“…Despite the duration of the experiment (18 months) and the huge quantity of data to manipulate (the individual V T evolution of more than 50 Gbits of memory cells has been stored and processed), the statistics of this first experiment remains relatively weak because of the extremely low rate of cell flips in this kind of memory. Nevertheless, the remarkable convergence of the experimental results and our numerical simulations (considering no fitting parameter in the complete simulation chain) for the neutron-SER indicates that this later value is more than two decades below the soft error rate usually measured in modern SRAMs [10,[21][22]. In the same way, the comparison of experimental data measured at sea-level and alpha-SER simulations clearly suggests that another mechanism than internal alpha-particle production in bulk materials may be responsible of charge loss from floating gates.…”
Section: Discussionmentioning
confidence: 55%
“…Two different radiation environments have thus been considered: the first one at sea-level in Rousset for reference and the second one in altitude on the ASTEP platform. The two sites are characterized by a relative atmospheric neutron flux of 1.04 and 6.02 with respect to New-York City, respectively [9][10]. After a period of exposition of several months, the wafers stored on ASTEP have been delivered to ST-Rousset for complete electrical characterization.…”
This work reports the combined characterization at mountain altitude (on the ASTEP Platform at 2552 m) and at sealevel of more than ~50 Gbit of 90 nm NOR flash memories subjected to natural radiation (atmospheric neutrons). This wafer-level experiment evidences a limited impact of the terrestrial radiation at ground level on the memory SER evaluated without ECC. Experimental values are compared to estimations obtained from Monte Carlo simulation using the TIARA-G4 code combined with a physical model for charge loss in such floating-gate devices.
“…Despite the duration of the experiment (18 months) and the huge quantity of data to manipulate (the individual V T evolution of more than 50 Gbits of memory cells has been stored and processed), the statistics of this first experiment remains relatively weak because of the extremely low rate of cell flips in this kind of memory. Nevertheless, the remarkable convergence of the experimental results and our numerical simulations (considering no fitting parameter in the complete simulation chain) for the neutron-SER indicates that this later value is more than two decades below the soft error rate usually measured in modern SRAMs [10,[21][22]. In the same way, the comparison of experimental data measured at sea-level and alpha-SER simulations clearly suggests that another mechanism than internal alpha-particle production in bulk materials may be responsible of charge loss from floating gates.…”
Section: Discussionmentioning
confidence: 55%
“…Two different radiation environments have thus been considered: the first one at sea-level in Rousset for reference and the second one in altitude on the ASTEP platform. The two sites are characterized by a relative atmospheric neutron flux of 1.04 and 6.02 with respect to New-York City, respectively [9][10]. After a period of exposition of several months, the wafers stored on ASTEP have been delivered to ST-Rousset for complete electrical characterization.…”
This work reports the combined characterization at mountain altitude (on the ASTEP Platform at 2552 m) and at sealevel of more than ~50 Gbit of 90 nm NOR flash memories subjected to natural radiation (atmospheric neutrons). This wafer-level experiment evidences a limited impact of the terrestrial radiation at ground level on the memory SER evaluated without ECC. Experimental values are compared to estimations obtained from Monte Carlo simulation using the TIARA-G4 code combined with a physical model for charge loss in such floating-gate devices.
“…Soft errors are one of the major concerns in the reliability of large-scale integrations. Understanding the failure modes and quantifying the error rate under natural radiation are primarily crucial for the development of high-density static random access memories (SRAMs), [19]. As a main memory of large-scale computing systems, dynamic random access memories (DRAMs) have been widely used, and have shown a remarkable reliability from the appllication point of view, though mostly owing to the implementation of redundancy and error correction coding (ECC).…”
Section: Fault Tolerance Analysismentioning
confidence: 99%
“…Consequently, robustness to memory errors is expected in embedded and safety-critical applications which will make use future memory generations subjected to expectedly decreasing fabrication yield. Robustness to memory errors is also a factor of cost reduction by enabling low redundancy of memory at the integrated circuit and system levels [18,19].…”
Remarkable hardware robustness of deep learning is revealed from an error-injection analysis performed using a custom hardware model implementing parallelized restricted Boltzmann machines (RBMs). RBMs used in deep belief networks (DBNs) demonstrate robustness against memory errors during and after learning. Fine-tuning has a significant impact on the recovery of accuracy under the presence of static errors that may modify structural data of RBMs. The proposed hardware networks with fine-graded memory distribution are observed to tolerate memory errors, thereby resulting in a reliable deep learning hardware platform, potentially suitable to safety-critical embedded applications.
“…This method is called "real-time" soft error rate (RTSER) test [25][26][27] or unaccelerated testing. In this method, as in accelerated testing, the intensity of the natural radiation can be increased by deploying the test in altitude (at least for neutrons).…”
Section: See Characterization Using Accelerated and Real-time Testsmentioning
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.