Online Adaptive Code Generation and Tuning
2011pp. 879–892
Citations Over TimeTop 10% of 2011 papers
Abstract
In this paper, we present a runtime compilation and tuning framework for parallel programs. We extend our prior work on our auto-tuner, Active Harmony, for tunable parameters that require code generation (for example, different unroll factors). For such parameters, our auto-tuner generates and compiles new code on-the-fly. Effectively, we merge traditional feedback directed optimization and just-in-time compilation. We show that our system can leverage available parallelism in today's HPC platforms by evaluating different code-variants on different nodes simultaneously. We evaluate our system on two parallel applications and show that our system can improve runtime execution by up to 46% compared to the original version of the program.
Related Papers
- → LLVM-Based Code Generation for B(2015)6 cited
- → Extraction of Efficient Instruction Schedulers from Cycle-True Processor Models(2003)3 cited
- → Some design aspects for VLIW architectures exploiting fine-grained parallelism(1993)4 cited
- → URPR-1: A single-chip VLIW architecture(1993)3 cited
- → Retargetable code generation for application-specific processors(2004)