Refine
Year of publication
Document Type
- In Proceedings (43)
- Article (21)
- ZIB-Report (11)
- Book chapter (2)
- Other (2)
- In Collection (1)
- Poster (1)
Keywords
- GPGPU (2)
- Bioinformatics (1)
- C-H...O interaction (1)
- Concurrent Kernel Execution (1)
- Connected Component Labeling (1)
- Fault-tolerance (1)
- Hyper-Q (1)
- Ising Model (1)
- Oversubscription (1)
- Performance and usage measurement (1)
Institute
- Distributed Algorithms and Supercomputing (76)
- Modeling and Simulation of Complex Processes (3)
- ZIB Allgemein (3)
- Parallel and Distributed Computing (2)
- Visual Data Analysis (2)
- Visual and Data-centric Computing (2)
- Image Analysis in Biology and Materials Science (1)
- Numerical Mathematics (1)
- Supercomputing (1)
- Visual Data Analysis in Science and Engineering (1)
Solving PDEs on unstructured grids is a cornerstone of engineering and scientific computing. Heterogeneous parallel platforms, including CPUs, GPUs, and FPGAs, enable energy-efficient and computationally demanding simulations.
In this article, we introduce the HPM C++-embedded DSL that bridges the abstraction gap between the mathematical formulation of mesh-based algorithms for PDE problems on the one hand and an increasing number of heterogeneous platforms with their different programming models on the other hand.
Thus, the HPM DSL aims at higher productivity in the code development process for multiple target platforms. We introduce the concepts as well as the basic structure of the HPM DSL, and demonstrate its usage with three examples. The mapping of the abstract algorithmic description onto parallel hardware, including distributed memory compute clusters, is presented.
A code generator and a matching back end allow the acceleration of HPM code with GPUs. Finally, the achievable performance and scalability are demonstrated for different example problems.
This work provides a brief description of Omni-Path Express and the current status of its development, stability, and performance. Basic benchmarks that highlight the gains of OPX over PSM2 are provided, and the results of an initial performance and scalability study of several applications are presented.
Next-Generation Sequencing technologies generate a vast and exponentially increasing amount of sequence data. The Interleaved Bloom Filter (IBF) is a novel indexing data structure which is state-of-the-art for distributing approximate queries with an in-memory data structure. With it, a main task of sequence analysis pipelines, (approximately) searching large reference data sets for sequencing reads or short sequence patterns like genes, can be significantly accelerated. To meet performance and energy-efficiency requirements, we chose a co-design approach of the IBF data structure on the FPGA platform. Further, our OpenCL-based implementation allows a seamless integration into the widely used SeqAn C++ library for biological sequence analysis. Our algorithmic design and optimization strategy takes advantage of FPGA-specific features like shift register and the parallelization potential of many bitwise operations. We designed a well-chosen schema to partition data across the different memory domains on the FPGA platform using the Shared Virtual Memory concept. We can demonstrate significant improvements in energy efficiency of up to 19x and in performance of up to 5.6x, respectively, compared to a well-tuned, multithreaded CPU reference.
Version 4.0 of the Message Passing Interface standard introduced the concept of Partitioned Communication
which adds support for multiple contributions to a communication buffer. Although initially targeted at multithreaded MPI applications, Partitioned Communication currently receives attraction in the context of accelerators, especially GPUs. In this publication it is demonstrated that this communication concept can also be implemented for SYCL-programmed FPGAs. This includes a discussion of the design space and the presentation of a prototypical implementation. Experimental results show that a lightweight implementation on top of an existing MPI library is possible. In addition, the presented approach also reveals issues in both the SYCL and the MPI standard which need to be addresses for improved support of the intended communication style.
The locality of solution features in cardiac electrophysiology simulations calls for adaptive methods. Due to the overhead incurred by established mesh refinement and coarsening, however, such approaches failed in accelerating the computations. Here we investigate a different route to spatial adaptivity that is based on nested subset selection for algebraic degrees of freedom in spectral deferred correction methods. This combination of algebraic adaptivity and iterative solvers for higher order collocation time stepping realizes a multirate integration with minimal overhead. This leads to moderate but significant speedups in both monodomain and cell-by-cell models of cardiac excitation, as demonstrated at four numerical examples.
Performance Evaluation and Improvements of the PoCL Open-Source OpenCL Implementation on Intel CPUs
(2021)
The Portable Computing Language (PoCL) is a vendor independent open-source OpenCL implementation that aims to support a variety of compute devices in a single platform. Evaluating PoCL versus the Intel OpenCL implementation reveals significant performance drawbacks of PoCL on Intel CPUs – which run 92 % of the TOP500 list. Using a selection of benchmarks, we identify and analyse performance issues in PoCL with a focus on scheduling and vectorisation. We propose a new CPU device-driver based on Intel Threading Building Blocks (TBB), and evaluate LLVM with respect to automatic compiler vectorisation across work-items in PoCL. Using the TBB driver, it is possible to narrow the gap to Intel OpenCL and even outperform it by a factor of up to 1.3× in our proxy application benchmark with a manual vectorisation strategy.
Solving partial differential equations on unstructured grids is a cornerstone of engineering and scientific computing. Nowadays, heterogeneous parallel platforms with CPUs, GPUs, and FPGAs enable energy-efficient and computationally demanding simulations. We developed the HighPerMeshes C++-embedded Domain-Specific Language (DSL) for bridging the abstraction gap between the mathematical and algorithmic formulation of mesh-based algorithms for PDE problems on the one hand and an increasing number of heterogeneous platforms with their different parallel programming and runtime models on the other hand. Thus, the HighPerMeshes DSL aims at higher productivity in the code development process for multiple target platforms. We introduce the concepts as well as the basic structure of the HighPer-Meshes DSL, and demonstrate its usage with three examples, a Poisson and monodomain problem, respectively, solved by the continuous finite element method, and the discontinuous Galerkin method for Maxwell’s equation. The mapping of the abstract algorithmic description onto parallel hardware, including distributed memory compute clusters is presented. Finally, the achievable performance and scalability are demonstrated for a typical example problem on a multi-core CPU cluster.
Recently, Intel released the oneAPI programming environment. With Data Parallel C++ (DPC++), oneAPI enables codes to target multiple hardware architectures like multi-core CPUs, GPUs, and even FPGAs or other hardware using a single source. For legacy codes that were written for Nvidia GPUs, a compatibility tool is provided which facilitates the transition to the SYCL-based DPC++ programming language. This paper presents early experiences when using both the compatibility tool and oneAPI as well the employed extension to the SYCL programming standard for the tsunami simulation code easyWave. A performance study compares the original code running on Xeon processors using OpenMP as well as CUDA with the performance of the DPC++ counter part on multicore CPUs as well as integrated GPUs.
Large capacity Storage Class Memory (SCM) opens new possibilities for workloads requiring a large memory footprint. We examine optimization strategies for a legacy Fortran application on systems with an heterogeneous memory configuration comprising SCM and DRAM. We present a performance study for the multigrid solver component of the large-eddy simulation framework PALM for different memory configurations with large capacity SCM. An important optimization approach is the explicit assignment of storage locations depending on the data access characteristic to take advantage of the heterogeneous memory configuration. We are able to demonstrate that an explicit control over memory locations provides better performance compared to transparent hardware settings. As on aforementioned systems the page management by the OS appears as critical performance factor, we study the impact of different huge page settings.
The NEC SX-Aurora TSUBASA is a new generation of vector processing architectures that combines a standard Intel Xeon host with the newly developed NEC Vector Engine co-processor cards. One way to use these co-processors is offloading suitable parts of the program from the host to the Vector Engines. Currently, the only vendor-provided offloading solutions are the low-level Vector Engine Offloading (VEO) library, and a builtin reverse-offloading mechanism named VHcall.
In this work, we extend the portable Heterogeneous Active Messages (HAM) based HAM-Offload framework with support for the NEC SX-Aurora TSUBASA. Therefore, we design, implement, and evaluate two messaging protocols aimed at minimising offloading cost. This sheds some light on how to achieve fast communication between host CPU and the Vector Engines of the NEC SX-Aurora TSUBASA. Compared with VEO, the DMA-based protocol reduces offloading overhead by a factor of 13×. The resulting framework enables users to write portable offload applications with low overhead, that do neither require a language extension like OpenMP, nor a special language like OpenCL. Existing HAM-Offload applications are now ready to run on the NEC SX-Aurora TSUBASA.
Field-programmable gate arrays (FPGAs) are of great interest for future high-performance computing and data analytics systems, since they are capable of efficient, highly-parallel data processing. Even though high-level synthesis became more popular in the last years, the effort of porting existing scientific software onto FPGAs is still considerable. We propose to use OpenMP target offloading as a solution, which we implement in a first prototype, making use of the preexisting OpenCL SDK of the FPGA vendor. Early results demonstrate the feasibility of this approach and also reveal that further optimizations will be necessary such that code can be written in an FPGA-agnostic way.
Computing the Hierarchical Equations of Motion (HEOM) is by itself a challenging problem, and so is writing portable production code that runs efficiently on a variety of architectures while scaling from PCs to supercomputers. We combined both challenges to push the boundaries of simulating quantum systems, and to evaluate and improve methodologies for scientific software engineering.
Our contributions are threefold: We present the first distributed memory implementation of the HEOM method (DM-HEOM), we describe an interdisciplinary development workflow, and we provide guidelines and experiences for designing distributed, performance-portable HPC applications with MPI-3, OpenCL and other state-of-the-art programming models. We evaluated the resulting code on multi- and many-core CPUs as well as GPUs, and demonstrate scalability on a Cray XC40 supercomputer for the PS I molecular light harvesting complex.
The Vienna Ab initio Simulation Package (VASP) is a widely used electronic structure code that originally exploits process-level parallelism through the Message Passing Interface (MPI) for work distribution within and across nodes.
Architectural changes of modern parallel processors urge programmers to address thread- and data-level parallelism as well to benefit most from the available compute resources within a node.
We describe for VASP how to approach for an MPI + OpenMP parallelization including data-level parallelism through OpenMP SIMD constructs together with a generic high-level vector coding scheme.
We can demonstrate an improved scalability of VASP and more than 20% gain over the MPI-only version, as well as a 2x increased performance of collective operations using the multiple-endpoint MPI feature.
The high-level vector coding scheme applied to VASP's general gradient approximation routine gives up 9x performance gain on AVX512 platforms with the Intel compiler.
Large-eddy simulations (LES) with the new ICOsahedral Non-hydrostatic atmosphere model (ICON) covering Germany are evaluated for four days in spring 2013 using observational data from various sources. Reference simulations with the established Consortium for Small-scale Modelling (COSMO) numerical weather prediction model and further standard LES codes are performed and used as a reference. This comprehensive evaluation approach covers multiple parameters and scales, focusing on boundary-layer variables, clouds and precipitation. The evaluation points to the need to work on parametrizations influencing the surface energy balance, and possibly on ice cloud microphysics. The central purpose for the development and application of ICON in the LES configuration is the use of simulation results to improve the understanding of moist processes, as well as their parametrization in climate models. The evaluation thus aims at building confidence in the model's ability to simulate small- to mesoscale variability in turbulence, clouds and precipitation. The results are encouraging: the high-resolution model matches the observed variability much better at small- to mesoscales than the coarser resolved reference model. In its highest grid resolution, the simulated turbulence profiles are realistic and column water vapour matches the observed temporal variability at short time-scales. Despite being somewhat too large and too frequent, small cumulus clouds are well represented in comparison with satellite data, as is the shape of the cloud size spectrum. Variability of cloud water matches the satellite observations much better in ICON than in the reference model. In this sense, it is concluded that the model is fit for the purpose of using its output for parametrization development, despite the potential to improve further some important aspects of processes that are also parametrized in the high-resolution model.
The effectiveness of ahead-of-time compiler optimization heavily depends on the amount of available information at compile time. Input-specific information that is only available at runtime cannot be used, although it often determines loop counts, branching predicates and paths, as well as memory-access patterns. It can also be crucial for generating efficient SIMD-vectorized code. This is especially relevant for the many-core architectures paving the way to exascale computing, which are more sensitive to code-optimization. We explore the design-space for using input-specific information at compile-time and present KART, a C++ library solution that allows developers tocompile, link, and execute code (e.g., C, C++ , Fortran) at application runtime. Besides mere runtime compilation of performance-critical code, KART can be used to instantiate the same code multiple times using different inputs, compilers, and options. Other techniques like auto-tuning and code-generation can be integrated into a KART-enabled application instead of being scripted around it. We evaluate runtimes and compilation costs for different synthetic kernels, and show the effectiveness for two real-world applications, HEOM and a WSM6 proxy.
The computational power and availability of graphics processing units (GPUs), such as the Nvidia Tesla, and Many Integrated Core (MIC) processors, such as the Intel Xeon Phi, on high performance computing (HPC) systems is rapidly evolving. However, HPC applications need to be ported to take advantage of such hardware. This paper is a report on our experience of porting the MPI+OpenMP parallelised large-eddy simulation model (PALM) to multi-GPU as well as to MIC processor environments using the directive-based high level programming paradigm OpenACC and OpenMP, respectively. PALM is a Fortran-based computational fluid dynamics software package, used for the simulation of atmospheric and oceanic boundary layers to answer questions linked to fundamental atmospheric turbulence research, urban modelling, aircraft safety and cloud physics. Development of PALM started in 1997, the project currently entails 140 kLOC and is used on HPC farms of up to 43,200 cores. The main challenges we faced during the porting process are the size and complexity of the PALM code base, its inconsistent modularisation and the complete lack of a unit-test suite. We report the methods used to identify performance issues as well as our experiences with state-of-the-art profiling tools. Moreover, we outline the required porting steps in order to properly execute our code on GPUs and MIC processors, describe the problems and bottlenecks that we encountered during the porting process, and present separate performance tests for both architectures. These performance tests, however, do not provide any benchmark information that compares the performance of the ported code between the two architectures.
The effectiveness of ahead-of-time compiler optimization heavily depends on the amount of available information at compile time. Input-specific information that is only available at runtime cannot be used, although it often determines loop counts, branching predicates and paths, as well as memory-access patterns. It can also be crucial for generating efficient SIMD-vectorized code. This is especially relevant for the many-core architectures paving the way to exascale computing, which are more sensitive to code-optimization. We explore the design-space for using input-specific information at compile-time and present KART, a C++ library solution that allows developers to compile, link, and execute code (e.g., C, C++ , Fortran) at application runtime. Besides mere runtime compilation of performance-critical code, KART can be used to instantiate the same code multiple times using different inputs, compilers, and options. Other techniques like auto-tuning and code-generation can be integrated into a KART-enabled application instead of being scripted around it. We evaluate runtimes and compilation costs for different synthetic kernels, and show the effectiveness for two real-world applications, HEOM and a WSM6 proxy.