Comments by "Toby" (@toby9999) on "Creel" channel.

  1. Yeah, but x86 is also a common label given to mean the Intel CPU rather than being about specific subgroups of the supported instruction set.
    1
  2. LOL, it won't.
    1
  3. They sure were, and I loved both of then. When I eventually saw the x86 spec, I said no way. Ugly.
    1
  4. I have heard if it and I have used it in chess engines where every clock cycle counts. For instance, bitboards and branchless algorthms.
    1
  5. Nope.
    1
  6. You're stating the bleeding obvious, which is to not optimise too early. But what is too early? The later one leaves it the harder it becomes, and the easier becomes to introduce bugs. When developing high performance software, optimisation should be in the forefront of one's mind before writing any code. It should be part of the design process.
    1
  7. ​​ @robegatt That's not true. You actually have it all backwards.
    1
  8. I doubt Python does any optimisations given how slow it runs.
    1
  9. It's not possible to represent an infinite number of floating point values into finite resources (which you likely know already). There'll always be a compromise.
    1
  10. Why?
    1
  11. It's personal opinion.
    1
  12. What bad advice? The concepts presented here are very good, and they have been used in high-performance software for decades. I've used them in my chess engine, as have others, and I'm using them in a compiler project I'm working on for a propriatory language. Roughly a 50% performance increase across a set of 20,000 unit tests just by eliminating 'if' statements. Perhaps your comment was sarcasm, but yes, there's plenty of bad advice, but the bad advice is all in the comments.
    1
  13. Why would that be "lucky"?
    1
  14. No.
    1
  15. Yeah, the 68000 instruction set was way cleaner. Avtually very nice.
    1