Search NASA⌕ Search

SEARCH · Search NASA

Results for “GEMM”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

32 records · Page 2
All sourcesNASA NTRSDOE OSTI
DOE OSTI2018

Attacking the Opioid Epidemic: Determining the Epistatic and Pleiotropic Genetic Architectures for Chronic Pain and Opioid Addiction

We describe the CoMet application for large-scale epistatic Genome-Wide Association Studies (eGWAS) and pleiotropy studies. High performance is attained by transforming the underlying vector comparison methods into highly performant generalized distributed dense linear algebra operations. The 2-way and 3-way Proportional Similarity metric and Custom Correlation Coefficient are implemented using native or adapted GEMM kernels optimized for GPU architectures. By aggressive overlapping of communications, transfers and computations, high efficiency with respect to single GPU kernel performance is maintained up to the full Titan and Summit systems. Nearly 300 quadrillion element comparisons per second and over 2.3 mixed precision ExaOps are reached on Summit by use of Tensor Core hardware on the Nvidia Volta GPUs. Performance is four to five orders of magnitude beyond comparable state of the art. CoMet is currently being used in projects ranging from bioenergy to clinical genomics, including for the genetics of chronic pain and opioid addiction.

Joubert, Wayne↗
DOE OSTI2022

A Length Adaptive Algorithm-Hardware Co-design of Transformer on FPGA Through Sparse Attention and Dynamic Pipelining

Transformers are considered one of the most important deep learning models since 2018, in part because it establishes state-of-the-art (SOTA) records and could potentially replace existing Deep Neural Networks (DNNs). Despite the remarkable triumphs, the prolonged turnaround time of Transformer models is a widely recognized roadblock. The variety of sequence lengths imposes additional computing overhead where inputs need to be zero-padded to the maximum sentence length in the batch to accommodate the parallel computing platforms. This paper targets the field-programmable gate array (FPGA) and proposes a coherent sequence length adaptive algorithm–hardware co-design for Transformer acceleration. Particularly, we develop a hardware-friendly sparse attention operator and a length-aware hardware resource scheduling algorithm. The proposed sparse attention operator brings the complexity of attention-based models down to linear complexity and alleviates the off-chip memory traffic. The proposed length-aware resource hardware scheduling algorithm dynamically allocates the hardware resources to fill up the pipeline slots and eliminates bubbles for NLP tasks. Experiments show that our design has very small accuracy loss and has 80.2 × and 2.6 × speedup compared to CPU and GPU implementation, and 4 × higher energy efficiency than state-of-the-art GPU accelerator optimized via CUBLAS GEMM.

Peng, Hongwu↗
DOE OSTI2023

Evaluation of OpenAI Codex for HPC Parallel Programming Models Kernel Generation

We evaluate AI-assisted generative capabilities on fundamental numerical kernels in high-performance computing (HPC), including AXPY, GEMV, GEMM, SpMV, Jacobi Stencil, and CG. We test the generated kernel codes for a variety of language-supported programming models, including (1) C++ (e.g., OpenMP [including offload], OpenACC, Kokkos, SyCL, CUDA, and HIP), (2) Fortran (e.g., OpenMP [including offload] and OpenACC), (3) Python (e.g., numpy, Numba, cuPy, and pyCUDA), and (4) Julia (e.g., Threads, CUDA.jl, AMDGPU.jl, and KernelAbstractions.jl). We use the GitHub Copilot capabilities powered by the GPT-based OpenAI Codex available in Visual Studio Code as of April 2023 to generate a vast amount of implementations given simple + + prompt variants. To quantify and compare the results, we propose a proficiency metric around the initial 10 suggestions given for each prompt. Results suggest that the OpenAI Codex outputs for C++ correlate with the adoption and maturity of programming models. For example, OpenMP and CUDA score really high, whereas HIP is still lacking. We found that prompts from either a targeted language such as Fortran or the more general purpose Python can benefit from adding code keywords, while Julia prompts perform acceptably well for its mature programming models (e.g., Threads and CUDA.jl). We expect for these benchmarks to provide a point of reference for each programming model's community. Overall, understanding the convergence of large language models, AI, and HPC is crucial due to its rapidly evolving nature and how it is redefining human-computer interactions.

Godoy, William↗
DOE OSTI2023

Mixed-Precision S/DGEMM Using the TF32 and TF64 Frameworks on Low-Precision AI Tensor Cores

Using NVIDIA graphics processing units (GPUs) equipped with Tensor Cores has enabled the significant acceleration of general matrix multiplication (GEMM) for applications in machine learning (ML) and artificial intelligence (AI) and in high-performance computing (HPC) generally. The use of such power-efficient, specialized accelerators can provide a performance increase between 8 × and 20 ×, albeit with a loss in precision. However, a high level of precision is required in many large scientific and HPC applications, and computing in single or double precision is still necessary for many of these applications to maintain accuracy. Fortunately, mixed-precision methods can be employed to maintain a higher level of numerical precision while also taking advantage of the performance increases from computing with lower-precision AI cores. With this in mind, we extend the state of the art by using NVIDIA’s new TF32 framework. This new framework not only burdens some constraints of the previous frameworks, such as costly 32 16-bit castings but also provides an equivalent precision and performance by using a much simpler approach. We also propose a new framework called TF64 that attempts double-precision arithmetic with low-precision Tensor Cores. Although this framework does not exist yet, we validated the correctness of this idea and achieved an equivalent of 64-bit precision on 32-bit hardware.

Valero Lara, Pedro↗
DOE OSTI2024

A graphics processing unit accelerated sparse direct solver and preconditioner with block low rank compression

We present the GPU implementation efforts and challenges of the sparse solver package STRUMPACK. The code is made publicly available on github with a permissive BSD license. STRUMPACK implements an approximate multifrontal solver, a sparse LU factorization which makes use of compression methods to accelerate time to solution and reduce memory usage. Multiple compression schemes based on rank-structured and hierarchical matrix approximations are supported, including hierarchically semi-separable, hierarchically off-diagonal butterfly, and block low rank. Here, in this paper, we present the GPU implementation of the block low rank (BLR) compression method within a multifrontal solver. Our GPU implementation relies on highly optimized vendor libraries such as cuBLAS and cuSOLVER for NVIDIA GPUs, rocBLAS and rocSOLVER for AMD GPUs and the Intel oneAPI Math Kernel Library (oneMKL) for Intel GPUs. Additionally, we rely on external open source libraries such as SLATE (Software for Linear Algebra Targeting Exascale), MAGMA (Matrix Algebra on GPU and Multi-core Architectures), and KBLAS (KAUST BLAS). SLATE is used as a GPU-capable ScaLAPACK replacement. From MAGMA we use variable sized batched dense linear algebra operations such as GEMM, TRSM and LU with partial pivoting. KBLAS provides efficient (batched) low rank matrix compression for NVIDIA GPUs using an adaptive randomized sampling scheme. The resulting sparse solver and preconditioner runs on NVIDIA, AMD and Intel GPUs. Interfaces are available from PETSc, Trilinos and MFEM, or the solver can be used directly in user code. We report results for a range of benchmark applications, using the Perlmutter system from NERSC, Frontier from ORNL, and Aurora from ALCF. For a high frequency wave equation on a regular mesh, using 32 Perlmutter compute nodes, the factorization phase of the exact GPU solver is about 6.5× faster compared to the CPU-only solver. The BLR-enabled GPU solver is about 13.8× faster than the CPU exact solver. For a collection of SuiteSparse matrices, the STRUMPACK exact factorization on a single GPU is on average 1.9× faster than NVIDIA’s cuDSS solver.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

Relating Oxidative Protein Damage to Antioxidant Status in Health and Disease (Full Technical Report for 24-LW-026)

This two-year project evaluated how dietary antioxidants influence oxidative damage in cancer using complementary analytical and in-vivo approaches. We initially developed a protein oxidation labeling workflow and a parallel accelerator and molecular mass spectrometry (PAMMS) quantification method, but ultimately discontinued the labeling strategy due to unresolved separation challenges; PAMMS was instead leveraged to quantify radiolabeled catechol in rat plasma as a methodological benchmark. The biological study used a genetically engineered murine model (GEMM) for breast cancer (n = 40; four groups of 10: cancer/high antioxidant diet, cancer/normal diet, healthy/high-antioxidant diet, healthy/normal diet). In lieu of the abandoned labeling assay, untargeted metabolomics profiled plasma across groups, revealing widespread treatment-dependent changes in metabolites.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2024

autoGEMM: Pushing the Limits of Irregular Matrix Multiplication on Arm Architectures

This paper presents an open-source library that pushes the limits of performance portability for irregular General Matrix Multiplication (GEMM) on the widely-used Arm architectures. Our library, autoGEMM, is designed to support a wide range of Arm processors: from edge devices to HPC-grade CPUs. autoGEMM generates optimized kernels for various hardware configurations by auto-combining fragments of autogenerated micro-kernels that employ hand-written optimizations to maximize computational efficiency. We optimize the kernel pipeline by tuning the register reuse and the data load/store overlapping. In addition, we use a dynamic tiling scheme to generate balanced tile shapes. Finally, we position autoGEMM on top of the TVM framework where our dynamic tiling scheme prunes the search space for TVM to identify the optimal combination of parameters for code optimization. Evaluations on five different classes of Arm chips demonstrate the advantages of autoGEMM. For small matrices, autoGEMM achieves 98% of peak and up to 2.0x speedup over state-of-the-art libraries such as LIBXSMM and LibShalom. For irregular matrices (i.e. tall skinny and long rectangles), autoGEMM is 1.3-2.0x faster than widely-used libraries such as OpenBLAS and Eigen. autoGEMM is publicly available at: https://github.com/wudu98/autoGEMM.

Wu, Du↗
DOE OSTI2020

Measurements of Single Event Upset in ATLAS IBL

Effects of Single Event Upsets (SEU) and Single Event Transients (SET) are studied in the FE-I4B chip of the innermost layer of the ATLAS pixel system. SEU/SET affect the FE-I4B Global Registers as well as the settings for the individual pixels, causing, among other things, occupancy losses, drops in the low voltage currents, noisy pixels, and silent pixels. Quantitative data analysis and simulations indicate that SET dominate over SEU on the load line of the memory. Operational issues and mitigation techniques are presented.

47 OTHER INSTRUMENTATION↗
DOE OSTI2020

New high precision measurements of the cosmic charged particle rate beyond the Arctic Circle with the PolarquEEEst experiment

The goal of the PolarquEEEst experiment was to measure the cosmic charged particle rate at latitudes greater than 66° N, where no systematic and accurate measurements at sea level have ever been performed. A latitude range well above the Arctic Circle was explored on board of a sailboat, up to the unprecedented northernmost value of 82°07' N. In this paper a description of the experimental set-up is reported, then the procedures for calibration and data analysis are described in detail. The results show that the rate measured in this latitude range stays constant within a novel accuracy of ±1%.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2021

The cosmic muon and detector simulation framework of the extreme energy events (EEE) experiment

Abstract This paper describes the simulation framework of the extreme energy events (EEE) experiment. EEE is a network of cosmic muon trackers, each made of three multi-gap resistive plate chambers (MRPC), able to precisely measure the absolute muon crossing time and the muon integrated angular flux at the ground level. The response of a single MRPC and the combination of three chambers have been implemented in a GEANT4-based framework (GEMC) to study the telescope response. The detector geometry, as well as details about the surrounding materials and the location of the telescopes have been included in the simulations in order to realistically reproduce the experimental set-up of each telescope. A model based on the latest parametrization of the cosmic muon flux has been used to generate single muon events. After validating the framework by comparing simulations to selected EEE telescope data, it has been used to determine detector parameters not accessible by analysing experimental data only, such as detection efficiency, angular and spatial resolution.

Abbrescia, M.↗
DOE OSTI2018

The INFN R&D: New pixel detector for the High Luminosity upgrade of the LHC

The High Luminosity upgrade of the CERN-LHC (HL-LHC) demands for a new high-radiation tolerant solid-state pixel sensor capable of surviving fluencies up to a few 10 16 particles/cm 2 at ~3 cm from the interaction point. To this extent the INFN ATLAS-CMS joint research activity, in collaboration with Fondazione Bruno Kessler-FBK, is aiming at the development of thin n-in-p type pixel sensors for the HL-LHC. The R&D covers both planar and single-sided 3D columnar pixel devices made with the Si-Si Direct Wafer Bonding technique, which allows for the production of sensors with 100 μm and 130 μm active thickness for planar sensors, and 130 μm for 3D sensors, the thinnest ones ever produced so far. The first prototypes of hybrid modules bump-bonded to the present CMS and ATLAS readout chips have been tested in beam tests. Furthermore, the preliminary results on their performance before and after irradiation are presented.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
NASA NTRS2018

Mars Habitat Commonality: CPP-HAB

This proposal addresses the challenge to create a habitation system that has commonality in both the in-space and surface habitat designs so the crew will be familiar with the layout, function, and location of everything in the surface habitat when they arrive on Mars.

Michael Fox↗