{"title":"CORE-VR:一种具有一致性和可重用性的多核低电压容错缓存","authors":"A. Choudhury, B. Sikdar","doi":"10.1109/DFT.2019.8875457","DOIUrl":null,"url":null,"abstract":"Voltage scaling to reduce power consumption for ensuring longer battery life expedites SRAM cell failures due to process variations. Cache, holding significant chip area, encounters these cell failures exponentially with voltage reduction. Several voltage reduction techniques have been proposed by tolerating faults at the cost of sacrificial cache portions affecting effective cache capacity. On this outset, this work attempts to minimize power below threshold by handling faults without affecting effective cache capacity. Words under priority blocks addressed on faulty cache portions are remapped in non-functional blocks to avoid cache pollution. Blocks are prioritized considering their coherence states and reusability. Non-reusable clean copies are invalidated to ensure adequate space for remapping and maintaining effective cache capacity. This work achieves minimum Vdd 325 mV with 7.77% area overhead with 6.7% leakage power and 0.5% dynamic power overhead in 90nm processor.","PeriodicalId":415648,"journal":{"name":"2019 IEEE International Symposium on Defect and Fault Tolerance in VLSI and Nanotechnology Systems (DFT)","volume":"28 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"CORE-VR: A Coherence and Reusability Aware Low Voltage Fault Tolerant Cache in Multicore\",\"authors\":\"A. Choudhury, B. Sikdar\",\"doi\":\"10.1109/DFT.2019.8875457\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Voltage scaling to reduce power consumption for ensuring longer battery life expedites SRAM cell failures due to process variations. Cache, holding significant chip area, encounters these cell failures exponentially with voltage reduction. Several voltage reduction techniques have been proposed by tolerating faults at the cost of sacrificial cache portions affecting effective cache capacity. On this outset, this work attempts to minimize power below threshold by handling faults without affecting effective cache capacity. Words under priority blocks addressed on faulty cache portions are remapped in non-functional blocks to avoid cache pollution. Blocks are prioritized considering their coherence states and reusability. Non-reusable clean copies are invalidated to ensure adequate space for remapping and maintaining effective cache capacity. This work achieves minimum Vdd 325 mV with 7.77% area overhead with 6.7% leakage power and 0.5% dynamic power overhead in 90nm processor.\",\"PeriodicalId\":415648,\"journal\":{\"name\":\"2019 IEEE International Symposium on Defect and Fault Tolerance in VLSI and Nanotechnology Systems (DFT)\",\"volume\":\"28 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2019-10-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2019 IEEE International Symposium on Defect and Fault Tolerance in VLSI and Nanotechnology Systems (DFT)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/DFT.2019.8875457\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2019 IEEE International Symposium on Defect and Fault Tolerance in VLSI and Nanotechnology Systems (DFT)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/DFT.2019.8875457","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
CORE-VR: A Coherence and Reusability Aware Low Voltage Fault Tolerant Cache in Multicore
Voltage scaling to reduce power consumption for ensuring longer battery life expedites SRAM cell failures due to process variations. Cache, holding significant chip area, encounters these cell failures exponentially with voltage reduction. Several voltage reduction techniques have been proposed by tolerating faults at the cost of sacrificial cache portions affecting effective cache capacity. On this outset, this work attempts to minimize power below threshold by handling faults without affecting effective cache capacity. Words under priority blocks addressed on faulty cache portions are remapped in non-functional blocks to avoid cache pollution. Blocks are prioritized considering their coherence states and reusability. Non-reusable clean copies are invalidated to ensure adequate space for remapping and maintaining effective cache capacity. This work achieves minimum Vdd 325 mV with 7.77% area overhead with 6.7% leakage power and 0.5% dynamic power overhead in 90nm processor.