![integrated circuits (ICs) affect their reliability and set error rate levels outside
specifications [1], [2]. Various mechanisms like coupling noise, power supply
disturbance, jitter and temperature fluctuations are accused for timing error
generation. Moreover, transistor aging mechanisms significantly impact the
performance of nanometer circuits resulting in the appearance of timing errors
continuously earlier with technology evolution during their normal lifetime [3],
[4]. Such cases are the Negative-Positive Bias Temperature Instability (NBTI-
PBTI) induced aging of PMOS-NMOS transistors respectively and the hot-carrier
injection (HCI) induced aging of NMOS transistors.
These phenomena degrade transistors’ threshold voltage over time
increasing path delays [5]. Furthermore, modern systems running at multiple
frequency and voltage levels (e.g. exploiting dynamic voltage-frequency scaling –
DVFS techniques) may present increased timing error rates due to numerous
environmental and process related as well as data dependent variabilities that affect
circuit performance [6]. Path delay deviations, due to process variations, and
manufacturing defects that affect circuit speed may also result in timing errors that
are not easily detectable (in terms of test cost) in high device count ICs. The
inability of fabrication test procedures to exhaustively exercise the huge number of
paths in nanometer circuit designs and effectively screen out all timing related
defective ICs, increases the probability of test escapes. Additionally, and for the](https://image.slidesharecdn.com/timingerrortoleranceinsmallcoredesignsforsocapplications-150928053458-lva1-app6891/85/Timing-error-tolerance-in-small-core-designs-for-so-c-applications-2-320.jpg)
Timing error tolerance in small core designs for so c applications
- 1. Timing Error Tolerance in Small Core Designs for SoC Applications ABSTRACT: Timing errors are an increasing reliability concern in nanometer technology, high complexity and multivoltage/ frequency integrated circuits. A local error detection and correction technique is presented in this work that is based on a new bit flipping flip-flop. Whenever a timing error is detected, it is corrected by complementing the output of the corresponding flip-flop. The proposed solution is characterized by very low silicon area and power requirements compared to previous design schemes in the open literature. To validate its efficiency, it has been applied in the design of a MIPS microprocessor core in a 90nm technology, while a demonstration version of the same core in an FPGA platform is presented. EXISTING SYSTEM: THE CMOS technology scaling, the increase of process variations, the susceptibility of nanometer devices to various performance degradation mechanisms, the power supply reduction and the increasing complexity of modern
- 2. integrated circuits (ICs) affect their reliability and set error rate levels outside specifications [1], [2]. Various mechanisms like coupling noise, power supply disturbance, jitter and temperature fluctuations are accused for timing error generation. Moreover, transistor aging mechanisms significantly impact the performance of nanometer circuits resulting in the appearance of timing errors continuously earlier with technology evolution during their normal lifetime [3], [4]. Such cases are the Negative-Positive Bias Temperature Instability (NBTI- PBTI) induced aging of PMOS-NMOS transistors respectively and the hot-carrier injection (HCI) induced aging of NMOS transistors. These phenomena degrade transistors’ threshold voltage over time increasing path delays [5]. Furthermore, modern systems running at multiple frequency and voltage levels (e.g. exploiting dynamic voltage-frequency scaling – DVFS techniques) may present increased timing error rates due to numerous environmental and process related as well as data dependent variabilities that affect circuit performance [6]. Path delay deviations, due to process variations, and manufacturing defects that affect circuit speed may also result in timing errors that are not easily detectable (in terms of test cost) in high device count ICs. The inability of fabrication test procedures to exhaustively exercise the huge number of paths in nanometer circuit designs and effectively screen out all timing related defective ICs, increases the probability of test escapes. Additionally, and for the
- 3. same reasons, timing verification turns to be a hard task escalating the probability of timing failures in a design. PROPOSED SYSTEM: The proposed error detection and correction technique is based on the bit-flipping flip-flop concept. This is synopsized as follows: in case of error detection at the output of a flip-flop the corresponding logic value is asynchronously complemented for error correction. Fig. 1(a) illustrates the new Error Detection / Correction Flip-Flop (EDC Flip-Flop) that is suitable to confront with timing errors. Apart from the original flipflop (Main Flip-Flop), it consists of two XOR gates and a Latch. The first XOR gate compares the D input and the F output of the Main Flip-Flop and provides the result to the Latch. The Latch feeds the second XOR gate at the output of the Main Flip-Flop. Depending on the comparison result within a specified time interval, either the F signal of the Main Flip-Flop or its complement is propagated to the output Q of the EDC Flip-Flop. The Q signal feeds the subsequent logic. Briefly, the proposed timing error detection and correction technique operates as follows. Suppose that a timing error is detected at one or more inputs of the combinational logic stage Sj+1, due to a delayed response of the previous stage Sj. Thus, the response of Sj+1 will be erroneous and must be corrected. To achieve error correction, the output of each flip-flop, at the
- 4. register between the two stages, where a timing error has been detected is complemented so that valid values feed the Sj+1 logic stage. Moreover, in case that this stage is not fast enough (not a shallow stage), the evaluation time of the circuit is extended by one clock cycle to guarantee its correct computation.
- 5. SOFTWARE IMPLEMENTATION: Modelsim 6.0 Xilinx 14.2 HARDWARE IMPLEMENTATION: SPARTAN-III, SPARTAN-VI