madanMus

3 ms·

1. 1.▲
   by madanMus 11y ago

   yes, you did read correctly. The 1-3% does assume a non-standard whole-program analysis. Something more practical on existing hardware will look more like the E numbers (for escape analysis): 17.5% on x86 and 6.8% on ARM. An even dumber ana
2. 2.▲
   by madanMus 11y ago

   Yes, it does, provided programmers restrict themselves to the safe subset of the language. Rust is a wonderful step in the right direction.
3. 3.▲
   by madanMus 11y ago

   [I am one of the authors of the paper.] The paper reports overhead numbers from existing research. For instance, see Figure 18 in http://arxiv.org/abs/1312.1411 , which shows the cost of SC for memcached - 1% on x86 and
4. 4.▲
   by madanMus 11y ago

   But the math is currently broken for Java. See http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.112... . Here the authors show that most common compiler optimizations are JMM incompatible.
5. 5.▲
   by madanMus 11y ago

   The article does not ignore architectures like ARM. You do need fences, but not all the time - the compiler can avoid fences in places where there is no danger of violating sequential consistency (e.g. on accesses to provably local objects)