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

help-circle





  • The main reason I’m quitting where I currently work is because it’s impossible to do the most basic of tasks without having 40 meetings about it and it being delayed by 6 months for no apparent reason.

    Similar, one issue for me is that I am not great at all the modern task-tracking systems, I can do them, but they take far more energy for me than the actual work itself.

    And the more people I have to work with, the slower I work, I can just go by myself, but when I have to keep coordinating I just bog down and get discouraged.










  • Hah, no, but I’m good otherwise.

    Talking to friends about starting one of my own ideas, reached an age where digging through the dirt is still fun, but I think I should be doing more.

    Thanks for the concern :) , I rarely join startups because I think they have a chance to succeed, it’s just the only place you can get stuff done and have a good time, big companies are… Horrifying :( Google might have been the worst.

    I need to get off my ass really.




  • So, trying not to dox myself, I worked with the architect twice.

    Knights Ferry was derived directly from Larrabee (GPU), P54Cs with pre-AVX-512, .

    KNC was a die shrink with more cores. Both of these were PCIe accelerators only.

    KNL had full Airmont Atom cores with smt4, basically meaningful cores with proper AVX-512. Also you could boot them with linux, or as a PCIe accelerator.

    KNM jadded ML instructions, basically 8/16bit float and faster SIMD.

    They cancelled KNH.

    I interviewed some of the actual Larrabee guys, they were wild, there was a lot of talk about dynamic translation, they were trying to do really complex things, but when people talk like that it makes me think they were floundering on the software and just looking for tech magic solutions to fundamental problems.

    Intel always dies when the software gets more complex than really simple drivers, it’s their achilles heel.

    KNL also had the whole MCDRAM on package for basically HBM bandwidth, but that didn’t actually work very well in practice, again due to software issues (you have to pick where you allocate, and using it as an l4 cache was not always effective).