By combining asynchronous I/O with a shared-nothing architecture, PyParallel research project is able to execute code in a parallel context faster than it can using CPython’s normal interpreter. And it does this without removing the GIL. The secret, no reference counting or garbage collection of any kind.
We interviewed Joe Duffy, author of Concurrent Programming on Windows, about his research into the use of type systems to ensure safe parallelism. This work was presented in the paper titled Uniqueness and Reference Immutability for Safe Parallelism. We asked for this interview because there seemed to be a lot of misconceptions about this research project.
InfoQ spoke with book author Venkat Subramaniam about strategies and design approaches for programming concurrency on JVM and hardware capabilities to achieve concurrency. 4
Joe Duffy talks about the future of concurrency and parallelism. This interview covers his thoughts on the language designs, libraries, and patterns that are becoming important in modern programming. 1
In this IEEE article, authors Frank Feinbube, Peter Troger and Andreas Polze discuss two major hardware trends in parallel programming space, multi-core CPU architectures and GPUs.
IBM’s Yao Qi, Raja Das, and Zhi Da Luo describe jucprofiler, an alphaWorks tool designed to profile multicore applications that make use of the java.util.concurrent classes introduced in Java 5. 2
Jeroen Borgers examines if biased locking, lock coarsening, lock elision by escape analysis and adaptive spin locking techniques in the latest JVMs actually work in highly-concurrent apps. 3
Jeroen Borgers examines if biased locking, lock coarsening, lock elision by escape analysis and adaptive spin locking techniques in the latest JVMs actually work in highly-concurrent apps. 16