On 3/28/2011 11:49 AM, dsimcha wrote:
Correct me if I'm wrong, but I think some important optimizations (like inlining) are performed in the front end. It's pretty obvious that DMD's inliner needs improvement, though I agree with Walter's decision to prioritize this below fixing major bugs, 64 support and (before that) finalizing the spec. Therefore, the performance of GDC is still a quality of implementation issue rather than a fundamental technical issue.
By fundamental technical issue, I mean things like Python's numeric types which require runtime testing for every operation, and are very resistant to known techniques of optimization.
D's perf problems have known causes that are not faults in the language, but are fixable in the compiler using known techniques. It's just a matter of priorities and manpower.
In fact, D has many attributes that offer the potential for significantly better code generation than C/C++. They've just not been exploited yet.
