Caching multi-bit hard error detection and fault tolerance device and method under near threshold voltage

Abstract

The invention relates to a caching multi-bit hard error detection and fault tolerance device and method under a near threshold voltage. The detection and fault tolerance device comprises a Cache datamemory, a Cache tag memory, a BIST circuit, an ECC memory, an error correction encoder, an error correction decoding, an error map memory, a compression map memory, an empty sub-block detector, a sub-block compressor and a sub-block decompressor, and ECC error correction codes and empty sub-block data compression methods of single-error correction double-error detection (SECDED) are used for detecting and conducting fault tolerance on multi-bit hard errors occurring in the Cache data memory according to a selected Cache management strategy. A microprocessor is achieved by the device and method, the multi-bit hard errors caused by fluctuations of process parameters in a cache memory can be detected in the microprocessor, fault tolerance can be conducted, and reliability when the microprocessor is under a low power consumption environment and uses a near threshold voltage technology is improved.

Description

technical field [0001] The invention relates to a detection and fault tolerance device for data errors in a microprocessor cache, in particular to a detection and fault tolerance device for multi-bit hard errors in a microprocessor cache under near-threshold voltage. The invention also relates to a detection and fault tolerance method for data errors in the cache memory of a microprocessor under a near-threshold voltage. Background technique [0002] As semiconductor technology improves, more transistors can be integrated on a chip, doubling every 18 months in accordance with Moore's Law. Therefore, a chip can theoretically integrate more transistor circuits. However, energy and power consumption limitations have largely prevented the normal use of more transistors, making it difficult for chip performance to continue to improve. The near-threshold voltage technology reduces the power supply voltage to near the transistor threshold voltage, which greatly reduces power cons...

AI Technical Summary

Problems solved by technology

[0008] Combining all existing technical solutions, there is no efficient and reliable fault-tolerant solution for multi-bit hard errors exceeding 4-bit hard errors in cache entries

Images