PB's simple optimisations maybe a saving grace

[Kruno]

http://channel9.msdn.com/Events/Build/2014/4-587

It is a talk about compiler optimisations and how next generation CPUs are able to see further into the code than before, thus allowing certain optimisations to get in the way of a faster program.

I remember every so often Fred bringing up the topic of complexity of modern day CPUs in regard to optimisation, usually whenever someone is bothering him about it , and I thought this may be interesting to the people here.

On another note, I've been building more tools for my game, it has been jugging along nicely.

[Thorium]

Thats true, i also dont do heavy optimization because it may run faster on my CPU but slower on another. However PureBasic could realy benefit from SIMD. Some PB SIMD commands would be nice. Also PB could use the additional registers for preserving and stuff.

One example of breaking the performance is using self modifieing code. It was been used in the old days to speed up some loops. Today it totaly breaks the instruction prefetching and loop optimization build in the CPU. If the CPU detects self modifieing code it deactivates all loop optimizations resulting in slower code.

[davido]

Interesting. Thanks.

[coco2]

Interesting. Also what I like about PB is how fast the IDE loads up. I can open and close PB about 8 times while Codeblocks is loading up