When the result of a previous instruction is needed in the pipeline before it is available, a “data hazard” occurs. Register Forwarding and Pipeline Interlock (RF&PI) are mechanisms suitable to avoid data corruption and to limit the performance penalty caused by data hazards in pipelined microprocessors. Data hazards handling is part of the 1microprocessor control logic; its test can hardly be achieved with a functional approach, unless a specific test algorithm is adopted. In this paper we analyze the causes for the low functional testability of the RF&PI logic and propose some techniques able to effectively perform its test. In particular, we describe a strategy to perform Software-Based Self-Test (SBST) on the RF&PI unit. The general structure of the unit is analyzed, a suitable test algorithm is proposed and the strategy to observe the test responses is explained. The method can be exploited for test both at the end of manufacturing and in the operational phase. Feasibility and effectiveness of the proposed approach are demonstrated on both an academic MIPS-like processor and an industrial System-on-Chip based on the Power ArchitectureTM.

On the Functional Test of the Register Forwarding and Pipeline Interlocking Unit in Pipelined Processors / Bernardi, Paolo; Cantoro, Riccardo; Ciganda, LYL MERCEDES; Du, Boyang; SANCHEZ SANCHEZ, EDGAR ERNESTO; SONZA REORDA, Matteo; Grosso, Michelangelo; O., Ballan. - STAMPA. - (2014), pp. 52-57. (Intervento presentato al convegno 14th International Workshop on Microprocessor Test and Verification (MTV) tenutosi a Austin, Tx, USA nel December 12–13, 2013) [10.1109/MTV.2013.10].

On the Functional Test of the Register Forwarding and Pipeline Interlocking Unit in Pipelined Processors

BERNARDI, PAOLO;CANTORO, RICCARDO;CIGANDA, LYL MERCEDES;DU, BOYANG;SANCHEZ SANCHEZ, EDGAR ERNESTO;SONZA REORDA, Matteo;GROSSO, MICHELANGELO;
2014

Abstract

When the result of a previous instruction is needed in the pipeline before it is available, a “data hazard” occurs. Register Forwarding and Pipeline Interlock (RF&PI) are mechanisms suitable to avoid data corruption and to limit the performance penalty caused by data hazards in pipelined microprocessors. Data hazards handling is part of the 1microprocessor control logic; its test can hardly be achieved with a functional approach, unless a specific test algorithm is adopted. In this paper we analyze the causes for the low functional testability of the RF&PI logic and propose some techniques able to effectively perform its test. In particular, we describe a strategy to perform Software-Based Self-Test (SBST) on the RF&PI unit. The general structure of the unit is analyzed, a suitable test algorithm is proposed and the strategy to observe the test responses is explained. The method can be exploited for test both at the end of manufacturing and in the operational phase. Feasibility and effectiveness of the proposed approach are demonstrated on both an academic MIPS-like processor and an industrial System-on-Chip based on the Power ArchitectureTM.
File in questo prodotto:
Non ci sono file associati a questo prodotto.
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2525509
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo