Near-data processing (NDP) mitigates the data movement bottleneck in modern computing systems by performing computation close to where the data resides. Solid-state drives (SSDs) are well suited for NDP because they: (1) store large application datasets that exceed main memory capacity, and (2) contain multiple heterogeneous computation resources, e.g., general-purpose embedded cores in the SSD controller, DRAM chips, and NAND flash chips, which enable three NDP paradigms: in-storage processing (ISP), processing using DRAM in the SSD (PuD-SSD), and in-flash processing (IFP). These resources offer massive internal parallelism and enable in-place computation, which reduces data movement across the memory hierarchy. A large body of prior SSD-based NDP techniques operate in isolation, mapping computations to only one or two NDP paradigms (i.e., ISP, PuD-SSD, or IFP) within the SSD. These techniques (1) are tailored to specific workloads or kernels, (2) do not offload computations across all three NDP paradigms in the SSD and thus fail to exploit the full computational potential of an SSD, and (3) lack programmer-transparency, often requiring significant manual effort to identify offloadable code regions and map them to the SSD computation resources, which limits their general applicability and ease of deployment. While several prior works propose techniques to partition computation between the host and near-memory accelerators, adapting these techniques to SSDs offers limited benefits because they (1) ignore the heterogeneity of the SSD computation resources, and (2) make offloading decisions based on limited factors such as bandwidth utilization, data movement cost, or memory intensity, while ignoring key factors such as resource utilization. We propose Conduit, a general-purpose, programmertransparent NDP framework for SSDs that accelerates a broad range of workloads by leveraging available SSD computation resources. Conduit operates in two stages. At compile time, Conduit executes a custom compiler (e.g., LLVM) pass that (i) vectorizes suitable application code segments into single-instruction multiple-data (SIMD) operations that align with the SSD's page layout, and (ii) embeds metadata (e.g., operation type, operand sizes) into the vectorized instructions to guide runtime offloading decisions. At runtime, within the SSD, Conduit performs instruction-granularity offloading by evaluating six key application and system features (e.g., operation type, computation resource utilization, data dependence delay), and uses a cost function to select the most suitable SSD computation resource to execute each vectorized instruction. We evaluate Conduit and two prior NDP offloading techniques using an in-house event-driven SSD simulator on six data-intensive applications (e.g., large language model inference and training, encryption). Conduit outperforms the best-performing prior offloading policy by$1.8 \times$and reduces energy consumption by 46 %, with small latency and storage overheads, and no additional hardware cost.

Modern high-performance computing (HPC) environments rely on hybrid storage systems (HSS) that combine multiple storage devices with diverse latency, bandwidth, endurance, and capacity characteristics to meet the performance, capacity, and cost requirements of data-intensive applications. The performance of an HSS highly depends on two key data-management policies: (1) data placement, which determines the most suitable storage device to store application data, and (2) data migration, which dynamically reorganizes previously-stored data across storage devices (i.e., prefetching hot data and evicting cold data) to sustain high HSS performance. These policies are tightly interdependent, and thus, improving one without considering the other leads to suboptimal HSS performance. Unfortunately, prior works optimize only one of the policies.

Address translation is a major performance bottleneck in modern computing systems. Predicting the physical address (PA) of requested data before address translation completes can hide this latency, but accurate virtual address (VA)-to-PA prediction is difficult because conventional operating systems make VA-to-PA mappings unpredictable. Prior work improves predictability but relies on large pages or VA-to-PA contiguity, or stores speculation metadata in costly hardware structures. We introduce Revelator, a hardware-OS cooperative technique that uses hashing to enable accurate speculative address translation with small system modifications. Revelator employs a tiered hash-based memory allocation policy for both program data and last-level page table entries (PTEs), creating predictable VA-to-PA and VA-to-PTE mappings. After an L2 TLB miss, a lightweight hardware speculation engine uses the OS hash functions to predict these mappings and prefetch the corresponding cache blocks before translation completes, hiding address translation latency and accelerating page table walks (PTWs). Revelator does not rely on large pages or VA-to-PA contiguity and requires only small OS and hardware changes. Across 11 data-intensive workloads, Revelator improves performance by 15.3% on average over the state-of-the-art speculative address translation technique under high memory fragmentation. In virtualized environments, it predicts both guest and host physical addresses, providing a 13.6% average speedup over Nested Paging. In 16-core systems, Revelator achieves 1.40x (1.50x) speedup over Transparent Huge Pages across 30 server workload mixes from Google under medium (high) memory fragmentation. RTL synthesis shows only 0.02% area and 0.03% power overheads on a high-end server-grade CPU. Revelator is freely available at \href{https://github.com/CMU-SAFARI/Virtuoso}{github.com/CMU-SAFARI/Virtuoso}.

The unprecedented growth in data demand from emerging applications has turned virtual memory (VM) into a major performance bottleneck. VM's overheads are expected to persist as memory requirements continue to increase. Researchers explore new hardware/OS co-designs to optimize VM across diverse applications and systems. To evaluate such designs, researchers rely on various simulation methodologies to model VM components. Unfortunately, current simulation tools (i) either lack the desired accuracy in modeling VM's software components or (ii) are too slow and complex to prototype and evaluate schemes that span across the hardware/software boundary.

As Large Language Models (LLMs) gain traction, their reliance on power-hungry GPUs places ever-increasing energy demands, raising environmental and monetary concerns. Inference dominates LLM workloads, presenting a critical challenge for providers: minimizing energy costs under Service-Level Objectives (SLOs) that ensure optimal user experience. In this paper, we present throttLL’eM, a framework that reduces energy consumption while meeting SLOs through the use of instance and GPU frequency scaling. throttLL’eM features mechanisms that project future Key-Value (KV) cache usage and batch size. Leveraging a Machine-Learning (ML) model that receives these projections as inputs, throttLL’eM manages performance at the iteration level to satisfy SLOs with reduced frequencies and instance sizes. We show that the proposed ML model achieves $R^{2}$ scores greater than 0.97 and miss-predicts performance by less than 1 iteration per second on average. Experimental results on LLM inference traces show that throttLL’eM achieves up to $\mathbf{4 3. 8 \%}$ lower energy consumption and an energy efficiency improvement of at least $1.71 \times$ under SLOs, when compared to NVIDIA’s Triton server. throttLL’eM is publicly available at https://github.com/WilliamBlaskowicz/throttLL-eM.

High-Level Synthesis (HLS) has played a pivotal role in making FPGAs accessible to a broader audience by facilitating high-level device programming and rapid microarchitecture customization through the use of directives. However, manually selecting the right directives can be a formidable challenge for programmers lacking a hardware background. This article presents CollectiveHLS, an ultra-fast, knowledge-driven approach to optimizing HLS designs. It automates the identification and application of optimal directive configurations from the original source code, focusing on minimizing design latency and ensuring synthesizability. This optimization approach is entirely data-driven, offering a generalized HLS tuning solution without reliance on Quality of Result (QoR) models or meta-heuristics. CollectiveHLS is designed, implemented, and evaluated using around 60 applications sourced from well-established benchmark suites and GitHub repositories, all running on a Xilinx UltraScale + MPSoC ZCU104. It achieves an average geometric mean speedup of up to \(23.1\times\) compared to the official source code without directives, while maintaining synthesizability and feasibility rates of 100% and 96.6%, respectively, matching those of Vitis, the industry-standard framework for FPGA acceleration. Comparisons with resource over-provisioning, traditional genetic algorithm-based Design Space Exploration (DSE), and State-of-the-Art (SotA) approaches demonstrate that CollectiveHLS produces designs of comparable quality \(14.6\times\) faster on average. These results underscore the potential of our approach as an ultra-fast and automated solution for HLS optimization.

High-Level Synthesis (HLS) has played a pivotal role in making FPGAs accessible to a broader audience by facilitating high-level device programming and rapid microarchitecture customization through the use of directives. However, manually selecting the right directives can be a formidable challenge for programmers lacking a hardware background. This paper introduces an ultra-fast, knowledge-based HLS design optimization method that automatically extracts and applies the most promising directive configurations to the original source code. This optimization approach is entirely data-driven, offering a generalized HLS tuning solution without reliance on Quality of Result (QoR) models or meta-heuristics. We design, implement, and evaluate our methodology using over 100 applications sourced from well-established benchmark suites and GitHub repositories, all running on a Xilinx ZCU104 FPGA. The results are promising, including an average geometric mean speedup of ×7.2 and ×1.35 compared to designer-optimized designs and resource over-provisioning strategies, respectively. Additionally, it demonstrates a high design feasibility score and maintains an average inference latency of 38ms. Comparative analysis with traditional genetic algorithm-based Design Space Exploration (DSE) methods and State-of-the-Art (SoA) approaches reveals that it produces designs of similar quality but at speeds 2-3 orders of magnitude faster. This suggests that it is a highly promising solution for ultra-fast and automated HLS optimization.

Deep Neural Networks (DNNs) are becoming extremely popular for many modern applications deployed at the edge of the computing continuum. Despite their effectiveness, DNNs are typically resource intensive, making it prohibitive to be deployed on resource- and/or energy-constrained devices found in such environments. To overcome this limitation, partitioning and offloading part of the DNN execution from edge devices to more powerful servers has been introduced as a prominent solution. While previous works have proposed resource management schemes to tackle this problem, they usually neglect the high dynamicity found in such environments, both regarding the diversity of the deployed DNN models, as well as the heterogeneity of the underlying hardware infrastructure. In this paper, we present RoaD-RuNNer, a framework for DNN partitioning and offloading for edge computing systems. RoaD-RuNNer relies on its prior knowledge and leverages collaborative filtering techniques to quickly estimate performance and energy requirements of individual layers over heterogeneous devices. By aggregating this information, it specifies a set of Pareto optimal DNN partitioning schemes that trade-off between performance and energy consumption. We evaluate our approach using a set of well-known DNN architectures and show that our framework i) outperforms existing state-of-the-art approaches by achieving 9.58× speedup on average and up to 88.73% less energy consumption, ii) achieves high prediction accuracy by limiting the prediction error down to 3.19% and 0.18% for latency and energy, respectively and iii) provides lightweight and dynamic performance characteristics.