|
|
|
|
|
|
by Someone
1518 days ago
|
|
|
Some programs spend over 50% of their time doing reference counting (http://iacoma.cs.uiuc.edu/iacoma-papers/pact18.pdf), and that will show up I’m your throughput measurements. I expect you can find papers giving numbers in the same ballpark for malloc/free, too. A MVP of JavaScript, for example, that boxes everything, combined with a benchmark that creates a d destroys lots of objects. |
|
|