State-of-the-art high-throughput DNA sequencers output terabytes of short reads that typically need to be aligned to a reference genome in order to perform downstream analyses. Because alignment typically dominates the total run time of bioinformatics pipelines, a number of recent work sought to accelerate it in hardware. However, existing FPGA implemen-tations did not fully optimize the alignment algorithms for the FPGA hardware and mainly focused on a subset of alignment problems, e.g., ungapped alignment with a limited number of mismatches, which hinder their practical utility. In this work, we analyze the existing alignment methods and identify and leverage opportunities for FPGA acceleration. Our alignment framework, SALIENT, first carries out an ultra-fast ungapped alignment, which supports a flexible number of mismatches. Based on the underlying bioinformatics pipeline and the information provided by the ungapped aligner, SALIENT then identifies a fraction of reads that need to go through its gapped aligner, thus improving alignment throughput. We extensively evaluate SALIENT using diverse datasets. Experimental results indicate that SALIENT, running on a single Xilinx Alveo U280 device, delivers an average throughput of 546 million bases/second, outperforming the state- of-the-art minimap2 software by 40x, and Bowtie2 by up to 107 x, with a similar or slightly better (~O.l %-0.5 %) alignment and error (false negative/positive) rate. Compared to the existing ungapped FPGA aligners [1]–[4], SALIENT has 9.4-18x higher throughput/Watt, while compared to the gapped aligners [5], [6], it is 28–35 x better. SALIENT achieves 7.6 x higher throughput than Illumina DRAGEN Bio-IT Platform [7].

As Processing-In-Memory (PIM) hardware matures and starts making its way into normal compute platforms, software has an important role to play in determining what to perform where, and when, on such heterogeneous systems. Taking an emerging class of PIM hardware which provisions a general purpose (RISC-V) processor at each memory bank, this paper takes on this challenging problem by developing a software compilation framework. This framework analyzes several application characteristics - parallelizability, vectorizability, data set sizes, and offload costs - to determine what, whether, when and how to offload computations to the PIM engines. In the process, it also proposes a vector engine extension to the bank-level RISC-V cores. Using several off-the-shelf C/C++ applications, we demonstrate that PIM is not always a panacea, and a framework such as ours is essential in carefully selecting what needs to be performed where, when and how. The choice of hardware platforms - number of memory banks, relative speeds and capabilities of host CPU and PIM cores, can further impact the "to PIM or not" question.

This paper conducts a design space exploration of placing general purpose RISCV cores within the DDR DRAM hierarchy to boost the performance of important data analytics applications in the datacenter. We investigate the hardware (where? how many? how to interface?) and software (how to place data? how to map computations?) choices for placing these cores within the rank, chip, and bank of the DIMM slots to take advantage of the locality vs. parallelism trade-offs. We use the popular MapReduce paradigm, normally used to scale out workloads across servers, to scale in these workloads into the DDR DRAM hierarchy. We evaluate the design space using diverse off-the-shelf Apache Spark Workloads to show the pros-and-cons of different hardware placement and software mapping strategies. Results show that bank-level RISCV cores can provide tremendous speedup (up to 363X) for the offload-able parts of these applications, amounting to 14X speedup overall in some applications. Even in the non-amenable applications, we get at least 31% performance boost for the entire application. To realize this, we incur an area overhead of 4% at the bank level, and increase in temperature of < 4°C over the chip averaged over all applications.

Emerging resistive memory (RRAM) based crossbar array is a promising technology to accelerate neural network applications. RRAM-based CNN accelerators support a high-degree of intra-layer and inter-layer parallelism. The intra-layer parallelism duplicates kernels for each network layer while the inter-layer parallelism allows execution of each layer when a portion of input data is available. However, previously proposed RRAM-based accelerators do not leverage data sharing between duplicate kernels leading to significant idleness of crossbar arrays during inference. This shared data creates data dependencies that stall the processing of the next layer in the pipeline. To address these issues, we propose Fine-grained Parallel RRAM Architecture (FPRA), a novel architectural design, to improve parallelism for pipeline-enabled RRAM-based accelerators. FPRA addresses the data sharing issue with kernel batching and data sharing aware memory. Kernel batching rearranges the layout of the kernels and minimizes the data dependencies created by the input shared data. The data sharing aware memory uniformly buffers the input and output data for each layer, efficiently dispatching data to duplicate kernels while reducing the amount of data transferred between layers. We evaluate FPRA on eight popular image recognition CNN models with various configurations in a cycle-accurate simulator. We find that FPRA manages to achieve 2.0 $\times$ average latency speedup, and 2.1 $\times$ average throughput increase, as compared to the state-of-the-art RRAM-based accelerators.

In-situ approaches process data very close to the memory cells, in the row buffer of each subarray. This minimizes data movement costs and affords parallelism across subarrays. However, current in-situ approaches are limited to only row-wide bitwise (or few-bit) operations applied uniformly across the row buffer. They impose a significant overhead of multiple row activations for emulating 32-bit addition and multiplications using bitwise operations and cannot support operations with data dependencies or based on predicates. Moreover, with current peripheral logic, communication among subarrays is inefficient, and with typical data layouts, bits in a word are not physically adjacent. The key insight of this work is that in-situ, single-word ALUs outperform in-situ, parallel, row-wide, bitwise ALUs by reducing the number of row activations and enabling new operations and optimizations. Our proposed lightweight access and control mechanism, Fulcrum, sequentially feeds data into the single-word ALU and enables operations with data dependencies and operations based on a predicate. For algorithms that require communication among subarrays, we augment the peripheral logic with broadcasting capabilities and a previously-proposed method for low-cost inter-subarray data movement. The sequential processor also enables overlapping of broadcasting and computation, and reuniting bits that are physically adjacent. In order to realize true subarray-level parallelism, we introduce a lightweight column-selection mechanism through shifting one-hot encoded values. This technique enables independent column selection in each subarray. We integrate Fulcrum with Compress Express Link (CXL), a new interconnect standard. Fulcrum with one memory stack delivers on average (up to) 23.4 (76) speedup over a server-class GPU, NVIDIA P100, with three stacks of HBM2 memory, (ii) 70 (228) times speedup per memory stack over the GPU, and (iii) 19 (178.9) times speedup per memory stack over an ideal model of the GPU, which only accounts for the overhead of data movement.

This paper proposes an energy-efficient deep in-memory architecture for NAND flash (DIMA-F) to perform machine learning and inference algorithms on NAND flash memory. Algorithms for data analytics, inference, and decision-making require processing of large data volumes and are hence limited by data access costs. DIMA-F achieves energy savings and throughput improvement for such algorithms by reading and processing data in the analog domain at the periphery of NAND flash memory. This paper also provides behavioral models of DIMA-F that can be used for analysis and large scale system simulations in presence of circuit non-idealities and variations. DIMA-F is studied in the context of linear support vector machines and k-nearest neighbor for face detection and recognition, respectively. An estimated 8×-to-23× reduction in energy and 9×-to-15× improvement in throughput resulting in EDP gains up to 345× over the conventional NAND flash architecture incorporating an external digital ASIC for computation.

In this paper, we propose the concept of compute memory, where computation is deeply embedded into the memory (SRAM). This deep embedding enables multi-row read access and analog signal processing. Compute memory exploits the relaxed precision and linearity requirements of pattern recognition applications. System-level simulations incorporating various deterministic errors from analog signal chain demonstrates the limited accuracy of analog processing does not significantly degrade the system performance, which means the probability of pattern detection is minimally impacted. The estimated energy saving is 63 % as compared to the conventional system with standard embedded memory and parallel processing architecture, for 256×256 target image.