dumblob

Btw recently released version of Ruby uses yet better GC algorithm. I think it's worth reading to get some inspiration: https://bugs.ruby-lang.org/issues/10137 .

Oops, I somehow missed it's incremental. Great news :)

> Besides the locks, the only field that may be accessed frequently by mutator threads is the `idleList` field. We can try to check the impact of false sharing by...

I realized, that guys from Go finally decided to speed up their GC and make it "predictable" in terms of latency. Their plan (already half-realized in the current Go 1.4)...

To make matters worse, I realized, that we don't use the C99 keyword `restrict` anywhere in the Dao source code. This might make a huge difference (an order of magnitude...

Today I read some details about Ulterior reference counting and it sounds really attractive in the matter of mutations being the bottle neck. URC is also described in _The Garbage...

On new x86 and x86_64 CPUs (Core 2 Duo, i3, i5), we might gain at least 5x speed-up of repeated division (e.g. tight loops) by using http://libdivide.com/ instead of hardware...

Following up on GCC optimization, there seems to be an [unpleasant bug](https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77287) in GCC - namely it doesn't at all produce good code for argument passing to functions (in comparison...

Another good optimization is [Profile-guided optimization](https://en.wikipedia.org/wiki/Profile-guided_optimization ) (I saw 7% performance gain for database applications in 2009; in a VM I expect definitely a higher percentage - e.g. in [Python](http://www.activestate.com/blog/2014/06/python-performance-boost-using-profile-guided-optimization...

What an inspirating discussion and I couldn't participate due to my time constraints. I've never used AOP directly in production code (only indirectly by using e.g. Enterprise Java Beans) -...