léa Coussy

Page 1. A Design Flow Dedicated to Multi-mode Architectures for DSP Applications ... A design flow dedicated to multi-mode architectures for DSP applications is thus needed. Ad-hoc scheduling, binding and register merging algorithms have to be proposed. ...

This paper addresses the design of multi-mode architectures for digital signal processing applications. We present a dedicated design flow and its associated high-level synthesis tool, named GAUT. Given a unified description of a set of time-wise mutually exclusive tasks and their associated throughput constraints, a single RTL hardware architecture optimized in area is generated. In order to reduce the register, steering logic (multiplexers) and controller (decoding logic) complexities, we propose a joint-scheduling algorithm which maximizes the similarities between control steps and specific binding approaches for both functional units and storage elements which maximize the similarities between the datapaths. We show through a set of test cases that our approach offers significant area saving relative to the state-of-the-art.

Digital Signal Processing (DSP) applications are widely used from wireless communications to automotive. Their ever growing complexity and throughput still require significant parts to be implemented as dedicated hardware accelerators. A High-Level Synthesis (HLS) flow to automatically generate hardware accelerators for DSP applications is proposed in this paper. By considering bit-width information during all the synthesis process both area and power consumption are optimized. Experimental results show that the proposed approach allows to generate architectures that offer better computation accuracy for a given area and/or power consumption. Effectiveness of the approach is shown through several design experiments in the DSP domain realized on a Xilinx Virtex-5 FPGA.

Experiments vary in their ability to distinguish between competing hypotheses. In tests on primate cognition the majority of this variation is due to an experimenter's ability to test primates in valid settings while providing the adequate amount of experimental control. While experimenters studying primate cognition can use methods of control perfected in captivity, it is still very unclear how to design and then objectively evaluate the external validity of new experimental paradigms. I recommend that more effort be allocated to specify how to create relevant test settings for primates. Primate social life is highly competitive. This means that all aspects of primates themselves, including their cognitive abilities, have likely been shaped by the need to out-compete conspecifics. Based on this hypothesis, sophisticated cognitive abilities of primates might best be demonstrated in competitive contexts. Thus, it is suggested that one possible measure of validity is whether investigators integrate a competitive component into their experimental designs. To evaluate this methodological prediction I review the literature on chimpanzee perspective-taking as a case study including several recent studies that include a competitive component in their experimental designs.

Dogs can learn effectively to detour around a V-shaped fence after observing a demonstration from either an unfamiliar human or dog demonstrator. We found earlier that there is substantial individual variation between the dogs’ performance, even when using the same experimental conditions. Here, we investigate if the subjects’ relative dominance rank with other dogs had an effect on their social learning performance. On the basis of the owners’ answers to a questionnaire, subjects from multi-dog homes were sorted into groups of dominant and subordinate dogs. In Experiment 1, dominant and subordinate dogs were tested without demonstration and we did not find any difference between the groups—they had similarly low detour performances on their own. In Experiment 2 and 3, dogs from single dog and multi-dog households were tested in the detour task with demonstration by an unfamiliar dog, or human, respectively. The results showed that social learning performance of the single dogs fell between the dominant and subordinate multi-dogs with both dog and human demonstration. Subordinate dogs displayed significantly better performance after having observed a dog demonstrator in comparison to dominant dogs. In contrast, the performance of dominant and subordinate dogs was almost similar, when they observed a human demonstrator. These results suggest that perceived dominance rank in its own group has a strong effect on social learning in dogs, but this effect seems to depend also on the demonstrator species. This finding reveals an intricate organization of the social structure in multi-dog households, which can contribute to individual differences existing among dogs.