SLIDE 1
Virtual Machine and Bytecode for Optimization on Heterogeneous Systems
Kerry A. Seitz, Jr. and Mark C. Lewis Trinity University, San Antonio, TX
Optimization on Heterogeneous Systems Kerry A. Seitz, Jr. and Mark - - PowerPoint PPT Presentation
Optimization on Heterogeneous Systems Kerry A. Seitz, Jr. and Mark - - PowerPoint PPT Presentation
Virtual Machine and Bytecode for Optimization on Heterogeneous Systems Kerry A. Seitz, Jr. and Mark C. Lewis Trinity University, San Antonio, TX Acknowledgments Trinity University Department of Computer Science Trinity University Mach
SLIDE 2
SLIDE 3
Overview
Introduction Bytecode Design Implementation Conclusion
SLIDE 4
Heterogeneous Computing
Use all hardware
components available
CPU, GPU, FPGA, DSP
Increase Performance Hardware independent
| Introduction | Bytecode Design | Implementation | Conclusion |
SLIDE 5
Heterogeneous Computing Challenges
Different programming models Different combinations of devices Little development tool support
OpenCL, CUDA
Developer training
| Introduction | Bytecode Design | Implementation | Conclusion |
SLIDE 6
Motivations
Make heterogeneous computing easier Runtime code optimizations Platform independence
| Introduction | Bytecode Design | Implementation | Conclusion |
SLIDE 7
Annotation
Provide optimization hints to VM Can be disregarded until optimization Backwards compatibility Two formats
Within instructions Annotation file
| Introduction | Bytecode Design | Implementation | Conclusion |
SLIDE 8
Annotation Examples
Immutability Escape thread/stack frame Generics/Type Erasure Specify Device
| Introduction | Bytecode Design | Implementation | Conclusion |
SLIDE 9
Popular Virtual Machines
Java – stack-based Dalvik – register-based (16 registers) Both have issues
| Introduction | Bytecode Design | Implementation | Conclusion |
SLIDE 10
Virtual Machine Design
Variables
216 per stack frame Type associated with each Annotations
Class Files
Header modeled after Dalvik Annotations
| Introduction | Bytecode Design | Implementation | Conclusion |
SLIDE 11
Instruction Format
| Introduction | Bytecode Design | Implementation | Conclusion |
SLIDE 12
Instruction Format
| Introduction | Bytecode Design | Implementation | Conclusion |
SLIDE 13
Instruction Format
| Introduction | Bytecode Design | Implementation | Conclusion |
SLIDE 14
Instruction Format
| Introduction | Bytecode Design | Implementation | Conclusion |
SLIDE 15
Programming Language
Scala OpenCL
High-level Object oriented/functional Compiles to JVM Lower-level Heterogeneous Platform independent
(with correct drivers)
| Introduction | Bytecode Design | Implementation | Conclusion |
SLIDE 16
GPU Execution
Uses OpenCL (through JavaCL) Compiles bytecode instructions into OpenCL kernel Parallelizes over an array of data
| Introduction | Bytecode Design | Implementation | Conclusion |
SLIDE 17
Summary
Heterogeneous computing is hard, but increasingly
essential
Designed bytecode to aid in heterogeneous computing
Optimization hints in annotations Platform independent
Implementation in Scala with OpenCL
| Introduction | Bytecode Design | Implementation | Conclusion |
SLIDE 18
Future research
Design more annotations Implement more annotations Develop lower level interpreter/JIT compiler Explore program analysis techniques
| Introduction | Bytecode Design | Implementation | Conclusion |
SLIDE 19