3 ms·

Eh, I get it. I also independently came up with these MPMC approaches before hearing about LMAX. They're not entirely trivial but they do follow fairly naturall

by atq2119 10mo ago

Eh, I get it. I also independently came up with these MPMC approaches before hearing about LMAX. They're not entirely trivial but they do follow fairly naturally from the problem space when you really think about it. It's a good piece of engineering, but the one thing that's really unique about LMAX is the amount of advertising it gets.

bob1029 10mo ago
> the one thing that's really unique about LMAX is the amount of advertising it gets. Virtually zero? I have to go out of my way to remind HN it exists while everyone is 300+ comments deep into reinventing the wheel on a ~quarterly basis.
- dpc_01234 10mo ago
  Right buffer is a relative obvious and simple idea. For some reason the Java-OOP crowd keeps thinking that LMAX deserves a nobel price for being neither first nor last to use it. > have to go out of my way Yeah, that's exactly the annoying part. Can't mention ring buffer ever without someone bring up LMAX. "But do you know, that some Java developers somewhere once wrote something that didn't completely ruin the hardware performance?! It's stunning and amazing."
  - bob1029 10mo ago
    I am not sure why this is annoying. Some problems can be solved comprehensively. This is a pretty good example of one. It might be better for us to focus our attention on other aspects of the problem space (the customer/business).
  - immibis 10mo ago
    IMO the take-away from LMAX is not ring buffers - it's the knowledge of how much useful work a single CPU core can do. It's a story of playing to hardware's strengths instead of wrapping yourself up in bullshit excuses. They realized their problem was fundamentally not parallelizable, so they wrote it to run serially as fast as possible instead of wrapping themselves up in bullshit excuses, and the resulting performance was much faster than anyone would have ever guessed if they hadn't done it.

by atq2119 10mo ago

bob1029 10mo ago

> the one thing that's really unique about LMAX is the amount of advertising it gets. Virtually zero? I have to go out of my way to remind HN it exists while everyone is 300+ comments deep into reinventing the wheel on a ~quarterly basis.

dpc_01234 10mo ago
Right buffer is a relative obvious and simple idea. For some reason the Java-OOP crowd keeps thinking that LMAX deserves a nobel price for being neither first nor last to use it. > have to go out of my way Yeah, that's exactly the annoying part. Can't mention ring buffer ever without someone bring up LMAX. "But do you know, that some Java developers somewhere once wrote something that didn't completely ruin the hardware performance?! It's stunning and amazing."
- bob1029 10mo ago
  I am not sure why this is annoying. Some problems can be solved comprehensively. This is a pretty good example of one. It might be better for us to focus our attention on other aspects of the problem space (the customer/business).
- immibis 10mo ago
  IMO the take-away from LMAX is not ring buffers - it's the knowledge of how much useful work a single CPU core can do. It's a story of playing to hardware's strengths instead of wrapping yourself up in bullshit excuses. They realized their problem was fundamentally not parallelizable, so they wrote it to run serially as fast as possible instead of wrapping themselves up in bullshit excuses, and the resulting performance was much faster than anyone would have ever guessed if they hadn't done it.