1. Cost-Effective Register File Soft Error reduction Pablo Montesinos, Wei Liu and Josep Torellas, University of Illinois at Urbana-Champaign

2. Overview  Study of register file vulnerability to SDC(Silent Data Corruption)  Shield – cost effective protection to register files  Highighting policies and techniques used in shield  Experiment - Results

3. Register File AVF  RF-AVF is the probability that a fault that occurs will lead to error.  Register lifetime is divided into PreWrite, Useful, and PostLastRead parts.  Based on AVF calculation we can divide lifetime of bit into ACE (Architecturally Correct Execution) and un-ACE cycles.

4. Register File AVF  During PreWrite Period – un-ACE  If used atleast once after write the reg switches to ACE state.  After last read on reg, switches back to un-ACE during PostLastRead

5. Highlighting Insights (1)  The combined %-USEFUL time of all registers is small

6. Highlighting Insights (1)  The average number of useful (live) registers is less than 20 (SPECint) and 17(SPECfp).  It is thus possible to redue the vulnerability of the register file by only protecting a subset of carefully chosen registers at a time.

7. Highlighting Insights (2)  Only a few long-lived registers contribute to overall Total useful time  On average less than 10% of register versions are long-lived.

8. Highlighting Insights (2)  On average 40% of useful time comes from the few long-lived versions.  In SPECfp, 5% of long-lived versions account for 46% of the useful time.

9. Motivation  Register files have a very high access rate.  High temperature thus leading to lesser Qcrit for the devices.  An error in an RF can propagate with hght failure probability  If we isolate a few register versions, predicting their life- time, and protect these register versions alone, high reliability can be achieved with limited overhead.

10. Shield - Architecture Life-Time Prediction Shielding Decision Register Error Check Error Recovery

11. Reg-Version Lifetime Prediction P12 => Used(1), Renamed(1) P7 => Used(0), Renamed(1)

12. Shielding Decision  These prediction bits are stored as status in the ECC table.  The decision to shield an incoming register version written is by: Availability of free ECC-Table entry Same register# present in the ECC table will be replaced with new entry. Existing reg-version with lesser lifetime than incoming reg- version will be replaced.  Replacement policy:

13. Register Error Check & Recovery  On a read request the register data is sent to the original datapath and shield.  If the Reg# matches with a tag entry, then the reg-data is checked for errors at the ECC-Checker.  If Error is detected Processor stalls the instruction I reading reg P Reg-data is corrected and written into RF Oldest read instruction reading reg P in ROB and all succeeding instructions is flushed. Processor resumes from flushed instruction.

14. Experiments- Results  AVF computation for RF with shield

15. Experiments-Results  AVF of intREG reduced by different replacement policies: LRU = 31% Effective = 63% OptEffective = 84% ( pinning of global pointers to particular ECC entries + Effective )  AVF for fpREG can be reduced maximum by 100%, because fewer fp-registers are in useful state.

16. Power and Area Impact  Shield only uses 3ECC generators and 3 ECC checkers.  Shield has 45% power overhead over a plain register file. (Full ECC has 2X)  Shield introduces an overall 10% area overhead.

17. Conclusion  A cost-effective architectural technique has been proposed to reduce the vulnerability of RF by 84%  The area and power overhead indicated is a marginal tradeoff for reliability achieved.