Sreenivas Subramoney
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Interests
Sreenivas Subramoney has a total of 57 co-authors. Here are the top co-authors by number of shared publications:
- Aravinda Prasad
Intel, Bengaluru, India
3 shared publications - Gurpreet Singh Kalsi
Processor Architecture Research Lab, Intel Labs
3 shared publications - Sandeep Kumar
Intel, Bengaluru, India
3 shared publications - Udayan Ganguly
Indian Institute of Technology Bombay
3 shared publications - Kamlesh Pillai
Intel Labs, USA
2 shared publications - Maryam Shojaei Baghini
Department of Electrical Engineering, IIT Bombay, Mumbai, India
2 shared publications
Latest Publications
- Open Access
- 28 September 2026
- Published by Association for Computing Machinery (ACM)
- p. 89-96
Speeding up FaaS Warm Starts with Memory Restoration Templates
- 28 September 2026
- Published by Association for Computing Machinery (ACM)
- p. 89-96
Open AccessSpeeding up FaaS Warm Starts with Memory Restoration Templates
- Open Access
- 13 May 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Embedded Computing Systems
- Vol. 25 (3) , 1-28
Ace-of-Spade s: Accelerating Spatially Sparse Convolution for 3D Scene Understanding
Semantic understanding of 3D scenes is fundamental to many applications like robotics, autonomous driving, AR/VR. State-of-the-art methods for different 3D scene understanding tasks use 3D convolutional neural networks (CNNs) operating on point clouds. Convolution on spatially sparse data like point cloud involves irregular data accesses and compute patterns leading to poor utilization and energy efficiency in CPU/GPU implementations. The existing CNN accelerators designed for weight/activation sparsity cannot be efficiently repurposed for 3D spatially sparse CNNs given the fundamental differences in locating non-zero operands and granularity of work-dispatches. To address the dataflow challenges due to spatial sparsity and the need for specialized microarchitecture for spatially sparse convolution we present Ace-of-Spade s (AoS), an algorithm-dataflow-architecture co-designed system. AoS enables the data reuse among spatially proximate points using a locality-aware metadata structure along with a surface orientation aware point cloud reordering algorithm. AoS uses a novel technique for spatial sparsity aware selection of optimal data tiles by modelling the sparsity induced variations in the point cloud with a near-zero latency overheads. To accelerate computation on spatially sparse data, we propose a novel hardware accelerator Ss p nna with a front-end to convert varying number of operations per point into a stream of dense work dispatches to the backend compute engine. The compute engine further exploits weight and input feature data reuse through dynamic systolic grouping and multicast interconnects. The Ss p nna core together with the 64 KB of L1 memory requires 0.31 mm 2 of area in 10nm process at 1 GHz. Overall, AoS achieves speedup/energy savings of 19.9x / 49.9x and 2.2x / 7.1x over the state-of-the-art CPU and GPU implementations respectively.
- 13 May 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Embedded Computing Systems
- Vol. 25 (3) , 1-28
Open AccessAce-of-Spade s: Accelerating Spatially Sparse Convolution for 3D Scene Understanding
Semantic understanding of 3D scenes is fundamental to many applications like robotics, autonomous driving, AR/VR. State-of-the-art methods for different 3D scene understanding tasks use 3D convolutional neural networks (CNNs) operating on point clouds. Convolution on spatially sparse data like point cloud involves irregular data accesses and compute patterns leading to poor utilization and energy efficiency in CPU/GPU implementations. The existing CNN accelerators designed for weight/activation sparsity cannot be efficiently repurposed for 3D spatially sparse CNNs given the fundamental differences in locating non-zero operands and granularity of work-dispatches. To address the dataflow challenges due to spatial sparsity and the need for specialized microarchitecture for spatially sparse convolution we present Ace-of-Spade s (AoS), an algorithm-dataflow-architecture co-designed system. AoS enables the data reuse among spatially proximate points using a locality-aware metadata structure along with a surface orientation aware point cloud reordering algorithm. AoS uses a novel technique for spatial sparsity aware selection of optimal data tiles by modelling the sparsity induced variations in the point cloud with a near-zero latency overheads. To accelerate computation on spatially sparse data, we propose a novel hardware accelerator Ss p nna with a front-end to convert varying number of operations per point into a stream of dense work dispatches to the backend compute engine. The compute engine further exploits weight and input feature data reuse through dynamic systolic grouping and multicast interconnects. The Ss p nna core together with the 64 KB of L1 memory requires 0.31 mm 2 of area in 10nm process at 1 GHz. Overall, AoS achieves speedup/energy savings of 19.9x / 49.9x and 2.2x / 7.1x over the state-of-the-art CPU and GPU implementations respectively.
- Open Access
- 26 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 247-262
TierScape: Harnessing Multiple Compressed Tiers to Tame Server Memory TCO
Tiered memory systems are the norm to effectively tackle the increasing memory total cost of ownership (TCO) in modern data centers. We propose TierScape to tame memory TCO through the novel creation and judicious management of multiple compressed memory tiers along with multiple byte-addressable tiers. As opposed to conventional tiering solutions that employ a single compressed memory tier, we harness multiple compressed tiers implemented through a combination of different compression algorithms, memory allocators for compressed objects, and backing media to store compressed objects. These compressed tiers represent distinct points in the access latency, data compressibility, and unit memory usage cost spectrum, allowing rich and flexible trade-offs between memory TCO savings and application performance impact. A key advantage with TierScape is that it enables aggressive memory TCO saving opportunities by placing warm data in low latency compressed tiers with a reasonable performance impact while simultaneously placing cold data in the best memory TCO saving tiers. TierScape presents a comprehensive, rigorous, and tunable analytical cost model for performance and TCO trade-off based on continuous monitoring of the application's data access profile. Guided by this model, TierScape takes informed actions to dynamically manage the placement and migration of application data across multiple compressed and byte-addressable tiers. We believe TierScape represents an important server system configuration and optimization capability to achieve the best SLA-aware performance per dollar for applications hosted in production data center environments. For a set of real-world benchmarks, TierScape reduces memory TCO by 15.1–23.6% percentage points while maintaining performance parity or improves performance by 2.61–10.0% percentage points while maintaining memory TCO parity compared to state-of-the-art tiering solutions.
- 26 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 247-262
Open AccessTierScape: Harnessing Multiple Compressed Tiers to Tame Server Memory TCO
Tiered memory systems are the norm to effectively tackle the increasing memory total cost of ownership (TCO) in modern data centers. We propose TierScape to tame memory TCO through the novel creation and judicious management of multiple compressed memory tiers along with multiple byte-addressable tiers. As opposed to conventional tiering solutions that employ a single compressed memory tier, we harness multiple compressed tiers implemented through a combination of different compression algorithms, memory allocators for compressed objects, and backing media to store compressed objects. These compressed tiers represent distinct points in the access latency, data compressibility, and unit memory usage cost spectrum, allowing rich and flexible trade-offs between memory TCO savings and application performance impact. A key advantage with TierScape is that it enables aggressive memory TCO saving opportunities by placing warm data in low latency compressed tiers with a reasonable performance impact while simultaneously placing cold data in the best memory TCO saving tiers. TierScape presents a comprehensive, rigorous, and tunable analytical cost model for performance and TCO trade-off based on continuous monitoring of the application's data access profile. Guided by this model, TierScape takes informed actions to dynamically manage the placement and migration of application data across multiple compressed and byte-addressable tiers. We believe TierScape represents an important server system configuration and optimization capability to achieve the best SLA-aware performance per dollar for applications hosted in production data center environments. For a set of real-world benchmarks, TierScape reduces memory TCO by 15.1–23.6% percentage points while maintaining performance parity or improves performance by 2.61–10.0% percentage points while maintaining memory TCO parity compared to state-of-the-art tiering solutions.
- Open Access
- 9 October 2025
- Published by Association for Computing Machinery (ACM) in ACM Journal on Emerging Technologies in Computing Systems
- Vol. 21 (4) , 1-18
Analog and Temporary On-chip Memory for ANN Training and Inference
On-chip training at the edge becomes a primary requisite for real-time and security-sensitive artificial neural network (ANN) applications. In-memory computation (IMC) techniques have been proposed to facilitate data-intensive computational operations in ANNs. IMC-based multiply-accumulate (MAC) accelerates ANN training but suffers from significant communication overhead between the MAC engine and the off-chip storage for the intermediate data. This article proposes an analog temporary on-chip memory (ATOM) to store this intermediate data during ANN training. The ANN training architecture with the proposed ATOM has two significant advantages. First, the energy required to store intermediate data is scaled down by \(\sim\) 40 \(\times\) due to the on-chip and analog nature of the memory. Second, the proposed architecture avoids power and area-consuming analog-to-digital converters (ADCs) between neural network stages. The ATOM cell measurements are carried out from 20 fabricated chips, and the impact of ATOM characteristics on ANN system performance accuracy is analyzed. This article shows significant latency improvement of \(\sim\) 9 \(\times\) and area savings of \(\sim\) 5 \(\times\) for intermediate data storage compared to the on-chip SRAM during ANN training’s forward and backward pass operations. An improvement in the area and latency will be beneficial to instrument the area- and energy-efficient hardware system for on-chip ANN applications.- 9 October 2025
- Published by Association for Computing Machinery (ACM) in ACM Journal on Emerging Technologies in Computing Systems
- Vol. 21 (4) , 1-18
Open AccessAnalog and Temporary On-chip Memory for ANN Training and Inference
On-chip training at the edge becomes a primary requisite for real-time and security-sensitive artificial neural network (ANN) applications. In-memory computation (IMC) techniques have been proposed to facilitate data-intensive computational operations in ANNs. IMC-based multiply-accumulate (MAC) accelerates ANN training but suffers from significant communication overhead between the MAC engine and the off-chip storage for the intermediate data. This article proposes an analog temporary on-chip memory (ATOM) to store this intermediate data during ANN training. The ANN training architecture with the proposed ATOM has two significant advantages. First, the energy required to store intermediate data is scaled down by \(\sim\) 40 \(\times\) due to the on-chip and analog nature of the memory. Second, the proposed architecture avoids power and area-consuming analog-to-digital converters (ADCs) between neural network stages. The ATOM cell measurements are carried out from 20 fabricated chips, and the impact of ATOM characteristics on ANN system performance accuracy is analyzed. This article shows significant latency improvement of \(\sim\) 9 \(\times\) and area savings of \(\sim\) 5 \(\times\) for intermediate data storage compared to the on-chip SRAM during ANN training’s forward and backward pass operations. An improvement in the area and latency will be beneficial to instrument the area- and energy-efficient hardware system for on-chip ANN applications. - Open Access
- 13 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 56-71
TierTrain: Proactive Memory Tiering for CPU-Based DNN Training
- 13 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 56-71
Open AccessTierTrain: Proactive Memory Tiering for CPU-Based DNN Training
- Open Access
- 15 February 2024
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Architecture and Code Optimization
- Vol. 21 (1) , 1-29
ApHMM: Accelerating Profile Hidden Markov Models for Fast and Energy-efficient Genome Analysis
Profile hidden Markov models (pHMMs) are widely employed in various bioinformatics applications to identify similarities between biological sequences, such as DNA or protein sequences. In pHMMs, sequences are represented as graph structures, where states and edges capture modifications (i.e., insertions, deletions, and substitutions) by assigning probabilities to them. These probabilities are subsequently used to compute the similarity score between a sequence and a pHMM graph. The Baum-Welch algorithm, a prevalent and highly accurate method, utilizes these probabilities to optimize and compute similarity scores. Accurate computation of these probabilities is essential for the correct identification of sequence similarities. However, the Baum-Welch algorithm is computationally intensive, and existing solutions offer either software-only or hardware-only approaches with fixed pHMM designs. When we analyze state-of-the-art works, we identify an urgent need for a flexible, high-performance, and energy-efficient hardware-software co-design to address the major inefficiencies in the Baum-Welch algorithm for pHMMs.We introduceApHMM, thefirstflexible acceleration framework designed to significantly reduce both computational and energy overheads associated with the Baum-Welch algorithm for pHMMs. ApHMM employs hardware-software co-design to tackle the major inefficiencies in the Baum-Welch algorithm by (1) designing flexible hardware to accommodate various pHMM designs, (2) exploiting predictable data dependency patterns through on-chip memory with memoization techniques, (3) rapidly filtering out unnecessary computations using a hardware-based filter, and (4) minimizing redundant computations.ApHMM achieves substantial speedups of 15.55×–260.03×, 1.83×–5.34×, and 27.97× when compared to CPU, GPU, and FPGA implementations of the Baum-Welch algorithm, respectively. ApHMM outperforms state-of-the-art CPU implementations in three key bioinformatics applications: (1) error correction, (2) protein family search, and (3) multiple sequence alignment, by 1.29×–59.94×, 1.03×–1.75×, and 1.03×–1.95×, respectively, while improving their energy efficiency by 64.24×–115.46×, 1.75×, and 1.96×.
- 15 February 2024
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Architecture and Code Optimization
- Vol. 21 (1) , 1-29
Open AccessApHMM: Accelerating Profile Hidden Markov Models for Fast and Energy-efficient Genome Analysis
Profile hidden Markov models (pHMMs) are widely employed in various bioinformatics applications to identify similarities between biological sequences, such as DNA or protein sequences. In pHMMs, sequences are represented as graph structures, where states and edges capture modifications (i.e., insertions, deletions, and substitutions) by assigning probabilities to them. These probabilities are subsequently used to compute the similarity score between a sequence and a pHMM graph. The Baum-Welch algorithm, a prevalent and highly accurate method, utilizes these probabilities to optimize and compute similarity scores. Accurate computation of these probabilities is essential for the correct identification of sequence similarities. However, the Baum-Welch algorithm is computationally intensive, and existing solutions offer either software-only or hardware-only approaches with fixed pHMM designs. When we analyze state-of-the-art works, we identify an urgent need for a flexible, high-performance, and energy-efficient hardware-software co-design to address the major inefficiencies in the Baum-Welch algorithm for pHMMs.We introduceApHMM, thefirstflexible acceleration framework designed to significantly reduce both computational and energy overheads associated with the Baum-Welch algorithm for pHMMs. ApHMM employs hardware-software co-design to tackle the major inefficiencies in the Baum-Welch algorithm by (1) designing flexible hardware to accommodate various pHMM designs, (2) exploiting predictable data dependency patterns through on-chip memory with memoization techniques, (3) rapidly filtering out unnecessary computations using a hardware-based filter, and (4) minimizing redundant computations.ApHMM achieves substantial speedups of 15.55×–260.03×, 1.83×–5.34×, and 27.97× when compared to CPU, GPU, and FPGA implementations of the Baum-Welch algorithm, respectively. ApHMM outperforms state-of-the-art CPU implementations in three key bioinformatics applications: (1) error correction, (2) protein family search, and (3) multiple sequence alignment, by 1.29×–59.94×, 1.03×–1.75×, and 1.03×–1.95×, respectively, while improving their energy efficiency by 64.24×–115.46×, 1.75×, and 1.96×.
- Open Access
Reclaimer: A Reinforcement Learning Approach to Dynamic Resource Allocation for Cloud Microservices
Many cloud applications are migrated from the monolithic model to a microservices framework in which hundreds of loosely-coupled microservices run concurrently, with significant benefits in terms of scalability, rapid development, modularity, and isolation. However, dependencies among microservices with uneven execution time may result in longer queues, idle resources, or Quality-of-Service (QoS) violations. In this paper we introduce Reclaimer, a deep reinforcement learning model that adapts to runtime changes in the number and behavior of microservices in order to minimize CPU core allocation while meeting QoS requirements. When evaluated with two benchmark microservice-based applications, Reclaimer reduces the mean CPU core allocation by 38.4% to 74.4% relative to the industry-standard scaling solution, and by 27.5% to 58.1% relative to a current state-of-the art method.
Open AccessReclaimer: A Reinforcement Learning Approach to Dynamic Resource Allocation for Cloud Microservices
Many cloud applications are migrated from the monolithic model to a microservices framework in which hundreds of loosely-coupled microservices run concurrently, with significant benefits in terms of scalability, rapid development, modularity, and isolation. However, dependencies among microservices with uneven execution time may result in longer queues, idle resources, or Quality-of-Service (QoS) violations. In this paper we introduce Reclaimer, a deep reinforcement learning model that adapts to runtime changes in the number and behavior of microservices in order to minimize CPU core allocation while meeting QoS requirements. When evaluated with two benchmark microservice-based applications, Reclaimer reduces the mean CPU core allocation by 38.4% to 74.4% relative to the industry-standard scaling solution, and by 27.5% to 58.1% relative to a current state-of-the art method.
- Open Access
VEGETA: Vertically-Integrated Extensions for Sparse/Dense GEMM Tile Acceleration on CPUs
Deep Learning (DL) acceleration support in CPUs has recently gained a lot of traction, with several companies (Arm, Intel, IBM) announcing products with specialized matrix engines accessible via GEMM instructions. CPUs are pervasive and need to handle diverse requirements across DL workloads running in edge/HPC/cloud platforms. Therefore, as DL workloads embrace sparsity to reduce the computations and memory size of models, it is also imperative for CPUs to add support for sparsity to avoid under-utilization of the dense matrix engine and inefficient usage of the caches and registers. This work presents VEGETA, a set of ISA and microarchitecture extensions over dense matrix engines to support flexible structured sparsity for CPUs, enabling programmable support for diverse DL models with varying degrees of sparsity. Compared to the state-of-the-art (SOTA) dense matrix engine in CPUs, a VEGETA engine provides 1.09x, 2.20x, 3.74x, and 3.28x speed-ups when running 4:4 (dense), 2:4, 1:4, and unstructured (95%) sparse DNN layers.
Open AccessVEGETA: Vertically-Integrated Extensions for Sparse/Dense GEMM Tile Acceleration on CPUs
Deep Learning (DL) acceleration support in CPUs has recently gained a lot of traction, with several companies (Arm, Intel, IBM) announcing products with specialized matrix engines accessible via GEMM instructions. CPUs are pervasive and need to handle diverse requirements across DL workloads running in edge/HPC/cloud platforms. Therefore, as DL workloads embrace sparsity to reduce the computations and memory size of models, it is also imperative for CPUs to add support for sparsity to avoid under-utilization of the dense matrix engine and inefficient usage of the caches and registers. This work presents VEGETA, a set of ISA and microarchitecture extensions over dense matrix engines to support flexible structured sparsity for CPUs, enabling programmable support for diverse DL models with varying degrees of sparsity. Compared to the state-of-the-art (SOTA) dense matrix engine in CPUs, a VEGETA engine provides 1.09x, 2.20x, 3.74x, and 3.28x speed-ups when running 4:4 (dense), 2:4, 1:4, and unstructured (95%) sparse DNN layers.
- 17 December 2022
- Published by Springer Nature
- p. 499-512
Disrupting Low-Write-Energy vs. Fast-Read Dilemma in RRAM to Enable L1 Instruction Cache
- 17 December 2022
- Published by Springer Nature
- p. 499-512
Disrupting Low-Write-Energy vs. Fast-Read Dilemma in RRAM to Enable L1 Instruction Cache
- Open Access
- 4 December 2021
- Published by MDPI AG in Mathematics
- Vol. 9 (23) , 3130
Early Prediction of DNN Activation Using Hierarchical Computations
Deep Neural Networks (DNNs) have set state-of-the-art performance numbers in diverse fields of electronics (computer vision, voice recognition), biology, bioinformatics, etc. However, the process of learning (training) from the data and application of the learnt information (inference) process requires huge computational resources. Approximate computing is a common method to reduce computation cost, but it introduces loss in task accuracy, which limits their application. Using an inherent property of Rectified Linear Unit (ReLU), a popular activation function, we propose a mathematical model to perform MAC operation using reduced precision for predicting negative values early. We also propose a method to perform hierarchical computation to achieve the same results as IEEE754 full precision compute. Applying this method on ResNet50 and VGG16 shows that up to 80% of ReLU zeros (which is 50% of all ReLU outputs) can be predicted and detected early by using just 3 out of 23 mantissa bits. This method is equally applicable to other floating-point representations.
- 4 December 2021
- Published by MDPI AG in Mathematics
- Vol. 9 (23) , 3130
Open AccessEarly Prediction of DNN Activation Using Hierarchical Computations
Deep Neural Networks (DNNs) have set state-of-the-art performance numbers in diverse fields of electronics (computer vision, voice recognition), biology, bioinformatics, etc. However, the process of learning (training) from the data and application of the learnt information (inference) process requires huge computational resources. Approximate computing is a common method to reduce computation cost, but it introduces loss in task accuracy, which limits their application. Using an inherent property of Rectified Linear Unit (ReLU), a popular activation function, we propose a mathematical model to perform MAC operation using reduced precision for predicting negative values early. We also propose a method to perform hierarchical computation to achieve the same results as IEEE754 full precision compute. Applying this method on ResNet50 and VGG16 shows that up to 80% of ReLU zeros (which is 50% of all ReLU outputs) can be predicted and detected early by using just 3 out of 23 mantissa bits. This method is equally applicable to other floating-point representations.