On 12.01.2009, at 06:41, Mark P wrote: > 1. Some of the algorithms I use have the potential > to be parallelized. I am hoping that as the number > of cores in PCs increase, at some point Clojure's > performance will beat C++'s due to Clojure's > superior utilization of multiple cores. (Any ideas > on how many cores are needed for this to become > true?)
It all depends on your algorithms and your code. Clojure has lots of support (data structures and algorithms) for concurrent programming, but you have to choose and combine them yourself to get efficient parallelization. > 2. The JVM is continually being improved. Hopefully > in a year or two, the performance of HotSpot will be > closer to that of C++. (But maybe this is just > wishful thinking.) Again this depends a lot on what you are doing. Hotspot is already doing an impressive job on some applications, but remains bad at others. Optimization of a JIT compiler is quite different from optimization by hand: the compiler knows nothing about your algorithm at large that could help it to speed up execution, but on the other hand it knows where optimization is really useful and it can create specialized code versions that no sane person would ever attempt to write by hand. > 3. Maybe I can implement certain performance critical > components in C++ via the JNI. (But I get the impression > that JNI itself isn't particularly efficient. Also, the more > I pull over into the C++ side, the fewer advantages to > using Clojure.) I also wish there were a better interface between the Java world and the C world than JNI. Perhaps there is? I'd happily trade some safety for performance. One source of hope for better interfaces is new virtual machines. One candidate is VMKit (http://vmkit.llvm.org/), an implementation of the JVM (and .Net as well) on top of LLVM. Combine this with the gcc version that produces LLVM code, and it should be possible to get Java-C interoperability with much less friction than through the JNI. On the other hand, it will be a long time before VMKit matches the performance of HotSpot. > If all else fails, maybe I could use Clojure as a prototyping > language. Then when I get it right, I code up the actual > programs in C++. But probably a lot would get lost in > the translation from Clojure -> C++ so would it be worth > it? You could also consider writing Clojure code that generates C++ code. Or C, or why not directly LLVM. > I'd love to be convinced that Clojure is a viable choice, > but I need to be a realist too. So what do people think? I am in a very similar situation if I ever want to use Clojure in my professional environment. For the moment I have decided that Clojure is fun enough to explore even if I never will use it professionally, so it's not an immediate concern. Time will tell what can be done realistically. Konrad. --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com To unsubscribe from this group, send email to clojure+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/clojure?hl=en -~----------~----~----~----~------~----~------~--~---
