Ram Kumar Krishnamurthy

Ram Kumar Krishnamurthy

Auto-generated profile, based on publicly available information

Ram Kumar Krishnamurthy

Intel Corporation, Hillsboro, OR, USA

Research Focus & Metrics

Academic Career Length

30Years (1997-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
3 December 2020 in IEEE Journal of Solid-State Circuits
Verified
Journal Article
Closed access
PKPhil C. Knag
Phil C. Knag
GCGregory K. Chen
Gregory K. Chen
HSH. Ekin Sumbul
H. Ekin Sumbul
RKRaghavan Kumar
Raghavan Kumar
SHSteven K. Hsu
Steven K. Hsu
RKRam K. Krishnamurthy
Ram K. Krishnamurthy
A binary neural network (BNN) chip explores the limits of energy efficiency and computational density for an all-digital deep neural network (DNN) inference accelerator. The chip intersperses data storage and computation using computation near memory (CNM) to reduce interconnect and data movement costs. It performs wide inner product operations to leverage parallelism inherent in DNN computations. The BNN chip leverages lightweight pipelining at a near-threshold voltage (NTV) to reduce the overhead of sequential elements. It employs optimized data access patterns to reduce memory accesses for convolutional operation with pooling layers. The combination of these techniques enables the BNN chip to achieve a peak energy efficiency of 617 TOPS/W. The digital BNN chip approaches the energy efficiency of analog in-memory techniques while also ensuring deterministic, scalable, and bit-accuracy operation. Moreover, the all-digital design leverages process scaling and does not require additional memory transistors or passive devices to attain a peak compute density of 418 TOPS/mm² and a memory density of 414 KB/mm². The binary design is extended to enable bit-serial integer precision operation with a reconfigurable 1-b multiplication circuit and element-wise partial sum shift and accumulate. This technique allows for fine-grain mixed precision and retains energy efficiency by exploiting parallelism inherent in DNNs. The bit-serial binary operation allows for bit-accurate operation and high DNN accuracy that multibit analog compute-in-memory designs struggle to attain. It provides favorable energy tradeoffs compared with small-integer digital DNN accelerators.

28 September 2023 in IEEE Journal of Solid-State Circuits
Verified
Journal Article
Closed access
CLChuan-Tung Lin
Chuan-Tung Lin
DWDewei Wang
Dewei Wang
BZBo Zhang
Bo Zhang
GCGregory K. Chen
Gregory K. Chen
PKPhil C. Knag
Phil C. Knag
MSMingoo Seok
Mingoo Seok
Recent SRAM-based in-memory computing (IMC) hardware demonstrates high energy efficiency and throughput for matrix–vector multiplication (MVM), the dominant kernel for deep neural networks (DNNs). Earlier IMC macros have employed analog-mixed-signal (AMS) arithmetic hardware. However, those so-called AIMCs suffer from process, voltage, and temperature (PVT) variations. Digital IMC (DIMC) macros, on the other hand, exhibit better robustness against PVT variations, but they tend to require more silicon area. This article proposes novel DIMC hardware featuring approximate arithmetic (DIMCA) to improve area efficiency without hurting compute density (CD). We also propose an approximation-aware training model and a customized number format to compensate for the accuracy degradation caused by the approximation hardware. We prototyped the test chip in 28-nm CMOS. It contains two versions: the DIMCA with single-approximate hardware (DIMCA1) and DIMCA with double-approximate hardware (DIMCA2). The measurement results show that DIMCA1 supports a 4 b-activation and 1 b-weight (4 b/1 b) CNN model, achieving 327 kb/mm2, 458–990 TOPS/W (normalized to 1 b/1 b), 8.27–392 TOPS/mm2 (normalized to 1 b/1 b), and 90.41% accuracy for CIFAR-10. DIMCA2 supports a 1 b/1 b CNN model, achieving 485 kb/mm2, 932–2219 TOPS/W, 14.4–607 TOPS/mm2, and 86.96% accuracy for CIFAR-10.

21 December 2022 in IEEE Journal of Solid-State Circuits
Verified
Journal Article
Closed access
GCGregory K. Chen
Gregory K. Chen
PKPhil C. Knag
Phil C. Knag
CTCarlos Tokunaga
Carlos Tokunaga
RKRam K. Krishnamurthy
Ram K. Krishnamurthy
An eight-core 64-b processor extends RISC-V to perform multiply–accumulate (MAC) within the shared last level cache (LLC). Instead of moving data from the LLC to the core, compute near last level cache (CNC) adds MAC to the LLC datapath and performs computation near where the data are stored. The RV64GC CNC instruction set architecture (ISA) extension performs digital MAC near unmodified SRAM arrays and has a low area overhead of 1.4%. CNC increases memory access width to 512 b per instruction by avoiding bottlenecks in the on- chip networks. The operation also reduces data movement by keeping MAC results and most input operands local to the LLC slices. CNC supports computation on cached data from main memory, coherent data sharing between cores, and virtual addressing. The CNC instructions are included in C++ programs and run either baremetal or in Linux. The 1.15-GHz chip reduces energy consumption by 52 $\times $ for fully connected and 29 $\times $ for convolutional deep neural network (DNN) layers, compared to scalar operation. Two benchmarks are characterized: MLPerf Tiny Anomaly Detection v0.5 latency is reduced by 4.25 $\times $ to 40 $\mu \text{s}$ versus previous work, and inference latency on memory-augmented neural networks is improved by 4.1 $\times $ versus scalar operation.

29 August 2022 in IEEE Transactions on Circuits and Systems I: Regular Papers
Verified
Journal Article
Closed access
ZZZhiyuan Zhou
Zhiyuan Zhou
NTNghia Tang
Nghia Tang
BNBai Nguyen
Bai Nguyen
WHWookpyo Hong
Wookpyo Hong
PPPartha Pratim Pande
Partha Pratim Pande
DHDeukhyoun Heo
Deukhyoun Heo
With the increasing complexity of highly integrated system on chips (SoCs), the power management system (PMS) is required to provide several power supplies efficiently for individual blocks. This paper presents a single-inductor multiple-output (SIMO) inductor-first hybrid converter that generates three outputs between 0.4V and 1.6V from a 1.8V input. The proposed multiple-output hybrid power stage can improve the conversion efficiency by reducing inductor current while extending the output voltage range compared with the existing hybrid topologies. In addition, the proposed converter employs an on-chip switched-capacitor power stage (SCPS) with a dual-switching frequency technique, resulting in a fast response time, low cross-regulation, and reduced number of on-chip pads. Measurement results show that the converter achieves a peak efficiency of 87.5% with the maximum output current of 450mA. The converter is integrated with a fast voltage regulation loop with 500MHz system clock to achieve a less than 0.01mA/mV cross-regulation and a maximum 20mV overshoot at full-load transient response. The design is fabricated in the standard 180nm CMOS technology.

16 February 2021 in IEEE Solid-State Circuits Letters
Verified
Journal Article
Closed access
DKDongkwun Kim
Dongkwun Kim
SKSung J. Kim
Sung J. Kim
ZJZhewei Jiang
Zhewei Jiang
SKSuhwan Kim
Suhwan Kim
ABAndres Blanco
Andres Blanco
MSMingoo Seok
Mingoo Seok
Emerging sub-mW near-threshold-voltage system-on-chips require new power management architecture that can create multiple voltage domains with the fewest possible off-chip passives. To fulfill this need, we propose an ultra-low-power single-inductor-multiple-output dc-dc buck converter in a 65-nm CMOS process. Featuring a comparator based output switch controller and a digital pulse-width modulation controller for ultra-low feedback latency, this dc-dc converter takes 1-V input voltage and produces ten independent output voltages from 0.4 to 0.8 V at the aggregated power delivery of 0.583 mW. It has ten 200-pF on-chip output capacitors and one 82-μH off-chip inductor and achieves the peak power conversion efficiency of 83.4%.

10 March 2021 in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
Verified
Journal Article
Closed access
SVSriram R. Vangal
Sriram R. Vangal
SPSomnath Paul
Somnath Paul
SHSteven Hsu
Steven Hsu
AAAmit Agarwal
Amit Agarwal
SKSaurabh Kumar
Saurabh Kumar
CKChris H. Kim
Chris H. Kim
The system-on-chip (SoC) designs for future Internet of Things (IoT) systems, spanning client platforms to cloud datacenters, need to deliver uncompromising and scalable performance with extreme energy efficiency for diverse workloads and applications, while satisfying a wide range of energy budgets, as well as platform cooling and power delivery constraints. Low-latency, burst-mode responsiveness, and scalable high-throughput performance must be delivered on demand for a range of thread-parallel, task-parallel, and data-parallel workloads covering traditional and emerging applications. This article discusses the challenges and opportunities for many-core SoC design in scaled CMOS process operating over a wide voltage-frequency range including near-threshold-voltage (NTV) that can meet the compute demands of the future at scale, flexibly, and efficiently. This article covers: 1) circuit design techniques for NTV cores; 2) mitigation techniques for within-die parameter variations via multivoltage frequency schemes; 3) digital integrated voltage regulators (VRs) for fine-grain and wide-range voltage modulation; and 4) radiation-induced soft error rate (SER) characterization and mitigation techniques to enable reliable operation at NTV. Silicon prototype examples will be used to illustrate the different techniques and highlight future research directions.