Combining Hardware- and Software-Based Self-Repair Methods for Statically Scheduled Data Paths

Abstract

This paper describes hardware-and software-based self-repair methods and how to combine them in order to obtain hybrid methods. All presented methods are able to handle multiple permanent faults in processors with a statically scheduled data path, e.g. a VLIW processor. They are based on adapting the executed program in the field to the current fault situation. The first method is a simple hardware-based technique. It binds dynamically operations to other execution units. A recently published first software-based method is briefly described, and a new second software-based method is introduced to overcome some weaknesses of the first one. Two hybrid methods are obtained by combining each software-based method with the hardware-based method. A detailed analysis of the advantages and disadvantages of each method is given. This includes a reliability analysis, the hardware overhead, the repair time, and the effect of multiple faults on the runtime of an executed application.

Run-time Instruction Replication for permanent and soft error mitigation in VLIW processors

“…Extending them for multiple permanent errors means that the instructions scheduled on a faulty unit have to be re-executed to another unit by adding an extra time slot. Similarly, hybrid approaches re-execute the instruction that can not be assigned to a sequential slot [6]. The behavior of these approaches can be estimated by using twice the execution cycles of the fault tolerance technique without any permanent errors, one time for the performance of the technique and a second time for the effect of the re-execution of the instructions scheduled to faulty units.…”

“…Some approaches combine software and hardware implementations. In [6] a software repair routine modifies the instructions permanently in the memory. During start-up, a self-test takes place to identify the faulty slots.…”

“…The use of spare units highly increases the area overhead for both the FUs and the control. The modification of the executed program is performed using software by re-writing the program in the memory [6] or by using multiple different versions of the program [7]. In the first case, time overhead is introduced, whereas in the second case the storage requirements are highly increased.…”

“…In the first case, time overhead is introduced, whereas in the second case the storage requirements are highly increased. A simple hardware approach has been combined with a software repair routine in [6], which sequentially schedules the instructions that cannot be scheduled to other slots, whereas the detection of the permanent errors has been performed up-front.…”

Run-Time Coarse-Grained Hardware Mitigation for Multiple Faults on VLIW Processors

“…Such type of processors are the Very Long Instruction Word (VLIW) processors, which usually consist of complex FUs able to execute all types of operations and simpler ones that cannot execute sophisticated operations, such as multiplications and divisions. Existing approaches on VLIW processors that detect the faulty units before the application execution, such as schedule multi-versioning [5] and instruction modification [6], cannot be applied for soft errors. Few approaches detect the faulty FUs during execution.…”