Many GC strategies were tried before we settled on our current mix that seem to perform well in the current problem domains. I am sure once it is proved, someon

Many GC strategies were tried before we settled on our current mix that seem to perform well in the current problem domains. I am sure once it is proved, someone will build a faster way. Not to mention that Hardware Transactional Memory is coming the next generation of intel chips (Q2 2013) and like virtualization it might just be that hardware support is required to give it the boost required. Either way, I think the research is important because its based around a real language, with real production code. All too often CS research is based around idealized languages and thus all the code tends to be coded to the capabilities of the language, instead of the code that is often suboptimal, because the developer was under the gun to meet a deadline.