Thanks for this presentation! The range of topics covered and the level of explanation (vs assumed viewer knowledge) worked out quite perfectly for my needs. Pretty much exactly what I was looking for. Cheers!
Great talk, but I would not replace multiplication with addition all the time. Loop carried dependencies are bad for compilers and CPUs. In case of integers it's usually no big deal because the latency of add is low, but for floating point I would think twice. Usually, multiplication is better for precision as well, because repetitive additions tend to accumulate rounding errors.
![Lock-free programming with modern C++ - Timur Doumler [ACCU 2017]](http://i.ytimg.com/vi/qdrp6k4rcP4/mqdefault.jpg)
![Lock-free programming with modern C++ - Timur Doumler [ACCU 2017]](/img/tr.png)
![Bluff your way in x64 assembler - Roger Orr [ACCU 2017]](http://i.ytimg.com/vi/RI7VL-g6J7g/mqdefault.jpg)
Thanks for this presentation! The range of topics covered and the level of explanation (vs assumed viewer knowledge) worked out quite perfectly for my needs. Pretty much exactly what I was looking for. Cheers!
Great talk, but I would not replace multiplication with addition all the time. Loop carried dependencies are bad for compilers and CPUs. In case of integers it's usually no big deal because the latency of add is low, but for floating point I would think twice. Usually, multiplication is better for precision as well, because repetitive additions tend to accumulate rounding errors.
Baguette