There are various tools you can use, and using a screwdriver instead of a hammer won't help you much if what you have is nails.
From my point of view, you may get back to me when Lush has data structures for concurrent programming and does (nearly) pauseless GC on multicore machines with heaps of 12GB composed of complex data structures, not just vectors of numbers. This is what I get with Clojure and for my applications it works great. Oh, we also do numerics, quite a bit in fact. But whenever I need actual raw-metal double-adding performance, I use an appropriate tool, e.g. an extension/library. For certain string operations we even went down to x86-64 assembly, with great results.
What you describe is a JVM limitation, not a Clojure limitation — and you're right in only one thing: Clojure is not the right tool for you.
Digression here, but I like the "(nearly)". This bit always amuses me about garbage collection wonks. The pauseless bit is a real time requirement. Saying your GC has great latencies or is "(nearly) pauseless" is tantamount to telling a real time engineer your system only fails some of the time. It makes you look dumb.
GC is great. GC makes a ton of things simpler. GC as implemented in popular environments still sucks for real time use.
The 'pauseless' bit is not just a real-time requirement. A website that stalls for 2 seconds once every 10 minutes is unacceptable to me. That application thus requires a garbage collector that is nearly pauseless: the pauses are small enough that your users won't notice them. So I don't think it makes anyone look dumb at all to speak of a 'nearly pauseless' garbage collector.
There are various tools you can use, and using a screwdriver instead of a hammer won't help you much if what you have is nails.
From my point of view, you may get back to me when Lush has data structures for concurrent programming and does (nearly) pauseless GC on multicore machines with heaps of 12GB composed of complex data structures, not just vectors of numbers. This is what I get with Clojure and for my applications it works great. Oh, we also do numerics, quite a bit in fact. But whenever I need actual raw-metal double-adding performance, I use an appropriate tool, e.g. an extension/library. For certain string operations we even went down to x86-64 assembly, with great results.
What you describe is a JVM limitation, not a Clojure limitation — and you're right in only one thing: Clojure is not the right tool for you.