arXiv subjects

Venkata Pavan Kumar Miriyala

Publications and source records attributed to Venkata Pavan Kumar Miriyala.

7 recordsLinked to original sources

arXiv2025

Data-aware Dynamic Execution of Irregular Workloads on Heterogeneous Systems

Current approaches to scheduling workloads on heterogeneous systems with specialized accelerators often rely on manual partitioning, offloading tasks with specific compute patterns to accelerators. This method requires extensive experimentation and human effort to identify the tasks suitable for the accelerator. To solve this problem, we introduce DyPe, a scheduling framework tailored for heterogeneous systems with specialized accelerators. Our method automatically partitions, deploys, and reschedules execution when necessary by dynamically analyzing the characteristics of the input data and leveraging the interoperator parallelism among heterogeneous devices. DyPe navigates a multi-objective, multi-constraint design space that considers both system constraints and application requirements, which allows it to discover Pareto-optimal mapping configurations, improving the system's overall performance and effectively managing energy-performance trade-offs. To demonstrate the benefits of our approach on real hardware, we build a heterogeneous system of GPUs and FPGAs with peer-to-peer data transfers. The experiments show that conventional static scheduling is optimal for 13 out of 86 cases for different workloads and system settings while DyPe is adaptable and able to find the optimal schedule in 77 out of 86 cases, with an average of only 3.95% performance or energy efficiency loss in the sub-optimal cases. Performance evaluation of DyPe shows an average of 1.53x throughput and 1.09x energy efficiency improvement over the static schedule baseline and 1.44x throughput and 1.66x energy efficiency over the GPU-only baseline.

cs.DC↗

arXiv2022

Multiply-and-Fire (MNF): An Event-driven Sparse Neural Network Accelerator

Machine learning, particularly deep neural network inference, has become a vital workload for many computing systems, from data centers and HPC systems to edge-based computing. As advances in sparsity have helped improve the efficiency of AI acceleration, there is a continued need for improved system efficiency for both high-performance and system-level acceleration. This work takes a unique look at sparsity with an event (or activation-driven) approach to ANN acceleration that aims to minimize useless work, improve utilization, and increase performance and energy efficiency. Our analytical and experimental results show that this event-driven solution presents a new direction to enable highly efficient AI inference for both CNN and MLP workloads. This work demonstrates state-of-the-art energy efficiency and performance centring on activation-based sparsity and a highly-parallel dataflow method that improves the overall functional unit utilization (at 30 fps). This work enhances energy efficiency over a state-of-the-art solution by 1.46$\times$. Taken together, this methodology presents a novel, new direction to achieve high-efficiency, high-performance designs for next-generation AI acceleration platforms.

cs.AR↗

arXiv2020

You Only Spike Once: Improving Energy-Efficient Neuromorphic Inference to ANN-Level Accuracy

In the past decade, advances in Artificial Neural Networks (ANNs) have allowed them to perform extremely well for a wide range of tasks. In fact, they have reached human parity when performing image recognition, for example. Unfortunately, the accuracy of these ANNs comes at the expense of a large number of cache and/or memory accesses and compute operations. Spiking Neural Networks (SNNs), a type of neuromorphic, or brain-inspired network, have recently gained significant interest as power-efficient alternatives to ANNs, because they are sparse, accessing very few weights, and typically only use addition operations instead of the more power-intensive multiply-and-accumulate (MAC) operations. The vast majority of neuromorphic hardware designs support rate-encoded SNNs, where the information is encoded in spike rates. Rate-encoded SNNs could be seen as inefficient as an encoding scheme because it involves the transmission of a large number of spikes. A more efficient encoding scheme, Time-To-First-Spike (TTFS) encoding, encodes information in the relative time of arrival of spikes. While TTFS-encoded SNNs are more efficient than rate-encoded SNNs, they have, up to now, performed poorly in terms of accuracy compared to previous methods. Hence, in this work, we aim to overcome the limitations of TTFS-encoded neuromorphic systems. To accomplish this, we propose: (1) a novel optimization algorithm for TTFS-encoded SNNs converted from ANNs and (2) a novel hardware accelerator for TTFS-encoded SNNs, with a scalable and low-power design. Overall, our work in TTFS encoding and training improves the accuracy of SNNs to achieve state-of-the-art results on MNIST MLPs, while reducing power consumption by 1.46$\times$ over the state-of-the-art neuromorphic hardware.

cs.NE↗

arXiv2020

Ultra-low power on-chip learning of speech commands with phase-change memories

Embedding artificial intelligence at the edge (edge-AI) is an elegant solution to tackle the power and latency issues in the rapidly expanding Internet of Things. As edge devices typically spend most of their time in sleep mode and only wake-up infrequently to collect and process sensor data, non-volatile in-memory computing (NVIMC) is a promising approach to design the next generation of edge-AI devices. Recently, we proposed an NVIMC-based neuromorphic accelerator using the phase change memories (PCMs), which we call as Raven. In this work, we demonstrate the ultra-low-power on-chip training and inference of speech commands using Raven. We showed that Raven can be trained on-chip with power consumption as low as 30~uW, which is suitable for edge applications. Furthermore, we showed that at iso-accuracies, Raven needs 70.36x and 269.23x less number of computations to be performed than a deep neural network (DNN) during inference and training, respectively. Owing to such low power and computational requirements, Raven provides a promising pathway towards ultra-low-power training and inference at the edge.

eess.AS↗

arXiv2020

Electrical generation and detection of terahertz signal based on spin-wave emission from ferrimagnets

Terahertz (THz) signals, mainly generated by photonic or electronic approaches, are being sought for various applications, whereas the development of magnetic source might be a necessary step to harness the magnetic nature of electromagnetic radiation. We show that the relativistic effect on the current-driven domain-wall motion induces THz spin-wave emission in ferrimagnets. The required current density increases dramatically in materials with strong exchange interaction and rapidly exceeds 1012 A m-2, leading to the device breakdown and thus the lack of experimental evidence. By translating the collective magnetization oscillations into voltage signals, we propose a three-terminal device for the electrical detection of THz spin wave. Through material engineering, wide frequency range from 264 GHz to 1.1 THz and uniform continuous signals with improved output power can be obtained. As a reverse effect, the spin wave generated in this system is able to move ferrimagnetic domain wall. Our work provides guidelines for the experimental verification of THz spin wave, and could stimulate the design of THz spintronic oscillators for wideband applications as well as the all-magnon spintronic devices.

cond-mat.mes-hall↗

arXiv2020

SIMBA: A Skyrmionic In-Memory Binary Neural Network Accelerator

Magnetic skyrmions are emerging as potential candidates for next generation non-volatile memories. In this paper, we propose an in-memory binary neural network (BNN) accelerator based on the non-volatile skyrmionic memory, which we call as SIMBA. SIMBA consumes 26.7 mJ of energy and 2.7 ms of latency when running an inference on a VGG-like BNN. Furthermore, we demonstrate improvements in the performance of SIMBA by optimizing material parameters such as saturation magnetization, anisotropic energy and damping ratio. Finally, we show that the inference accuracy of BNNs is robust against the possible stochastic behavior of SIMBA (88.5% +/- 1%).

cs.ET↗

arXiv2019

Spin-wave mediated interactions for Majority Computation using Skyrmions and Spin-torque Nano-oscillators

Recent progress in all-electrical nucleation, detection and manipulation of magnetic skyrmions has unlocked the tremendous potential of skyrmion-based spintronic devices. Here, we show via micromagnetic simulations that the stable magnetic oscillations of STNO radiate spin waves (SWs) that can be scattered in the presence of skyrmions in the near vicinity. Interference between SWs emitted by the STNO and SWs scattered by the skyrmion gives rise to interesting dynamics that leads to amplification or attenuation of STNO's magnetic oscillations. In the presence of strong Dzyaloshinskii-Moriya interaction (DMI), the amplified magnetic oscillations evolve into a new skyrmion. These interactions between skyrmions and STNOs are found to be identical for both Neel-type and Bloch-type skyrmions, and are not observed between domain walls and STNOs. These findings offer a novel perspective in processing information using single skyrmions and we propose a 3-bit majority gate for logic applications.

cond-mat.mes-hall↗