A. Orailoglu and R. Karri. Coactive Scheduling and Checkpoint Determination during High-Level Synthesis of SelfRecovering Microarchitectures. IEEE Transactions on VLSI Systems, 2(3):304--311, September 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....and recovery techniques. For these applications, micro rollback techniques which roll back only a few instructions upon occurrence of a fault [15] are an ideal alternative. In this paper, we consider a micro rollback technique that has been proposed for high level synthesis [2] 3] 7] [13], This work was supported by the National Science Foundation under Grant CCR 9318495, by Hughes Electronics Corporation, and by the University of California MICRO Program. 14] In this technique, fault free states are maintained in a computation (specified by a control and data flow graph ....

....under Grant CCR 9318495, by Hughes Electronics Corporation, and by the University of California MICRO Program. 14] In this technique, fault free states are maintained in a computation (specified by a control and data flow graph (CDFG) scheduled in [2] 3] 14] and unscheduled in [7] [13]) by artificially extending the lifetimes of certain variables and using a hardened register file. The fault free states provide points to which execution can be rolled back upon occurrence of a fault and, hence, they are referred to as recovery points. A fundamental problem to be solved, called ....

[Article contains additional citation context not shown here]

A. Orailoglu and R. Karri, "Coactive Scheduling and Checkpoint Determination during High-Level Synthesis of Self-Recovering Microarchitectures, " IEEE Transactions on VLSI Systems, vol. 2, pp. 304--311, September 1994.

....in the presence of transient faults. In a self recovering microarchitecture, intermediate results are compared at regular intervals, and if correct saved in registers (checkpoints) On the other hand, on detecting a fault, it rolls back to a checkpoint and retries. Orailoglu and Karri [69] proposed a self recovering microarchitecture synthesis system. They proposed an algorithm for the selection of good checkpoints that have low overhead while meeting the constraint on the number of clock cycles of a retry period. Self testing can be carried out concurrently with normal operations ....

A. Orailoglu and R. Karri, "Coactive Scheduling and Checkpoint Determination During High Level Synthesis of Self-Recovering Microarchitectures," IEEE Transactions on Very Large Scale Integration (VLSI) Systems, Vol. 2, No. 3, pp. 304-311, Sep. 1994.

....microarchitectures has been previously addressed. An algorithm that intertwines checkpoint insertion and scheduling (of operations in the input algorithm to clock cycles) to synthesize self recovering microarchitectures for supporting fault recovery in hardware was first presented in [19, 18]. Guerra et. al have developed synthesis for built in self repair using redundant modules[3] More recently, Blough, Kurdahi and Ohm [14] presented an algorithm for recovery point insertion in recoverable microarchitectures that minimizes the number of rollback points given constraints on the ....

A. Orailoglu and R. Karri. "Coactive Scheduling and Checkpoint Determination during the High Level Synthesis of Self Recovering Microarchitectures," IEEE Trans on VLSI Systems, 2(3):304-311, 1994.

....reconfigurable design problem requires sophisticated planning in both operation scheduling and hardware binding, and these tasks fall squarely within the problem domain of high level synthesis. In fact, high level synthesis of fault tolerant designs has been investigated and implemented in [3] [7] and [8] Moreover, the procedural approaches in high level # This work is supported by the National Science Foundation under grant number MIP 9308535. synthesis advocate a systematic definition of the underlying architecture, which in turn assists efficient planning of reconfigurable system ....

A. Orailoglu and R. Karri. Coactive Scheduling and Checkpoint Determination during High-Level Synthesis of SelfRecovering Microarchitectures. IEEE Transactions on VLSI Systems, 2(3):304--311, September 1994.

....use redundant functional units to overcome permanent fabrication time faults. In the area of high level synthesis most researchers have addressed the problems of area and performance optimization[6] More recently, other important objectives, such as power [7] testability [8] and fault tolerance [9] have also been addressed. 2 PHANTOM REDUNDANCY Phantom redundancy yields gracefully degradable data paths with minimal hardware overhead by using additional interconnects instead of spare functional units to achieve fault tolerance. Upon detecting a faulty functional unit at fabrication time, ....

A. Orailoglu and R. Karri, "Coactive Scheduling and Checkpoint Determination During High Level Synthesis of SelfRecovering Microarchitectures", in IEEE Trans. VLSI, pp. 429--433, Sep. 1994.

....points of failure in an otherwise fault tolerant design and hence are meticulously handcrafted for fault tolerance. 3. 1 Self Recovering Data Path Synthesis Towards optimizing the area overhead attendant upon checkpointing we have developed a self recovering microarchitecture synthesis system [16, 14]. A novel self recovery scheduler [16] has been developed to optimize the register and voter overhead. Furthermore, the scheduler inserts checkpoints into the computation. In contrast to traditional applications, the data transfers are as important as the operations in a computation. This is ....

....design and hence are meticulously handcrafted for fault tolerance. 3. 1 Self Recovering Data Path Synthesis Towards optimizing the area overhead attendant upon checkpointing we have developed a self recovering microarchitecture synthesis system [16, 14] A novel self recovery scheduler [16] has been developed to optimize the register and voter overhead. Furthermore, the scheduler inserts checkpoints into the computation. In contrast to traditional applications, the data transfers are as important as the operations in a computation. This is because, these data transfers represent (i) ....

A. Orailoglu and R. Karri. Coactive Scheduling and Checkpoint Determination during High Level Synthesis of Self Recovering Microarchitectures. IEEE Transactions on VLSI Systems, 2(3):304--311, 1994.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC