Personal tools
You are here: Home Events ICSA Colloquium Talk

ICSA Colloquium Talk

— filed under:

Presentation by Igor Bohm. Title: Efficiently Exploiting Parallelism in a Just-In-Time Compiler using Generalised Just-In-Time Trace Compilation based on the Parallel Task Farm Design Pattern.

  • Colloquium Series
When Mar 03, 2011
from 03:30 PM to 04:30 PM
Where G07a
Add event to calendar vCal


In any kind of Just-In-Time (JIT) compiler the main challenge is translating frequently executed program regions as fast as possible into highly efficient native code. As time for JIT compilation adds to the overall execution time, the JIT compiler is often decoupled and operates in a separate thread independent from the main simulation loop to reduce the overhead of JIT compilation.


In this talk I want to present three innovative contributions. The first contribution is a generalised trace compilation approach that considers all frequently executed paths in a program for JIT compilation, as opposed to previous approaches where trace compilation is restricted to paths through loops. The second contribution reduces JIT compilation cost by compiling several hot traces in a concurrent task farm. The third contribution extends our approach to multi-threaded applications. Altogether we combine generalised light-weight tracing, large translation units, parallel JIT compilation and dynamic work scheduling to ensure timely and efficient processing of hot traces.


We have implemented and evaluated our generalised trace compiler using a parallel JIT compilation task farm in ArcSim - our industry-strength instruction set simulator. Using more than 60 industry standard benchmarks from various domains we demonstrate speedups of up to 2.08 on a standard quad-core machine. Across short- and long-running benchmarks our scheme is robust and never results in a slowdown. 

In fact, using four processors total execution time can be reduced by on average 11.5% over state-of-the-art decoupled, parallel (or asynchronous) JIT compilation. Results from our latest research where we applied our technique to a multi-core processor simulator version of ArcSim show simulation rates of up to 25,307 MIPS for standard parallel benchmarks.


Finally, I want to show two live demos to showcase the capability and successful application of our technology to real world applications ranging from video decoding and playback using Google's latest video codec WebM, and the simulation of many-core processors.



Bio: Igor Bohm is a PhD candidate at ICSA in his final year. His research focuses on trace-based parallel Just-In-Time compilation as an enabling technology for ultra-high speed processor simulation, virtual machines and dynamic programming language interpreters.




Document Actions