Precision-aware soft error protection for GPUs
Authors:
- 1 February 2014
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 15300897,p. 49-59
Abstract
With the advent of general-purpose GPU computing, it is becoming increasingly desirable to protect GPUs from soft errors. For high computation throughout, GPUs must store a significant amount of state and have many execution units. The high power and area costs of full protection from soft errors make selective protection techniques attractive. Such approaches provide maximum error coverage within a fixed area or power limit, but typically treat all errors equally. We observe that for many floating-point-intensive GPGPU applications, small magnitude errors may have little effect on results, while large magnitude errors can be amplified to have a significant negative impact. We therefore propose a novel precision-aware protection approach for the GPU execution logic and register file to mitigate large magnitude errors. We also propose an architecture modification to optimize error coverage for integer computations. Our approach combines selective logic hardening, targeted checker circuits, and intelligent register file encoding for best error protection. We demonstrate that our approach can reduce the mean error magnitude by up to 87% compared to a traditional selective protection approach with the same overhead.Keywords
This publication has 30 references indexed in Scilit:
- Power-efficient computing for compute-intensive GPGPU applications2013
- Warped-DMR: Light-weight Error Detection for GPGPU2012
- Exploring the feasibility of selective hardening for combinational logicMicroelectronics Reliability, 2012
- A compile-time managed multi-level register file hierarchy2011
- Signal probability for reliability evaluation of logic circuitsMicroelectronics Reliability, 2008
- Approximate logic circuits for low overhead, non-intrusive concurrent error detection2008
- System Level Approaches for Mitigation of Long Duration Transient Faults in Future Technologies12th IEEE European Test Symposium (ETS'07), 2007
- Low-Cost Hardening of Image Processing Applications Against Soft Errors2006
- Microarchitectural power modeling techniques for deep sub-micron microprocessors2004
- Reliable floating-point arithmetic algorithms for error-coded operandsIEEE Transactions on Computers, 1994