectors. This gives the normal vector of a plane containing the origin and the trajectory of A, both of which the thrown stone must intersect. So, the trajectory of the thrown stone lies in that plane somewhere.
    3. Take two more arbitrary hailstones B and C and find the points and times that they intersect the plane. The thrown stone must strike B and C at those points, so those points are coordinates on the line representing the thrown stone. The velocity of the thrown stone is calculated by dividing the displacement between the two points by the difference of the time points of the intersections.
    4. Use the velocity of the thrown stone and the time and position info the intersection of B or C to determine the position of the thrown stone at t = 0
    5. Translate that position and velocity back to the original reference frame.

    It’s a suboptimal solution in that it uses 4 hailstones instead of the theoretical minimum of 3, but was a lot easier to wrap my head around. Incidentally, it is not too hard to adapt the above algorithm to not need C (i.e., to use only 3 hailstones) by using line intersections. Such a solution is not much more complicated than what I gave and still has a simple geometric interpretation, but I’ll leave that as an exercise for the reader :)



  • Didn’t realize this was happening and yay -Syu went brrr and it broke my shit. Probably doesn’t help that I’m running nvidia with linux (endeavouros). Wayland doesn’t work at all (black screen on login with only mouse ptr, wrong resolution), while Xorg is now much less smooth e.g. on the switching desktop animations. Moving windows around and in-window graphics are fine. Some graphical config stuff changed too; I’m still taking inventory.

    I’m also currently playing with nvidia vs nvidia-dkms with different kernels to see if that solves anything.

    EDIT: Looks like that my configuration was failing to set nvidia_drm modeset=1 correctly due to my unfamiliarity with dracut. Manually adding nvidia_drm.modeset=1 to my kernel cmdline makes Wayland work (and quite well at that), though Xorg is still laggy.







  • I think water is rather rare as a coolant these days. Organics (chemical sense not farming sense) like propylene glycol or some kind of glyme aren’t potentially corrosive to metals if spilled, are harder to grow shit in, have lower volatility, and have a higher thermal limit. Maybe also with a little bit of antifouling agent thrown in. My main gripe with them is that if you do spill them, they don’t evaporate and you’re slipping over the floor for the next few days because you missed a spot.

    But yeah, air cooling ftw



  • ornery_chemist@mander.xyztoMemes@lemmy.mlThe reason
    link
    fedilink
    arrow-up
    4
    arrow-down
    1
    ·
    edit-2
    1 year ago

    I think rz is linguistically equivalent to a soft r, so in this case rze would be “ре”, not “ж”. In some areas, rz is pronounced closer to the Czech ř. IIRC, ж transliterates to ż (not to be confused with ź, which is a soft z). The Polish Roman alphabet is very regular and well adapted to the language, representing palatalization and other non-Latin sounds as digraphs in a similar way to Italian or English.

    The cyrillicization of Polish was historically done to a limited extent, but carried with it some, shall we say, sociopolitical baggage. There are also some peculiarities to Polish that either don’t exist or have ambiguous transliterations into Cyrillic, such as the Polish nasals ą and ę or ó (historically a long o, but currenly pronouned /u/).