Cherry-MP: Correctly Integrating Checkpointed Early Resource Recycling in Chip Multiprocessors
Authors:
- M. KyrmanMKM. Kyrman
- N. KyrmanNKN. Kyrman
- 11 October 2006
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 245-256
Abstract
Checkpointed early resource recycling (Cherry) is a recently-proposed microarchitectural technique that aims at improving critical resource utilization by performing aggressive resource recycling decoupled from instruction retirement, using a checkpoint/rollback mechanism to recover from occasional incorrect execution. In this paper, we explore correctness and performance issues that arise when Cherry-enabled processors are used in chip multiprocessor architectures. We propose mechanisms to address cache coherence, memory consistency, and forward progress issues in such environments. We also provide quantitative insight on the performance impact of the Cherry mechanism on parallel processing.Keywords
This publication has 28 references indexed in Scilit:
- A first glance at Kilo-instruction based multiprocessors2004
- Register file design considerations in dynamically scheduled processors2002
- Coordinated checkpointing-rollback error recovery for distributed shared memory multicomputers2002
- Memory consistency and event ordering in scalable shared-memory multiprocessors2002
- Speculative versioning cache2002
- POWER4 system microarchitectureIBM Journal of Research and Development, 2002
- ARB: a hardware mechanism for dynamic reordering of memory referencesIEEE Transactions on Computers, 1996
- The Mips R10000 superscalar microprocessorIEEE Micro, 1996
- Error recovery in shared memory multiprocessors using private cachesIEEE Transactions on Parallel and Distributed Systems, 1990
- A class of compatible cache consistency protocols and their support by the IEEE futurebusACM SIGARCH Computer Architecture News, 1986