| 1. |
- Aarrestad, Thea, et al.
(författare)
-
Fast convolutional neural networks on FPGAs with hls4ml
- 2021
-
Ingår i: Machine Learning: Science and Technology. - : IOP Publishing. - 2632-2153. ; 2:4
-
Tidskriftsartikel (refereegranskat)abstract
- We introduce an automated tool for deploying ultra low-latency, low-power deep neural networks with convolutional layers on field-programmable gate arrays (FPGAs). By extending the hls4ml library, we demonstrate an inference latency of 5 mu s using convolutional architectures, targeting microsecond latency applications like those at the CERN Large Hadron Collider. Considering benchmark models trained on the Street View House Numbers Dataset, we demonstrate various methods for model compression in order to fit the computational constraints of a typical FPGA device used in trigger and data acquisition systems of particle detectors. In particular, we discuss pruning and quantization-aware training, and demonstrate how resource utilization can be significantly reduced with little to no loss in model accuracy. We show that the FPGA critical resource consumption can be reduced by 97% with zero loss in model accuracy, and by 99% when tolerating a 6% accuracy degradation.
|
|
| 2. |
- Izosimov, Viacheslav, et al.
(författare)
-
Optimization of Assertion Placement in Time-Constrained Embedded Systems
- 2011
-
Ingår i: European Test Symposium (ETS2011), Trondheim, Norway, May 23-27, 2011.. - : IEEE. - 9781457704833 - 9780769544335
-
Konferensbidrag (refereegranskat)abstract
- We present an approach for optimization of assertion placement in time-constrained HW/SW modules for detection of errors due to transient and intermittent faults. During the design phases, these assertions have to be inserted into the executable code and, hence, will always be executed with the corresponding code branches. As the result, they can significantly increase execution time of a module, in particular, contributing to a much longer execution of the worst case, and cause deadline misses. Assertions have different characteristics such as tightness (or "local error coverage") and execution latency. Taking into account these properties can increase efficiency of assertion checks in time-constrained embedded HW/SW modules. We have developed a design optimization framework, which (1) identifies candidate locations for assertions, (2) associates a candidate assertion to each location, and (3) selects a set of assertions in terms of performance degradation and assertion tightness. Experimental results have shown the efficiency of the proposed techniques.
|
|
| 3. |
- Izosimov, Viacheslav, et al.
(författare)
-
Time-Constraint-Aware Optimization of Assertions in Embedded Software
- 2012
-
Ingår i: Journal of electronic testing. - : Springer Verlag (Germany). - 0923-8174 .- 1573-0727. ; 28:4, s. 469-486
-
Tidskriftsartikel (refereegranskat)abstract
- Technology shrinking and sensitization have led to more and more transient faults in embedded systems. Transient faults are intermittent and non-predictable faults caused by external events, such as energetic particles striking the circuits. These faults do not cause permanent damages, but may affect the running applications. One way to ensure the correct execution of these embedded applications is to keep debugging and testing even after shipping of the systems, complemented with recovery/restart options. In this context, the executable assertions that have been widely used in the development process for design validation can be deployed again in the final product. In this way, the application will use the assertion to monitor itself under the actual execution and will not allow erroneous out-of-the-specification behavior to manifest themselves. This kind of software-level fault tolerance may represent a viable solution to the problem of developing commercial off-the-shelf embedded systems with dependability requirements. But software-level fault tolerance comes at a computational cost, which may affect time-constrained applications. Thus, the executable assertions shall be introduced at the best possible points in the application code, in order to satisfy timing constraints, and to maximize the error detection efficiency. We present an approach for optimization of executable assertion placement in time-constrained embedded applications for the detection of transient faults. In this work, assertions have different characteristics such as tightness, i.e., error coverage, and performance degradation. Taking into account these properties, we have developed an optimization methodology, which identifies candidate locations for assertions and selects a set of optimal assertions with the highest tightness at the lowest performance degradation. The set of selected assertions is guaranteed to respect the real-time deadlines of the embedded application. Experimental results have shown the effectiveness of the proposed approach, which provides the designer with a flexible infrastructure for the analysis of time-constrained embedded applications and transient-fault-oriented executable assertions.
|
|
| 4. |
- Ponticelli, Francesco, et al.
(författare)
-
Safety and efficacy of coronary sinus narrowing in chronic refractory angina: Insights from the RESOURCE study.
- 2021
-
Ingår i: International journal of cardiology. - : Elsevier BV. - 1874-1754 .- 0167-5273. ; 337, s. 29-37
-
Tidskriftsartikel (refereegranskat)abstract
- Refractory angina (RA) is considered the end-stage of coronary artery disease, and often has no interventional treatment options. Coronary sinus Reducer (CSR) is a recent addition to the therapeutic arsenal, but its efficacy has only been evaluated on small populations. The RESOURCE registry provides further insights into this therapy.The RESOURCE is an observational, retrospective registry that includes 658 patients with RA from 20 centers in Europe, United Kingdom and Israel. Prespecified endpoints were the amelioration of anginal symptoms evaluated with the Canadian Cardiovascular Society (CCS) score, the rates of procedural success and complications, and MACEs as composite of all-cause mortality, acute coronary syndromes, and stroke.At a median follow-up of 502days (IQR 225-1091) after CSR implantation, 39.7% of patients improved by ≥2 CCS classes (primary endpoint), and 76% by ≥1 class. Procedural success was achieved in 96.7% of attempts, with 3% of procedures aborted mostly for unsuitable coronary sinus anatomy. Any complication occurred in 5.7% of procedures, but never required bailout surgery nor resulted in intra- or periprocedural death or myocardial infarction. One patient developed periprocedural stroke after inadvertent carotid artery puncture. At the last available follow-up, overall mortality and MACE were 10.4% and 14.6% respectively. At one, three and five years, mortality rate at Kaplan-Meier analysis was 4%, 13.7%, and 23.4% respectively.CSR implantation is safe and reduces angina in patients with refractory angina.
|
|
