hpc.md

HPC (High Performance Computing) bookmarks

• Learning and practice of high performance computing

• CFD

  • Algebraic Flux Correction I. Scalar Conservation Laws

  • Algebraic Flux Correction II. Compressible Euler Equations

  • CFD Notes by Hiroaki Nishikawa

    • CFD codes in f90

  • Tim Warburton's github repositories

    • Nodal Discontinuous Galerkin

  • Hybrid and Easy Discontinuous Galerkin Environment

  • Element-based-Galerkin-Methods

  • Extreme-scale Discontinuous Galerkin Environment (EDGE)

  • The Development, Verification, and Validation of a Discontinuous Galerkin Method for the Navier-Stokes Equations

  • What are possible methods to solve compressible Euler equations

  • I do like CFD, VOL.1, Second Edition

  • Free CFD Codes

  • CFD Julia: A Learning Module Structuring an Introductory Course on Computational Fluid Dynamics

    • CFD_Julia

  • Lattice Boltzmann codes

    • A lattice Boltzmann code for complex fluids

  • Hydrodynamics in OpenCL

    • Roe, HLL, HLLC, Burgers Scheme; 1D, 2D, 3D; Euler Equation, (~Maxwell), (~MHD), ADM Solver in OpenCL
    • Differential Geometry Tensor Library

  • Implementing the discontinuous Galerkin method in CUDA

    • Marty Fuhry's Homepage
      • Master's Thesis: An Implementation of the Discontinuous Galerkin Method on Graphics Processing Units, defended May, 2013
      • A GPU-accelerated adaptive discontinuous Galerkin method for level set equation
      • A GPU Accelerated Discontinuous Galerkin Incompressible Flow Solver
    • The Development, Verification, and Validation of a Discontinuous Galerkin Method for the Navier-Stokes Equations

  • SU2

    • github repo for SU2

  • HiFiLES: High Fidelity Large Eddy Simulation

    • github repo for HiFiLES

  • PyWENO + PyPFASST

  • Clawpack Repositories

  • Riemann Problems and Jupyter Solutions

  • Shenfun is a high performance computing platform for solving partial differential equations (PDEs) by the spectral Galerkin method

  • Multilayered Abstractions for Partial Differential Equations by Graham Robert Markall: good review on Nektar++ etc.

    • Making Faster FEM Solvers, Faster MPhil Transfer Report By Graham Markall
    • Graham Markall

  • Nektar++

    • Nektar++: An efficient h to p finite element framework
    • Nektar++: a high-order finite element framework

  • Simple (and not-so-simple) CFD solvers written in Fortran with Python plotting routines

  • MAESTROeX solves the equations of low Mach number hydrodynamics for stratified atmospheres/full spherical stars with a general equation of state, and nuclear reaction networks in an adaptive-grid finite-volume framework. It includes reactions and thermal diffusion and can be used on anything from a single core to 100,000s of processor cores with MPI + OpenMP or 1,000s of GPUs

    • Model stars and atomspheres with MAESTROeX

  • Is there a good tutorial or textbook-like source on implementing ENO/WENO with limiters in one (and more than one) dimension?

    • PyWENO

  • blitzdg is an open-source library offering discontinuous Galerkin (dg) solvers for common partial differential equations systems using blitz++ for array and tensor manipulations in a C++ environment or NumPy as a Python 3 library

    • Derek Steinmoeller's blog

  • DGSWEM V2

  • adaptive multiresolution DG

  • Exasim - Generating Discontinuous Galerkin Codes For Extreme Scalable Simulations

  • Chebyshev Pseudo-Spectral Method (PSM)

    • Chebyshev Polynomials J.C. Mason D.C. Handscomb

      • Chapter 1. Definitions
      • Chapter 2. Basic Properties and Formulae
      • Chapter 3. The Minimax Property and Its Applications
      • Chapter 4. Orthogonality and Least-Squares Approximation
      • Chapter 5. Chebyshev Series
      • Chapter 6. Chebyshev Interpolation
      • Chapter 7. Near-Best L∞, L1 and Lp Approximations
      • Chapter 8. Integration Using Chebyshev Polynomials
      • Chapter 9. Solution of Integral Equations
      • Chapter 10. Solution of Ordinary Differential Equations
      • Chapter 11. Chebyshev and Spectral Methods for Partial DifferentialEquations
      • Chapter 12. Conclusion
      • Appendices:
      • Summary of Notations, Definitions and ImportantProperties
      • Tables of Coefficients
      • FFTW Discrete Cosine Transform Derivative
      • FAST ALGORITHMS FOR DISCRETE POLYNOMIAL TRANSFORMS
      • A New Method for Chebyshev Polynomial Interpolation Based on Cosine Transforms

    • A brief introduction to pseudo-spectral methods: application to diffusion problems

    • Spectral methods in python

    • An Introduction to Domain Decomposition Methods:algorithms, theory and parallel implementation

    • Chebyshev-Legendre Spectral Domain Decomposition Method for Two-Dimensional Vorticity Equations

    • Domain Decomposition Methods for Mortar Finite Elements

    • An efficient domain-decomposition pseudo-spectral method for solving elliptic differential equations

    • A Pseudospectral Multi-Domain Method for the Incompressible Navier-Stokes Equations

    • Deep Domain Decomposition Method: Elliptic Problems

    • How to Design an Efficient Pseudospectral Code

      • code for How to Design an Efficient Pseudospectral Code

    • Dedalus is a framework for solving a broad range of partial differential equations using spectral methods, including initial-value, boundary-value, and generalized eigenvalue problems

      • Dedalus is a flexible framework for solving partial differential equations using spectral methods

    • multiple-interval pseudospectral methods to solve optimal control problems

    • pizza is a high-performance numerical code for quasi-geostrophic and non-rotating convection in a 2-D annulus geometry

    • FDBB (Fluid Dynamics Building Blocks) is a C++ expression template library for fluid dynamics

      • FDBB - Fluid Dynamics Building Blocks

  • Finite Element Methods (FEM) and Spectral Element Methods (SEM)

    • deal.II — an open source finite element library
      • Amandus: Simulations based on multilevel Schwarz methods Documentation
    • Feel++ finite element embedded library in C++
      • Feel++: Finite Element Embedded Library in C++
    • Veamy: an extensible object-oriented C++ library for the virtual element method
      • Veamy: an extensible object-oriented C++ library for the virtual element method
    • Two dimensional high-order spectral element method fluid dynamics solver
      • Two dimensional high-order spectral element method fluid dynamics solver
    • github: ITHACA-SEM - In real Time Highly Advanced Computational Applications for Spectral Element Methods
      • THACA-SEM - In real Time Highly Advanced Computational Applications for Spectral Element Methods
    • AxiSEM is a parallel spectral-element method to solve 3D wave propagation in a sphere with axisymmetric or spherically symmetric visco-elastic, acoustic, anisotropic structures
    • HDGlab: An open-source implementation of the hybridisable discontinuous Galerkin method in MATLAB
      • HDGlab - A Matlab implementation of the hybridisable discontinuous Galerkin (HDG) method
    • Euler Equations for Ideal Gases
      • Split form nodal discontinuous Galerkin schemes with summation-by-parts property for the compressible Euler equations

  • Siemens

    • Embedded Multicore Building Blocks (EMB²)

  • Maxeler

    • Maxeler Technologies - Maximum Performance Computing
      • AirfoilDFE - An unstructured mesh finite volume solver on DFE.
      • Lattice QCD is a discretization of Quantum Chromodynamics
      • LatticeBoltzmann

  • facilities to experiment with Discontinuous Petrov Galerkin (DPG) methods

    • Research papers of Jay Gopalakrishnan

  • Free CFD codes

    • Code_Saturne
    • Large-Scale CFD Parallel Computing Dealing with Massive Mesh016c0bd28b2435d468ce3cd1771426de9f264af6

  • Open source tools in technical photorealistic large-scale visualisation

  • An Open Source CFD-DEM Perspective

  • 3D, block structured, explicit/implicit, Navier-Stokes solver

    • An evaluation of the Eigen linear algebra library for use in the aither CFD solver
      • A look at the performance of expression templates in C++: Eigen vs Blaze vs Fastor vs Armadillo vs XTensor

  • CFD + GPU

    • Recent progress and challenges in exploiting graphics processors in computational fluid dynamics: slightly outdated but interesting
      • Laplace solver running on GPU using CUDA, with CPU version for comparison, slightly outdated
    • PyFR
      • PyFR is an open-source Python based framework for solving advection-diffusion type problems on streaming architectures using the Flux Reconstruction approach of Huynh
        • vincentlab/PyFR
      • New PyFR Paper “Heterogeneous Computing on Mixed Unstructured Grids with PyFR”
      • PyFR: An open source framework for solving advection–diffusion type problems on streaming architectures using the flux reconstruction approach
      • High Performance Parallelism Pearls Volume Two: Multicore and Many-core Programming Approaches

  • Camellia

    • Camellia Discontinuous Petrov-Galerkin github repository

• Co-design at Lawrence Livermore National Lab

  • Livermore Unstructured Lagrangian Explicit Shock Hydrodynamics (LULESH)
  • DoE Exascale Co-Design Center for Materials in Extreme Environments : Extreme Materials at Extreme Scale
    • Programming Models - Languages and tools for developing multi-scale applicatins.
    • Terra is a new low-level system programming language that is designed to interoperate seamlessly with the Lua programming language

• List of quantum chemistry and solid-state physics software

  • CP2K
    • Mirror of official svn repository at sourceforge. Synced every 5 minutes.
    • Accelerated Sparse Matrix Multiplication for Quantum Chemistry with CP2K on Hyprid Supercomputers

• Evaluation of C, Go, and Rust in the HPC environment

• Modern Fortran

  • NNSA, national labs team with Nvidia to develop open-source Fortran compiler technology
  • Flang is a ground-up implementation of a Fortran front end written in modern C++. It started off as the f18 project
    • F18 is a front-end for Fortran intended to replace the existing front-end in the Flang compiler
      tl;dr 301 moved The code from this repository can now be found at flang
    • Flang and F18
    • Installing LLVM Flang Fortran compiler
      tl;dr

    git clone https://github.com/llvm/llvm-project
    mkdir -p llvm-project/build
    cd llvm-project/build
    cmake ../llvm -DLLVM_ENABLE_PROJECTS=flang

      + [Unknown CMake command “tablegen”](https://stackoverflow.com/questions/59691069/unknown-cmake-command-tablegen)
    

• libCEED: the CEED Library: Code for Efficient Extensible Discretization

  • CEED Library: Code for Efficient Extensible Discretization
    • MFEM is a free, lightweight, scalable C++ library for finite element methods
      • MFEM is a free, lightweight, scalable C++ library for finite element methods: examples
      • GPU support in MFEM
      • Finite Element Discretization Library __ _ __ ___ / | ___ _ __ ___ | ' _ \ | |_ / _ \| '_ _
        | | | | | || || __/| | | | | | || || |||| ___||| || ||
      • High-order Lagrangian Hydrodynamics Miniapp

• Modern trends in programming of GPUs DAQFEET 2021

  • Toward Performance-Portable PETSc for GPU-based Exascale Systems

• AMG

  • AMG intro
    • Iteration methods
    • Algebraic multigrid method by smoothed agglomeration for a Stokes problem
    • Convergence of Algebraic Multigrid Based on Smoothed Aggregation II: Extension to a Petrov-Galerkin Method
    • Lawrence Livermore National Laboratory Robert D. Falgout Center for Applied Scientific Computing An Algebraic Multigrid Tutorial
    • An Introduction to Algebraic Multigrid
    • An Algebraic Multigrid Tutorial IMA Tutorial – FastSolution Techniques November28-29, 2010
    • Multigrid Methods: From Geometrical to Algebraic Versions Gundolf HAASE
    • A root-node based algebraic multigrid method
    • Iterative methods for linear, non-linear and eigenvalue problems
    • A Multigrid Tutorial by William L. Briggs
    • Algebraic Multigrid Code
    • Performance of Preconditioners for Large-Scale Simulations Using Nek5000
    • Reducing Complexity in Parallel Algebraic Multigrid Preconditioners, Hans de Sterck, Ulrike Meier Yang and Jeffrey J. Heys
    • 3.2.5. Block Compressed Sparse Row Format (BSR)
      • I don't find the LU decomposition on the device with cuSolver
  • AMGX
    • AMGX in Julia
    • pyamgx: Python interface to NVIDIA's AMGX library
      • pyamgx - GPU accelerated multigrid library for Python
    • AmgXWrapper
      • An example and benchmark of AmgX and PETSc with Poisson system
      • PetIBM - toolbox and applications of the immersed-boundary method on distributed-memory architectures
      • geoclaw-landspill
      • High-productivity, high-performance workflow for virus-scale electrostatic simulations with Bempp-Exafmm
      • Alexa: Simulating Shock Hydrodynamics on the GPU using Kokkos
      • GPGPU acceleration a case study of algebraic multigrid preconditioned GMRES
      • AmgX: A Library for GPU Accelerated Algebraic Multigrid and Preconditioned Iterative Methods
        • Comparison of AMGX and Hypre
    • rocALUTION is a sparse linear algebra library with focus on exploring fine-grained parallelism
  • amgcl
    • amgcl
    • C++ library for solving large sparse linear systems with algebraic multigrid method
    • Triggering C++11 support in NVCC with CMake
      tl;dr

    diff --git a/CMakeLists.txt b/CMakeLists.txt
    index 6ca3264..b63e326 100644
    --- a/CMakeLists.txt
    +++ b/CMakeLists.txt
    @@ -161,9 +161,9 @@ if(CMAKE_CXX_COMPILER_ID MATCHES "GNU" OR CMAKE_CXX_COMPILER_ID MATCHES "MSVC")
     
             if (CMAKE_CXX_COMPILER_ID MATCHES "GNU")
                list(APPEND CUDA_NVCC_FLAGS
    -                ${CUDA_ARCH_FLAGS} -std=c++11 -Wno-deprecated-gpu-targets)
    +                ${CUDA_ARCH_FLAGS} -std=c++17 -Wno-deprecated-gpu-targets)
     
    -            list(APPEND CUDA_NVCC_FLAGS -Xcompiler -std=c++11 -Xcompiler -fPIC -Xcompiler -Wno-vla)
    +            list(APPEND CUDA_NVCC_FLAGS -Xcompiler -std=c++17 -Xcompiler -fPIC -Xcompiler -Wno-vla)
             endif()
     
             add_library(cuda_target INTERFACE)
    

    • Stokes problem gives NaN by AMG but GMRES works fine
    • Pressure projection solver for Incompressible Navier-Stokes FEM
    • how to perform matrix construction in GPU deveces without the data transfer
    • Block preconditioners
      • amg_corrector_solver
      • schur pressure correction
        • code accompanying "Accelerating linear solvers for Stokes problems with C++ metaprogramming"
          • Accelerating linear solvers for Stokes problems with C++ metaprogramming
  • SPARSH-AMG
    • SPARSH-AMG: A LIBRARY FOR HYBRID CPU-GPU ALGEBRAIC MULTIGRID AND PRECONDITIONED ITERATIVE METHODS
  • Ginkgo is a high-performance linear algebra library for manycore systems, with a focus on sparse solution of linear systems. It is implemented using modern C++ (you will need at least C++14 compliant compiler to build it), with GPU kernels implemented in CUDA and HIP. HAS support for AMG
    • GPGPU acceleration - a case study of algebraic multigrid preconditioned GMRES
  • BootCMatchG
  • multigrid solver for solving elliptic PDEs using finite differences on a rectangular grid
  • Multigrid HowTo (Part I): A simple Multigrid solver in C++ in less than 200 lines of code
  • Multigrid HowTo (Part II): An Open Source Algebraic Multigrid Solver in C++
    • Multigrid solver prototype (GMG) and simple Lid Cavity solver
  • ExaStencils: Advanced Multigrid Solver Generation
    • EvoStencils - Constructing efficient multigrid solvers through evolutionary computation

• Sparse Linear System Solvers on GPUs

  • SPARSE LINEAR SYSTEM SOLVERS ON GPUS: PARALLEL PRECONDITIONING, WORKLOAD BALANCING, AND COMMUNICATION REDUCTION
  • High performance sparse multifrontal solvers on modern GPUs
    • STRUMPACK -- STRUctured Matrix PACKage, Copyright (c) 2014-2021

• Как SpaceX использует GPU для обсчёта ракетных двигателей

  • Rockets Shake And Rattle, So SpaceX Rolls Homegrown CFD

• Modern C++ Parallel Task Programming

  • docs for Modern C++ Parallel Task Programming

• Freud, a tool to create Performance Annotations for C/C++ programs

• Eyal Rozenberg, Ph.D.

  • Eyal Rozenberg
    • Thin C++-flavored wrappers for the CUDA APIs: Runtime, Driver, NVRTC and NVTX
    • GPU Kernel Runner

• RAPIDS - Open GPU Data Science

  • RAFT: Reusable Accelerated Functions and Tools
  • cuDF - GPU DataFrames
    tl;dr

    cd cpp && mkdir -p build && cd build && cmake .. -DCMAKE_BUILD_TYPE=Release  -DOPENSSL_INCLUDE_DIR=/usr/include/openssl  -DOPENSSL_CRYPTO_LIBRARY=/usr/lib/libcrypto.so -DOPENSSL_SSL_LIBRARY=/usr/lib/libssl.so

  • cuSpatial - GPU-Accelerated Spatial and Trajectory Data Management and Analytics Library

• CUDA rehab & NVidia docs

  • Documentation of NVIDIA chip/hardware interfaces

  • CS344 : CUDA Programming in C

  • UD281 : High Performance Computing

    • Parallel Computer Architecture and Programming (CMU 15-418/618)
    • Parallel Computer Architecture and Programming (CMU 15-418/618)
      • CMU 15418 Assignment 1: Analyzing Program Performance on a Multi-Core CPU
      • Assignment 1: Analyzing Program Performance on a Multi-Core CPU
      • Assignment 2: A Simple CUDA Renderer

  • Course on CUDA Programming on NVIDIA GPUs, July 22-26, 2019

    • Lecture 3: control flow and synchronisation: Warp divergence
      • Is branch divergence really so bad?
    • Lecture 5: libraries and tools
    • Maximizing Unified Memory Performance in CUDA
    • CUDA OPTIMIZATION TIPS, TRICKS AND TECHNIQUES Stephen Jones, GTC 2017
    • HIGH THROUGHPUT WITH GPUS
    • Small tips of optimizing CUDA programs
    • Error using __ldg in cuda kernel at compile time tl;dr

    nvcc -arch=sm_35 ...

  • Open-Arch-Group

    • Matrix Multiplication (MMul) Benchmarks
    • Performance engineer that's always happy to answer questions!
      • GPGPU Programming with CUDA
        • From Scratch: Histograms in CUDA using Atomics
      • Parallel Programming in Modern C++
        • This program shows off the basics of stop tokens in C++20

  • Matrix multiplication in cuSparse (cusparseDcsrgemm) outputs wrong results

    • C++ (Cpp) cusparseDcsrgemm примеры использования
    • Problem of two large sparse matrices multiplication in cuParse
    • spgemm_example.c
    • CusparseManager.cu

  • how to cast thrust::device_vector to raw pointer

  • Параллельные вычисления с использованием стандартов MPI, OpenMP, OpenACC

• Memory Model

  • C++11 introduced a standardized memory model. What does it mean? And how is it going to affect C++ programming?
  • A Primer on Memory Consistency and Cache Coherence

• LPC2018 - Open Source GPU compute stack - Not dancing the CUDA dance

• OpenCL

  • OpenCL 3.0 Specification Released With New Khronos Open-Source OpenCL SDK
  • The State of OpenCL for Scientific Computing in 2018
  • OpenCL: History & Future
  • Tuned OpenCL BLAS
    • CLBlast:ATunedBLASLibrary forFasterDeepLearning
  • OpenCL vloadn
  • Could not find a package configuration file provided by "OpenCLHeaders"
  • Using OpenCL on Adreno & Mali GPUs is slower than CPU
  • Zero copy buffer allocation on arm mali midgard gpus?
  • SYCL - C++ Single-source Heterogeneous Programming for OpenCL
    • Khronos SYCL
    • An open-source implementation of OpenCL SYCL from Khronos Group
    • codeplaysoftware
      • SYCL BLAS
      • SYCL DNN
      • SYCL VisionCpp
    • Implementation of the SYCL specification.
      • Building a brain with SYCL and modern C++

• OneAPI

  • Run simple DPC++ application
  • oneAPI Direct Programming
  • Port a CUDA App to oneAPI and DPC++ in 5 Minutes
  • How to run dpc++ code on Intel HD Graphic atop Nvidia GPU

• Kompute

  • The general purpose GPU compute framework for cross vendor graphics cards (AMD, Qualcomm, NVIDIA & friends)
    • Kompute github repo

• HCC is an Open Source, Optimizing C++ Compiler for Heterogeneous Compute currently for the ROCm GPU Computing Platform

  • Why did AMD open source ROCm’s OpenCL driver-stack?
  • wiki for HCC
  • github HCC repository
  • Portable Computing Language
  • A collection of Arch Linux PKGBUILDS for the ROCm platform tl;dr

  yay -S rocm-opencl-runtime

  • aur package rocm-opencl-runtime

  • Arch GPGPU

    • Arch ROCm
      • ROCm for Arch Linux

  • rocm OpenCL Programming Guide

    • clinfo ERROR: clBuildProgram(-11)
    • rock-dkms kernel vs mainline clarification
    • Error during installation of rock-dkms 4.0 on 5.4 kernel
    • dkms build on unsported kernel and supported which makes errors
    • ROCm support in upstream Linux kernels
    • Information for rock-dkms
    • Radeon ROCm 4.1 Released - Still Without RDNA GPU Support
    • ROCm 4.1 - Vega 20 (Radeon VII) with upstream amdgpu
    • AMD dkms fails

    dkms install --no-depmod -m amdgpu-4.0 -v 23 -k 5.11.16-arch1-1
    Error! Bad return status for module build on kernel: 5.11.16-arch1-1 (x86_64)
    Consult /var/lib/dkms/amdgpu-4.0/23/build/make.log for more information.
    ==> Warning, `dkms install --no-depmod -m amdgpu-4.0 -v 23 -k 5.11.16-arch1-1' returned 10
    
    pacman -Qo /usr/src/amdgpu-4.0-23
    /usr/src/amdgpu-4.0-23/ принадлежит rock-dkms-bin 4.0-3
    /usr/src/amdgpu-4.0-23/ принадлежит rock-dkms-firmware-bin 4.0-3
    

  • Radeon Instinct like : Radeon VII

    • RTX 2080 vs. Radeon VII vs. 5700 XT: Rendering and Compute Performance
    • AMD Radeon VII Review: This Isn’t the 7nm GPU You’re Looking For
    • Is a used Radeon VII worth it in 2020?
      • AMD Radeon Instinct MI50 1725MHz PCI-E 4.0 16384MB 1000MHz 4096 bit

  • hipSYCL - a SYCL implementation for CPUs and GPUs

    • hipSYCL performance

  • OpenCL => Vulkan

    • a prototype implementation of OpenCL 1.2 on top of Vulkan using clspv as the compiler
    • clspv is a prototype compiler for a subset of OpenCL C to Vulkan compute shaders

• How To Set The CPU Affinity Of A Running Process In Linux

• OpenMP

  • Ждали, ждали и дождались! OpenMP 4.0
  • Parallelization of a prefix sum (Openmp)
    • Parallel Prefix Sum (Scan) with CUDA
    • Parallel prefixsum algorithm in fastflow
    • GPU prefix scan

• OpenACC

  • IPMACC is a framework for translating/executing OpenACC for C API to/over CUDA or OpenCL runtime
  • IPMACC – An Open Source OpenACC to CUDA/OpenCL Translator

• MATOG - GPU Access Auto Tuning

  • MATOG Auto-Tuning on GPUs is a tool to automatically optimize performance of NVIDIA CUDA code
  • MATOG preprint
  • MATOG: CUDA Array Access Auto-Tuner

• OCCA (Open Concurrent Compute Abstraction)

  • github repository for OCCA

• LCSE - Linked Cluster Series Expansions - a framework for high-temperature series expansions

• VLI is a llibrary for high but fixed (128 to 512-bit) arithmetic and symbolic polinomials computations

  • Series Expansion Methods for Quantum Lattice Models

• Apache Arrow

  • Apache Arrow is a development platform for in-memory analytics. It contains a set of technologies that enable big data systems to process and move data fast.

• Sandia

  • Trilinos is a collection of open-source software libraries, called packages, intended to be used as building blocks for the development of scientific applications.
    • github repo fo Trilinos
      tl;dr

    $ yay -s trilinos
    3 aur/trilinos 12.14.1-2 (+0 0.00%) 
        algorithms for the solution of large-scale scientific problems
    2 aur/mingw-w64-trilinos 12.12.1-1 (+0 0.00%) 
        Framework for the solution of large-scale, complex multi-physics engineering and scientific problems (mingw-w64)
    1 aur/trilinos-git 12.12.0.gd3b096f4f1-1 (+1 0.00%) (Out-of-date 2019-06-21) 
        An effort to develop algorithms and enabling technologies within an object-oriented software framework for the solution of large-scale, complex multi-physics engineering and scientific problems.
    

    • Add option to turn off the install of gtest header and lib even if Gtest package is enabled

• ARM

  • The ARM Computer Vision and Machine Learning library
  • HPCG for Arm
    • Parallelizing HPCG's main kernels
  • ARM Neon
    • Coding for ARM NEON: How to start?
    • SIMD Assembly Tutorial:ARM NEON
    • ARM NEON скининг

• CPU, GPU & DRAM Architecture Simulators

  • GPGPU-Sim
  • Integrated gem5 + GPGPU-Sim Simulator
    • Getting gem5
  • SimpleScalar LLC
    • SimpleScalar LLC Intro
      • Todd Austin : the author
  • DRAMSim2
    • github repos for DRAMSim2 etc. from University of Maryland
      • Write-back vs Write-Through
      • Study of Different Cache Line Replacement Algorithms in Embedded Systems
  • Chisel: Constructing Hardware in a Scala Embedded Language
    • UC Berkeley Architecture Research
      • The RISC-V Instruction Set Architecture
        • Rocket Chip Generator
      • Rocket Microarchitectural Implementation of RISC-V ISA
      • Rocket uncore: L2 cache, etc.

CUDA and friends related surveys, papers

• A Survey of CPU-GPU Heterogeneous Computing Techniques
• Гибридная реализация алгоритма MST с использованием CPU и GPU
• Понимание конфликтов банков разделяемой (shared) памяти в NVIDIA CUDA
• Vulkan: The next Khronos graphics API… that is not OpenGL
• AMD supported project: HIP : Convert CUDA to Portable C++ Code
  • Examples for HIP

DSLs targeting GPU

• CARP: Correct and Efficient Accelerator Programming

  • CARP dessimination
    • A taste of CARP: benchmark analysis, language design and kernel verification
  • PENCIL: a C99-based intermediate language for compute & optimization
    • PENCIL summary in one slide: poster
    • PENCIL: A Platform-Neutral Language for Accelerator Programming
    • PENCIL support in pet and PPCG
  • see also PPCG (below)

• Framework for performance-portable parallel computations on unstructured meshes

  • OP2: Developing an open-source framework for the execution of unstructured grid applications
  • Optimising Unstructured Mesh Computational Fluid Dynamics Applications on Multicores via Machine Learning and Code Transformation
  • Compiler Optimizations for Industrial Unstructured Mesh CFD Applications on GPUs

• Copperhead Data Parallel Python

  • github CU copperhead

• Delite

• Scalan

  • Scalan Community Edition

• Generating Performance Portable Code using Rewrite Rules: From High-level Functional Expressions to High-Performance OpenCL Code

• Performance Comparison of GPU, DSP and FPGA implementations of image processing and computer vision algorithms in embedded systems, Fykse, Egil

• ROSE compiler + Mint for C-to-CUDA code generation

  • ROSE compiler github
  • MINT
    • ROSE project MINT
    • MINT google project
      • Mint: Realizing CUDA performance in 3D Stencil Methods with Annotated C: claims 78% of handwritten CUDA performance
      • MINT PhD thesis

• Nested Data Parallelism, Haskell, and friends

  • Nested Data Parallelism on GPU
  • Compiling a high-level language for GPUs: (via language support for architectures and compilers)
  • NOVA: A Functional Language for Data Parallelism
  • CuNesl: Compiling Nested Data-Parallel Languages for ...
  • A Haskell EDSL for Nested Data-parallel Design-space ...
  • Functional programming for nested data parallelism on GPUs
  • Platform-Specific Optimization and Mapping of Stencil Codes through Refinement
  • High-Performance Domain-Specific Languages for GPU Computing
  • Monoids and their efficiency in practice

• CUDA kernels generation using C++ expression templates technique

  • CU++ -- an interesting approach
    • CU++, An Object Oriented Tool for CFD Applications: GTC 2012
    • CU++(ET) / UGC- CUDA With C++ Expression Templates with the Unified GPU-CPU Compiler
    • A Hybrid Multi-GPU/CPU Computational Framework
  • VexCL is a C++ vector expression template library for OpenCL/CUDA
    • VexCL is a C++ vector expression template library for OpenCL/CUDA
    • Generating OpenCL/CUDA source code from C++ expressions in VexCL

• AnyDSL - A Framework for Rapid Development of Domain-Specific Libraries; thorin (The Higher-ORder INtermediate representation) / impala (An imperative and functional programming language)

  • A Framework for Rapid Development of Domain-Specific Libraries
    • AnyDSL Build Instructions
    • Shallow Embedding of DSLs via Online Partial Evaluation.(Best Paper Award)
  • thorin - The Higher-ORder INtermediate representation
  • impala - An imperative and functional programming language
  • A DSL for Stencil Codes
  • AnyDSL ports from http://benchmarksgame.alioth.debian.org

parallelforall

• An Efficient Matrix Transpose in CUDA C/C++

• BIDMach: Machine Learning at the Limit with GPUs

• High-Performance Geometric Multi-Grid with GPU Acceleration

• Inside Pascal: NVIDIA’s Newest Computing Platform

• GPU Programming in Functional Languages

• HIP : Convert CUDA to Portable C++ Code

Pencil computations

• Ускоряем трафаретные вычисления: сборка и запуск YASK на процессорах Intel

• flexible package manager that supports multiple versions, configurations, platforms, and compilers. https://spack.io

• Tutorial: Spack 101

• NASA: High Performance Fast Computing Challenge

• Why Rust fails hard at scientific computing

  • Why Rust fails hard at scientific computing
    • technicalities: interactive scientific computing #2 of 2, goldilocks languages

Nim links

• Laser - Primitives for high performance computing

• NimTorch

• A matrix library https://unicredit.github.io/neo/

• A fast, ergonomic and portable tensor library with a deep learning focus

• high performance tensor library in Nim

  • Arraymancer - A n-dimensional tensor (ndarray) library

• A curated list of awesome Nim frameworks, libraries and software

  • Find the nim package

• Meta Nim Are we scientists yet?

• Quantum EXpressions lattice field theory framework

  • QEX: a framework for lattice field theories

• tl;dr

nimble refresh
nimble install neo
nimble install Arraymancer

• Why is nim and nimble in official repo so outdated?

• parallel-computing resources list

• Portable Hardware Locality (hwloc)

• Overview of the Efficient Programming Languages (v.3) 2018.4

• Intel Level Zero

  • oneAPI Level Zero

• Code Generation for High Performance PDE Solvers on Modern Architectures

  • PhD Thesis Software Stack
  • Loopy: Transformation-Based Generation of High-Performance CPU/GPU Code
  • HyperHDG - a C++ based library implementing hybrid discontinuous Galerkin methods on extremely general domains

• GPU roof model

  • Elias Konstantinidis publications
    • A quantitative roofline model for GPU kernel performance estimation using micro-benchmarks and hardware metric profiling
      • mixbench - The purpose of this benchmark tool is to evaluate performance bounds of GPUs on mixed operational intensity kernels
  • Analysis-Driven Optimization: Preparing for Analysis with NVIDIA Nsight Compute, Part 1
  • GPU Performance Analysis
  • Roofline and NVIDIA Ampere GPU Architecture Analysis
  • Nsight Compute Feature Spotlight: Roofline Analysis, Asynchronous Copy, Sparse Data Compression
  • Optimizing CUDA Memory Allocations Using NVIDIA Nsight Systems
  • Roofline Hackathon 2020 part 1 and 2

• YouTube videos on GPU embedded profiling/optimization

  • Presentation: Mali Graphics Debugger (GDC 2014)
  • GPU Compute Optimisation with Hardware Counters
  • ARM Mali GPU Architecture Overview

• AMD Radeon and NVIDIA GeForce FP32/FP64 GFLOPS Table

• RICOS Co. Ltd. Research Institute for Computational Science Co.Ltd.

• Load-link/store-conditional