SUMMARYA self-checkpointing microprocessor periodically copies the state of the currently executing program to on-chip non-volatile storage, allowing it to resume execution of the program at the last checkpoint after a power supply interruption or shutdown. In this paper, we present a register file cell for a self-checkpointing microprocessor that integrates a magnetoelectronic non-volatile memory cell into a register cell similar to the one used in the Itanium 2 microprocessor. Our design allows the contents of each bit in the register file to be checkpointed simultaneously, reducing the time required to take a checkpoint to a few clock cycles, without compromising the performance of the register file during normal read and write operations. Because the area of the base register file cell is determined by the wiring tracks it requires, adding self-checkpointing to the cell only increases its area by 6%. Similarly, adding self-checkpointing has effectively no impact on the cell's performance, only increasing read and write times from 299 to 300 ps, even when the additional load on the bit lines from the larger cells is considered.

No abstract

No abstract

Architectures that exploit control independence (CI) promise to remove in-order fetch bottlenecks, like branch mispredicts, instruction-cache misses and fetch unit stalls, from the critical path of single-threaded execution. By exposing more fetch options, however, CI architectures also expose more performance tradeoffs. These tradeoffs make it hard to design policies that deliver good performance.This paper presents a criticality-based model for reasoning about CI architectures, and uses that model to describe the tradeoffs between gains from control independence versus increased costs of honoring data dependences. The model is then used to derive the design of a criticality-aware task selection policy that strikes the right balance between fetch-criticality and execute-criticality. Finally, the papervalidates the model by attacking branch-misprediction induced fetch-criticality through the above derived spawnpolicy. This leads to as high as 100% improvements in performance, and in the region of 40% or more improvements for four of the benchmarks where this is the main problem. Criticality analysis shows that this improvement arises due to reduced fetch-criticality.

No abstract