Searcharxiv⌕ Search

arXiv subjects

Miquel Pericas

Publications and source records attributed to Miquel Pericas.

10 recordsLinked to original sources
arXiv2026

Analytical Resource Management for Fine-grained MoE Computation-Communication Overlap

Fine-grained computation--communication overlap in distributed Mixture-of-Experts (MoE) inference allows communication to begin as partial compute results become ready. However, cooperative thread arrays (CTAs) performing computation and communication contend for finite residency capacity on streaming multiprocessors (SMs). Because a resident CTA generally retains its allocated SM resources until completion, CTAs that cannot be co-resident must wait for resources, resulting in wave-like execution. A fixed resource partition cannot adapt to changes in input size, routed expert load, and kernel configuration, potentially causing a communication backlog or reducing expert compute parallelism. We present a wave-quantized analytical model and launch-time resource manager for dependency-coupled overlap pipelines. Using routed-tile counts, kernel occupancy, GPU residency constraints, and split-level readiness dependencies, it selects the communication-CTA count and resource partition before each launch without candidate execution, per-workload profiling, or kernel recompilation. We integrate the method into the public COMET A100 implementation in FLUX. We evaluate three MoE models on four NVIDIA A100 GPUs under several parallelism strategies at the GEMM2+GatherRS operator, complete post-router MoE layer, and complete-model prefill levels. Across 15 real-p90 workloads, the analytical selector achieves 3.22 percent mean regret relative to the measured oracle with a mean solver overhead of 0.157 microseconds. Over COMET, our method achieves geometric-mean speedups of 2.528x at the GEMM2+GatherRS operator, 1.771x at the complete post-router MoE layer, and 1.185x for complete-model prefill, with maxima of 4.218x, 2.584x, and 1.439x, respectively. At every feasible TP=2/EP=2 sequence length of at least 4,096, our implementation outperforms COMET, Megatron core-TE, and FastMoE TP+NCCL.

cs.DC↗
arXiv2026

DEFT: Joint Task Placement and DVFS for Energy-Efficient Multi-GPU Runtimes

Energy efficiency has become a first-order concern in modern high-performance computing systems, as it directly determines achievable throughput under fixed power budgets. Although Dynamic Voltage and Frequency Scaling (DVFS) provides an effective mechanism for reducing GPU energy consumption, existing runtime systems decouple DVFS from task placement and inter-GPU communication, focus on single-GPU execution, or cannot adapt frequency to task granularity and runtime contention in multi-GPU environments. Consequently, current schedulers fail to capture the tight coupling between task placement, frequency selection, and inter-GPU data movement that fundamentally governs energy-performance trade-offs on multi-GPU systems. This paper presents DEFT, an energy-aware scheduling framework that jointly optimizes task-to-device assignment and per-GPU DVFS configuration for task-based multi-GPU applications. DEFT employs a cost-model-driven strategy that integrates slack awareness, throughput awareness, and explicit modeling of task execution cost, inter-GPU data movement, and DVFS transition overheads, enabling coordinated placement and frequency decisions at task granularity under dynamic runtime conditions. We prototype DEFT within the CUDASTF runtime and demonstrate its effectiveness across five optimization objectives. The evaluation shows that DEFT reduces energy consumption by 14.8% and 4.8% on average on NVIDIA L40S and L4, and reduces EDP by 9.9% and 3.7%, respectively, while maintaining performance within 1.5% of the fastest baseline.

cs.DC↗
arXiv2026

Resource-aware Computation-Communication Overlap for multi-GPU ML Workloads

The rapid growth of large-scale machine learning (ML) has made distributed training across multiple GPUs a fundamental component of modern ML systems. As model sizes and computational throughput continue to increase, communication overhead has become a dominant bottleneck in multi-GPU training, particularly when computation and communication are executed sequentially. This work explores concurrent execution of computation and collective communication using two portable runtime controls: shared-memory-driven occupancy shaping for computation kernels and elevated scheduling priority for communication kernels. Our approach regulates computation-kernel residency through per-block shared-memory allocation, leaving sufficient on-chip resources for communication kernels to make progress. In addition, assigning higher priority to communication streams ensures steady communication progress once resources become available. Experiments on NVIDIA A40, A100, H100, and AMD MI250X GPUs demonstrate that the proposed method enables effective computation-communication overlap and reduces total execution time by up to 25.5 percent, without modifying vendor libraries or kernel implementations.

cs.DC↗
arXiv2022

Accelerating CNN inference on long vector architectures via co-design

CPU-based inference can be an alternative to off-chip accelerators, and vector architectures are a promising option due to their efficiency. However, the large design space of convolutional algorithms and hardware implementations makes it challenging to select the best options. This paper presents ongoing research into co-designing vector architectures for CPU-based CNN inference, focusing on the im2col+GEMM and Winograd kernels. Using the Gem5 simulator, we examine the impact of various hardware microarchitectural features on RISC-V Vector and ARM-SVE ISAs. We also study the impact of several BLIS-like algorithmic optimizations on im2col+GEMM. Our co-design study shows that longer vector lengths and larger caches can improve performance by 5x with our optimized CNN kernels, compared to a vector length of 512-bit and 1MB of L2 cache. For Winograd, we present a novel approach of inter-tile parallelization that exploits longer vector lengths and offers high memory reuse, resulting in up to 2.4x performance improvement for non-strided convolutional layers with 3x3 kernel size. Our study also shows that Winograd requires smaller cache sizes compared to im2col+GEMM.

cs.DC↗
arXiv2021

Mitigating inefficient task mappings with an Adaptive Resource-Moldable Scheduler (ARMS)

Efficient runtime task scheduling on complex memory hierarchy becomes increasingly important as modern and future High-Performance Computing (HPC) systems are progressively composed of multisocket and multi-chiplet nodes with nonuniform memory access latencies. Existing locality-aware scheduling schemes either require control of the data placement policy for memory-bound tasks or maximize locality for all classes of computations, resulting in a loss of potential performance. While such approaches are viable, an adaptive scheduling strategy is preferred to enhance locality and resource sharing efficiency using a portable programming scheme. In this paper, we propose the Adaptive Resource-Moldable Scheduler (ARMS) that dynamically maps a task at runtime to a partition spanning one or more threads, based on the task and DAG requirements. The scheduler builds an online platform-independent model for the local and non-local scheduling costs for each tuple consisting of task type (function) and task topology (task location within DAG). We evaluate ARMS using task-parallel versions of SparseLU, 2D Stencil, FMM, and MatMul as examples. Compared to previous approaches, ARMS achieves up to 3.5x performance gain over state-of-the-art locality-aware scheduling schemes.

cs.DC↗
arXiv2020

Scheduling Task-parallel Applications in Dynamically Asymmetric Environments

Shared resource interference is observed by applications as dynamic performance asymmetry. Prior art has developed approaches to reduce the impact of performance asymmetry mainly at the operating system and architectural levels. In this work, we study how application-level scheduling techniques can leverage moldability (i.e. flexibility to work as either single-threaded or multithreaded task) and explicit knowledge on task criticality to handle scenarios in which system performance is not only unknown but also changing over time. Our proposed task scheduler dynamically learns the performance characteristics of the underlying platform and uses this knowledge to devise better schedules aware of dynamic performance asymmetry, hence reducing the impact of interference. Our evaluation shows that both criticality-aware scheduling and parallelism tuning are effective schemes to address interference in both shared and distributed memory applications

cs.DC↗
arXiv2019

Coordinated Management of DVFS and Cache Partitioning under QoS Constraints to Save Energy in Multi-Core Systems

Reducing the energy expended to carry out a computational task is important. In this work, we explore the prospects of meeting Quality-of-Service requirements of tasks on a multi-core system while adjusting resources to expend a minimum of energy. This paper considers, for the first time, a QoS-driven coordinated resource management algorithm (RMA) that dynamically adjusts the size of the per-core last-level cache partitions and the per-core voltage-frequency settings to save energy while respecting QoS requirements of every application in multi-programmed workloads run on multi-core systems. It does so by doing configuration-space exploration across the spectrum of LLC partition sizes and Dynamic Voltage Frequency Scaling (DVFS) settings at runtime at negligible overhead. We show that the energy of 4-core and 8-core systems can be reduced by up to 18% and 14%, respectively, compared to a baseline with even distribution of cache resources and a fixed mid-range core voltage-frequency setting. The energy savings can potentially reach 29% if the QoS targets are relaxed to 40% longer execution time.

cs.AR↗
arXiv2019

Coordinated Management of Processor Configuration and Cache Partitioning to Optimize Energy under QoS Constraints

An effective way to improve energy efficiency is to throttle hardware resources to meet a certain performance target, specified as a QoS constraint, associated with all applications running on a multicore system. Prior art has proposed resource management (RM) frameworks in which the share of the last-level cache (LLC) assigned to each processor and the voltage-frequency (VF) setting for each processor is managed in a coordinated fashion to reduce energy. A drawback of such a scheme is that, while one core gives up LLC resources for another core, the performance drop must be compensated by a higher VF setting which leads to a quadratic increase in energy consumption. By allowing each core to be adapted to exploit instruction and memory-level parallelism (ILP/MLP), substantially higher energy savings are enabled. This paper proposes a coordinated RM for LLC partitioning, processor adaptation, and per-core VF scaling. A first contribution is a systematic study of the resource trade-offs enabled when trading between the three classes of resources in a coordinated fashion. A second contribution is a new RM framework that utilizes these trade-offs to save more energy. Finally, a challenge to accurately model the impact of resource throttling on performance is to predict the amount of MLP with high accuracy. To this end, the paper contributes with a mechanism that estimates the effect of MLP over different processor configurations and LLC allocations. Overall, we show that up to 18% of energy, and on average 10%, can be saved using the proposed scheme.

cs.AR↗
arXiv2019

High performance scheduling of mixed-mode DAGs on heterogeneous multicores

Many HPC applications can be expressed as mixed-mode computations, in which each node of a computational DAG is itself a parallel computation that can be molded at runtime to allocate different amounts of processing resources. At the same time, modern HPC systems are becoming increasingly heterogeneous to address the requirements of energy efficiency. Effectively using heterogeneous devices is complex, requiring the developer to be aware of each DAG nodes' criticality, and the relative performance of the underlying heterogeneous resources. This paper studies how irregular mixed-mode parallel computations can be mapped on a single-ISA heterogeneous architecture with the goals of performance and portability. To achieve high performance we analyze various schemes for heterogeneous scheduling, including both criticality-aware and performance-only schemes, and extend them with task molding to dynamically adjust the amount of resources used for each task. To achieve performance portability we track each DAG nodes' performance and construct an online model of the system and its performance. Using a HiKey960 big.LITTLE board as experimental system, the resulting scheduler implementations achieve large speed-ups when executing irregular DAGs compared to traditional random work stealing.

cs.DC↗