Girish Venkataramani

Model-based hardware design

Ever shrinking design cycles, ever increasing design complexities, ever changing design standards, high NRE and design re-spin costs are a few of the major challenges in system-level design. In such an design environment, clock frequency, silicon footprint and power profile are no longer the sole design considerations. To this, we would have to add re-usability, platform re-targetability, testability, verifiability, standards compliance check-ability and design cycle time as first-order constraints. This paper presents model-based hardware design to address these challenges. It is a design methodology that bridges the world of system designers, software programmers, hardware engineers and verification engineers to increase productivity and produce correct-by-construction designs that match the system specification. We describe the various components of the methodology and present hard evidence of the realization of these ideals as cited by several real world case studies. In almost all cases, designers cite massive shrinkage of their design cycle time without sacrificing design quality.

System-Level Retiming and Pipelining

Various Emotion Models (e.g. Circumplex Model [1], Vector Model [2], PANA (Positive Activation - Negative Activation) Model [3], PAD (Pleasure Arousal Dominance) Model [4], etc...) can be used to represent the different emotional states of a robot. In this paper we chose to use a PAD Emotional Space [5] to simulate the emotional state of a NAO Robot (NAO is the name of a humanoid robot developed and commercialized by Aldebaran Robotics). But one difficulty is to write an algorithm that makes evolve the emotional state according to external and internal stimuli. Simulating a coherent Emotional State over the time is a real challenge for designing a virtual personality. As human emotions are fuzzy by nature, many researchers explored using Fuzzy State Machine [6], [7] or Fuzzy Logic [8], [9], [10], [11], [12] to control the emotions of a virtual character or robot. Following the same approach, in this paper we propose to use a Fuzzy Control System [13] to control the emotional state of a NAO robot by injecting attractors in his PAD Emotional Space. Fuzzy Control Systems are usually good controller of uncertainty and easy to write thanks to their syntax close to the human language. The result of our experiments showed that the evolution of the Emotional State of our NAO robot was smooth and coherent with the situation met when interacting with humans. Moreover with different set of rules it becomes easy to switch the personality of the robot (meaning the robot can react differently to the same stimuli according to the different set of rules used).In this paper, we examine retiming and pipelining in the context of system-level optimization techniques. Our main contributions are: (a) functionally equivalent retiming and delay balancing as necessary techniques for pipelining and retiming system-level graphs while maintaining numerical fidelity, and (b) clock-rate pipelining, as a new technique that leverages the knowledge of multi-rate design spec to pipeline multi-cycle paths. All these techniques have been implemented within HDL Coder, a tool that generates synthesizable HDL code from Simulink® and MATLAB®.Recently, in Washington D.C., National Aeronautics and Space Administration (NASA) and European Space Agency (ESA) officials decided pursuing a mission to study Jupiter and its moons, and another to visit the largest moons of Saturn. These initiatives represent important actions that set the future of the planetary science and could answer questions concerning how our Solar System formed and whether habitable conditions exist elsewhere in it. Those missions need a robust, efficient and flexible communication infrastructure. This opens the doors to new communications and networking challenges, which are the ones of the so called InterPlaNetary (IPN) Internet and they are the objects of the survey presented in this work.Energy and communication capability of sensor nodes are limited in Wireless Sensor Networks. Directed diffusion (DD) is a classic data-centric routing protocol. For the deficiency of large energy consumption in periodical interest flooding and data detection, as well as the ignorance of nodes energy in path reinforcement, we propose an improved directed diffusion protocol based on Visible Forwarding path and Promoted Evaluation criterion (VFPE). During interest propagation, the maximum and minimum numbers of hops to sink node are recorded. Visible forwarding path is set up to reduce the forwarding scope of detection data. Considering energy of nodes and transmission distance, promoted evaluation criterion is adopted in path reinforcement phase to select an optimal node, so as to balance network energy consumption. Simulation results show that VFPE protocol has a great improvement in network performances such as nodes energy and average end-to-end delay.While hybrid computing environments provide good potential for achieving high performance and low economic cost, it also introduces a broad set of unpredictable overheads especially for running data-intensive applications. This paper describes a novel approach which refines workflow structures and optimizes intermediate data transfers for large-scale scientific workflows containing thousands (or even millions) of tasks. The proposed method includes pre- and post-partitioning of workflows and data-flow optimization. Firstly, it partitions a workflow by identifying the critical path of the task graph. Secondly, it controls the granularity of partitions to reduce the complexity of task graph in order to process large-scale workflows. Thirdly, it optimizes the data-flow based on the scheduling to minimize its communication overheads. Our proposed approach is able to handle complex data flows and significantly reduce data transfer by replacing individual tasks according to data dependencies. We conducted experiments using real applications such as Montage and Broadband, and the results demonstrated the effectiveness of our methods in achieving low execution time with low communication overhead in a hybrid computing environments.

Comprehensive isomorphic subtree enumeration

A fundamental problem in program analysis and optimization concerns the discovery of structural similarities between different sections of a given program and/or across different programs. Specifically, there is a need to find topologically identical segments within compiler intermediate representations (IRs). Such topological isomorphism has many applications. For example, finding isomorphic sub-trees within different expression trees points to common computational resources that can be shared when targeting application-specific hardware. Isomorphism in the controlflow graph can be used to discovery of custom instructions for customizable processors. Discovering isomorphism in context call trees during program execution is invaluable to several JIT compiler optimizations. Thus, all these different applications rely on the fundamental ability to find topologically identical segments within a given tree or graph representation. In this paper, we present a generic formulation of the subtree isomorphism problem that is more powerful than previous proposals. We prove that an optimal quadratic time solution exists for this problem. We employ a dynamic programming based algorithm to efficiently enumerate all isomorphic sub-trees within given reference trees and also demonstrate its efficacy in a production compiler.

Tuning SoCs using the global dynamic critical path

We propose using a profiling-based technique (Dynamic Critical Path) to guide SoC optimization. Optimizing SoCs composed of many modules involves exploring a large space of possible configurations (exponential in the number of component modules). We present this optimization technique applied to a Globally Asynchronous Locally Synchronous (GALS) RTL design. Furthermore, we investigate the loss of precision when abstract versions of hardware modules are used for the critical path computation. Using the critical path provides very fast convergence towards optimal or near-optimal solutions when analyzing large configuration spaces by optimizing the design for composite optimization metrics, such as energy-delay.