OpenCL Applications Research Articles

Evaluation of ‘OpenCL for FPGA’ for Data Acquisition and Acceleration in High Energy Physics

The increase in the data acquisition and processing needs of High Energy Physics experiments has made it more essential to use FPGAs to meet those needs. However harnessing the capabilities of the FPGAs has been hard for anyone but expert FPGA developers. The arrival of OpenCL with the two major FPGA vendors supporting it, offers an easy software-based approach to taking advantage of FPGAs in applications such as High Energy Physics. OpenCL is a language for using heterogeneous architectures in order to accelerate applications. However, FPGAs are capable of far more than acceleration, hence it is interesting to explore if OpenCL can be used to take advantage of FPGAs for more generic applications. To answer these questions, especially in the context of High Energy Physics, two applications, a DAQ module and an acceleration workload, were tested for implementation with OpenCL on FPGAs2. The challenges on using OpenCL for a DAQ application and their solutions, together with the performance of the OpenCL based acceleration are discussed. Many of the design elements needed to realize a DAQ system in OpenCL already exists, mostly as FPGA vendor extensions, but a small number of elements were found to be missing. For acceleration of OpenCL applications, using FPGAs has become as easy as using GPUs. OpenCL has the potential for a massive gain in productivity and ease of use enabling non FPGA experts to design, debug and maintain the code. Also, FPGA power consumption is much lower than other implementations. This paper describes one of the first attempts to explore the use of OpenCL for applications outside the acceleration workloads.

LTTng CLUST: A System-Wide Unified CPU and GPU Tracing Tool for OpenCL Applications

As computation schemes evolve and many new tools become available to programmers to enhance the performance of their applications, many programmers started to look towards highly parallel platforms such as Graphical Processing Unit (GPU). Offloading computations that can take advantage of the architecture of the GPU is a technique that has proven fruitful in recent years. This technology enhances the speed and responsiveness of applications. Also, as a side effect, it reduces the power requirements for those applications and therefore extends portable devices battery life and helps computing clusters to run more power efficiently. Many performance analysis tools such as LTTng, strace and SystemTap already allow Central Processing Unit (CPU) tracing and help programmers to use CPU resources more efficiently. On the GPU side, different tools such as Nvidia’s Nsight, AMD’s CodeXL, and third party TAU and VampirTrace allow tracing Application Programming Interface (API) calls and OpenCL kernel execution. These tools are useful but are completely separate, and none of them allow a unified CPU-GPU tracing experience. We propose an extension to the existing scalable and highly efficient LTTng tracing platform to allow unified tracing of GPU along with CPU’s full tracing capabilities.

Developing adaptive multi-device applications with the Heterogeneous Programming Library

The usage of heterogeneous devices presents two main problems. One is their complex programming, a problem that grows when multiple devices are used. The second issue is that even if the codes for these devices can be portable on top of OpenCL, they lack performance portability, effectively requiring specialized implementations for each device to get good performance. In this paper we extend the Heterogeneous Programming Library (HPL), which improves the usability of heterogeneous systems on top of OpenCL, to better handle both issues. First, we provide HPL with mechanisms to support the implementation of any multi-device application that requires arbitrary patterns of communication between several devices and a host memory. In a second stage HPL is improved with an adaptive scheme to optimize communications between devices depending on the execution environment. An evaluation using benchmarks with very different nature shows that HPL reduces the SLOCs and programming effort of OpenCL applications by 27 and 43 %, respectively, while improving the performance of applications that exchange data between devices by 28 % on average.

The Impact of the SIMD Width on Control-Flow and Memory Divergence

Power consumption is a prevalent issue in current and future computing systems. SIMD processors amortize the power consumption of managing the instruction stream by executing the same instruction in parallel on multiple data. Therefore, in the past years, the SIMD width has steadily increased, and it is not unlikely that it will continue to do so. In this article, we experimentally study the influence of the SIMD width to the execution of data-parallel programs. We investigate how an increasing SIMD width (up to 1024) influences control-flow divergence and memory-access divergence, and how well techniques to mitigate them will work on larger SIMD widths. We perform our study on 76 OpenCL applications and show that a group of programs scales well up to SIMD width 1024, whereas another group of programs increasingly suffers from control-flow divergence. For those programs, thread regrouping techniques may become increasingly important for larger SIMD widths. We show what average speedups can be expected when increasing the SIMD width. For example, when switching from scalar execution to SIMD width 64, one can expect a speedup of 60.11, which increases to 62.46 when using thread regrouping. We also analyze the frequency of regular (uniform, consecutive) memory access patterns and observe a monotonic decrease of regular memory accesses from 82.6 at SIMD width 4 to 43.1% at SIMD width 1024.

OpenCL Performance Evaluation on Modern Multicore CPUs

Utilizing heterogeneous platforms for computation has become a general trend, making the portability issue important. OpenCL (Open Computing Language) serves this purpose by enabling portable execution on heterogeneous architectures. However, unpredictable performance variation on different platforms has become a burden for programmers who write OpenCL applications. This is especially true for conventional multicore CPUs, since the performance of general OpenCL applications on CPUs lags behind the performance of their counterparts written in the conventional parallel programming model for CPUs. In this paper, we evaluate the performance of OpenCL applications on out-of-order multicore CPUs from the architectural perspective. We evaluate OpenCL applications on various aspects, including API overhead, scheduling overhead, instruction-level parallelism, address space, data location, data locality, and vectorization, comparing OpenCL to conventional parallel programming models for CPUs. Our evaluation indicates unique performance characteristics of OpenCL applications and also provides insight into the optimization metrics for better performance on CPUs.

Design of OpenCL framework for embedded multi-core processors

In modern mobile embedded systems, various energy-efficient hardware acceleration units are employed in addition to a multi-core CPU. To fully utilize the computational power in such heterogeneous systems, Open Computing Language (OpenCL) has been proposed. A key benefit of OpenCL is that it works on various computing platforms. However, most vendors offer OpenCL software development kits (SDKs) that support their own computing platforms. The study of the OpenCL framework for embedded multi-core CPUs is in a rudimentary stage. In this paper, an OpenCL framework for embedded multi-core CPUs that dynamically redistributes the time-varying workload to CPU cores in real time is proposed. A compilation environment for both host programs and OpenCL kernel programs was developed and OpenCL libraries were implemented. A performance evaluation was carried out with respect to various definitions of the device architecture and the execution model. When running on embedded multi-core CPUs, applications parallelized by OpenCL C showed much better performance than the applications written in C without parallelization. Furthermore, since programmers are capable of managing hardware resources and threads using OpenCL application programming interfaces (APIs) automatically, highly efficient computing both in terms of the performance and energy consumption on a heterogeneous computing platform can be easily achieved.

A GPGPU based program to solve the TDSE in intense laser fields through the finite difference approach

We present a General-purpose computing on graphics processing units (GPGPU) based computational program and framework for the electronic dynamics of atomic systems under intense laser fields. We present our results using the case of hydrogen, however the code is trivially extensible to tackle problems within the single-active electron (SAE) approximation. Building on our previous work, we introduce the first available GPGPU based implementation of the Taylor, Runge–Kutta and Lanczos based methods created with strong field ab-initio simulations specifically in mind; CLTDSE. The code makes use of finite difference methods and the OpenCL framework for GPU acceleration. The specific example system used is the classic test system; Hydrogen. After introducing the standard theory, and specific quantities which are calculated, the code, including installation and usage, is discussed in-depth. This is followed by some examples and a short benchmark between an 8 hardware thread (i.e. logical core) Intel Xeon CPU and an AMD 6970 GPU, where the parallel algorithm runs 10 times faster on the GPU than the CPU. Program summaryProgram title: CLTDSECatalogue identifier: AESM_v1_0Program summary URL:http://cpc.cs.qub.ac.uk/summaries/AESM_v1_0.htmlProgram obtainable from: CPC Program Library, Queen’s University, Belfast, N. IrelandLicensing provisions: GNU General Public License, version 3 and aboveNo. of lines in distributed program, including test data, etc.: 15988No. of bytes in distributed program, including test data, etc.: 233518Distribution format: tar.gzProgramming language: C99 and OpenCL C. C99 conformance is ensured through use of C99 and pedantic flags under GCC and Clang.Computer: Single compute node.Operating system: GNU/Linux. It should, in principle, work with little modification for other Unix-Like systems.Has the code been vectorised or parallelised?: OpenCL is a parallel language. Thus CLTDSE can use all cores on a processor or GPU. OpenCL supports using all available compute units (GPUs and CPUs etc.), although this is not currently implemented in CLTDSE.RAM: Negligible RAM and GPU global memory (20MiB)Classification: 2.5.External routines: An OpenCL library; the current major packages are APP by AMD (CPU and AMD GPU), the NVIDIA driver (GPU), and the Intel SDK for OpenCL Applications (CPU and Intel HD Graphics). libconfig for processing configuration files.Nature of problem:Describing the dynamics of electrons under intense laser fields in atoms or molecules.Solution method:Solving the discretised system through finite difference using the Lanczos, Runge–Kutta, and Taylor methods.Restrictions:The example problem which is implemented is under the single active electron approximation.Unusual features:Focused on GPGPU acceleration through OpenCL.Running time:The running time depends on the nature of the pulse. This can vary from seconds to tens of minutes.

I/o paravirtualization at the device file boundary

Paravirtualization is an important I/O virtualization technology since it uniquely provides all of the following benefits: the ability to share the device between multiple VMs, support for legacy devices without virtualization hardware, and high performance. However, existing paravirtualization solutions have one main limitation: they only support one I/O device class, and would require significant engineering effort to support new device classes and features. In this paper, we present Paradice, a solution that vastly simplifies I/O paravirtualization by using a common paravirtualization boundary for various I/O device classes: Unix device files. Using this boundary, the paravirtual drivers simply act as a class-agnostic indirection layer between the application and the actual device driver. We address two fundamental challenges: supporting cross-VM driver memory operations without changes to applications or device drivers and providing fault and device data isolation between guest VMs despite device driver bugs. We implement Paradice for x86, the Xen hypervisor, and the Linux and FreeBSD OSes. Our implementation paravirtualizes various GPUs, input devices, cameras, an audio device, and an Ethernet card for the netmap framework with ~7700 LoC, of which only ~900 are device class-specific. Our measurements show that Paradice achieves performance close to native for different devices and applications including netmap, 3D HD games, and OpenCL applications.

I/o paravirtualization at the device file boundary

Paravirtualization is an important I/O virtualization technology since it uniquely provides all of the following benefits: the ability to share the device between multiple VMs, support for legacy devices without virtualization hardware, and high performance. However, existing paravirtualization solutions have one main limitation: they only support one I/O device class, and would require significant engineering effort to support new device classes and features. In this paper, we present Paradice, a solution that vastly simplifies I/O paravirtualization by using a common paravirtualization boundary for various I/O device classes: Unix device files. Using this boundary, the paravirtual drivers simply act as a class-agnostic indirection layer between the application and the actual device driver. We address two fundamental challenges: supporting cross-VM driver memory operations without changes to applications or device drivers and providing fault and device data isolation between guest VMs despite device driver bugs. We implement Paradice for x86, the Xen hypervisor, and the Linux and FreeBSD OSes. Our implementation paravirtualizes various GPUs, input devices, cameras, an audio device, and an Ethernet card for the netmap framework with ~7700 LoC, of which only ~900 are device class-specific. Our measurements show that Paradice achieves performance close to native for different devices and applications including netmap, 3D HD games, and OpenCL applications.

An OpenCL micro-benchmark suite for GPUs and CPUs

Open computing language (OpenCL) is a new industry standard for task-parallel and data-parallel heterogeneous computing on a variety of modern CPUs, GPUs, DSPs, and other microprocessor designs. OpenCL is vendor independent and hence not specialized for any particular compute device. To develop efficient OpenCL applications for the particular platform, we still need a more profound understanding of architecture features on the OpenCL model and computing devices. For this purpose, we design and implement an OpenCL micro-benchmark suite for GPUs and CPUs. In this paper, we introduce the implementations of our OpenCL micro benchmarks, and present the measuring results of hardware and software features like performance of mathematical operations, bus bandwidths, memory architectures, branch synchronizations and scalability, etc., on two multi-core CPUs, i.e. AMD Athlon II X2 250 and Intel Pentium Dual-Core E5400, and two different GPUs, i.e. NVIDIA GeForce GTX 460se and AMD Radeon HD 6850. We also compared the measuring results with existing benchmarks to demonstrate the reasonableness and correctness of our benchmark suite.

A Two-Level Task Scheduler on Multiple DSP System for OpenCL

This paper addresses the problem that multiple DSP system does not support OpenCL programming. With the compiler, runtime, and the kernel scheduler proposed, an OpenCL application becomes portable not only between multiple CPU and GPU, but also between embedded multiple DSP systems. Firstly, the LLVM compiler was imported for source-to-source translation in which the translated source was supported by CCS. Secondly, two-level schedulers were proposed to support efficient OpenCL kernel execution. The DSP/BIOS is used to schedule system level tasks such as interrupts and drivers; however, the synchronization mechanism resulted in heavy overhead during task switching. So we designed an efficient second level scheduler especially for OpenCL kernel work-item scheduling. The context switch process utilizes the 8 functional units and cross path links which was superior to DSP/BIOS in the aspect of task switching. Finally, dynamic loading and software managed CACHE were redesigned for OpenCL running on multiple DSP system. We evaluated the performance using some common OpenCL kernels from NVIDIA, AMD, NAS, and Parboil benchmarks. Experimental results show that the DSP OpenCL can efficiently exploit the computing resource of multiple cores.

Methods for Optimizing OpenCL Applications on Heterogeneous Multicore Architectures

Heterogeneous multicore architectures with CPU and add-on GPUs or streaming processors are now widely used in computer systems. These GPUs provide substantially more computation capability and memory bandwidth compared to traditional multi-cores. Also, because they are highly programmable, they provide the computational performance needed for realistic graphics rendering. Applications with general computations can also be leveraged onto these GPUs. This study discusses the architectures of these highly efficient GPUs and applies a unified programming standard c alled OpenCL to fully utilize their capabilities. Despite their great potential, applications of these GPUs are challenging because of the ir diverse underlying architectural characteristics. In this study, several optimizing techniques are applied on OpenCL-compatible heterogeneous multicore architectures to achieve thread-level and data-level parallelisms. The architectural implications of these techniq ues are discussed. Finally, optimization principles for these architectures will be are proposed. The experimental reveal average speedups of 24 and 430 for non-optimized and optimized kernels, respectively.

DOpenCL: Towards uniform programming of distributed heterogeneous multi-/many-core systems

Modern computer systems become increasingly distributed and heterogeneous by comprising multi-core CPUs, GPUs, and other accelerators. Current programming approaches for such systems usually require the application developer to use a combination of several programming models (e.g., MPI with OpenCL or CUDA) in order to exploit the system’s full performance potential. In this paper, we present dOpenCL (distributed OpenCL)—a uniform approach to programming distributed heterogeneous systems with accelerators. dOpenCL allows the user to run unmodified existing OpenCL applications in a heterogeneous distributed environment. We describe the challenges of implementing the OpenCL programming model for distributed systems, as well as its extension for running multiple applications concurrently. Using several example applications, we compare the performance of dOpenCL with MPI + OpenCL and standard OpenCL implementations.

Invitation to a Standard Programming Interface for Massively Parallel Computing Environment: OpenCL

Multicore/manycore architecture accelerates demand for a new programming environment to utilize the massive processors integrated in an LSI. GPU (Graphics Processing Unit) is one of the typical hardware environments. The programming environments on GPU are traditionally vendor-/hardware-specific, where complicate the management of uniform programs that access computing resources of the massively parallel platform. The recently released OpenCL is expected to become a standard for providing a uniform programming environment for the heterogeneous processors from different vendors. This tutorial paper introduces the overview of the OpenCL that motivates the programmers who are going to program the massively parallel hardware or who migrates the programming method from another vendor specific programming interface to the OpenCL. This paper explains the haracteristics of the OpenCL interface with describing in detail the basic structures used in the program. Moreover, this paper discusses performance aspects to evaluate advanced programming techniques that improve the performance of the OpenCL applications.

Discrete dipole approximation simulations on GPUs using OpenCL—Application on cloud ice particles

The global distribution and climatology of ice clouds are among the main uncertainties in climate modeling and prediction. In order to retrieve ice cloud properties from remote sensing measurements, the scattering properties of all cloud ice particle types must be known. The discrete dipole approximation (DDA) simulates scattering of radiation by arbitrarily shaped particles and is thus suitable for cloud ice crystals. The DDA models the particle as a collection of equal dipoles on a lattice, and is computationally much more expensive than approximations restricted to more regularly shaped particles. On a single computer the calculation for an ice particle of a specific size, for a given scattering plane at one specific wavelength can take several days. We have ported the core routines of the scattering suite “ADDA” to the open computing language (OpenCL), a framework for programming parallel devices like PC graphics cards (graphics processing units, GPUs) or multi-core CPUs. In a typical case we can achieve a speed-up on a GPU as compared to a CPU by a factor of 5 in double precision and a factor of 15 in single precision. Spreading the work load over multiple GPUs will allow calculating the scattering properties even of large cloud ice particles.

Achieving a single compute device image in OpenCL for multiple GPUs

In this paper, we propose an OpenCL framework that combines multiple GPUs and treats them as a single compute device. Providing a single virtual compute device image to the user makes an OpenCL application written for a single GPU portable to the platform that has multiple GPU devices. It also makes the application exploit full computing power of the multiple GPU devices and the total amount of GPU memories available in the platform. Our OpenCL framework automatically distributes at run-time the OpenCL kernel written for a single GPU into multiple CUDA kernels that execute on the multiple GPU devices. It applies a run-time memory access range analysis to the kernel by performing a sampling run and identifies an optimal workload distribution for the kernel. To achieve a single compute device image, the runtime maintains virtual device memory that is allocated in the main memory. The OpenCL runtime treats the memory as if it were the memory of a single GPU device and keeps it consistent to the memories of the multiple GPU devices. Our OpenCL-C-to-C translator generates the sampling code from the OpenCL kernel code and OpenCL-C-to-CUDA-C translator generates the CUDA kernel code for the distributed OpenCL kernel. We show the effectiveness of our OpenCL framework by implementing the OpenCL runtime and two source-to-source translators. We evaluate its performance with a system that contains 8 GPUs using 11 OpenCL benchmark applications.

A programming package for the calculation of cross-sections and probabilities for charge-exchange processes

We present a General-purpose computing on graphics processing units (GPGPU) based computational program and framework for the electronic dynamics of atomic systems under intense laser fields. We present our results using the case of hydrogen, however the code is trivially extensible to tackle problems within the single-active electron (SAE) approximation. Building on our previous work, we introduce the first available GPGPU based implementation of the Taylor, Runge–Kutta and Lanczos based methods created with strong field ab-initio simulations specifically in mind; CLTDSE. The code makes use of finite difference methods and the OpenCL framework for GPU acceleration. The specific example system used is the classic test system; Hydrogen. After introducing the standard theory, and specific quantities which are calculated, the code, including installation and usage, is discussed in-depth. This is followed by some examples and a short benchmark between an 8 hardware thread (i.e. logical core) Intel Xeon CPU and an AMD 6970 GPU, where the parallel algorithm runs 10 times faster on the GPU than the CPU.Program title: CLTDSECatalogue identifier: AESM_v1_0Program summary URL:http://cpc.cs.qub.ac.uk/summaries/AESM_v1_0.htmlProgram obtainable from: CPC Program Library, Queen’s University, Belfast, N. IrelandLicensing provisions: GNU General Public License, version 3 and aboveNo. of lines in distributed program, including test data, etc.: 15 988No. of bytes in distributed program, including test data, etc.: 233 518Distribution format: tar.gzProgramming language: C99 and OpenCL C. C99 conformance is ensured through use of C99 and pedantic flags under GCC and Clang.Computer: Single compute node.Operating system: GNU/Linux. It should, in principle, work with little modification for other Unix-Like systems.Has the code been vectorised or parallelised?: OpenCL is a parallel language. Thus CLTDSE can use all cores on a processor or GPU. OpenCL supports using all available compute units (GPUs and CPUs etc.), although this is not currently implemented in CLTDSE.RAM: Negligible RAM and GPU global memory (20MiB)Classification: 2.5.External routines: An OpenCL library; the current major packages are APP by AMD (CPU and AMD GPU), the NVIDIA driver (GPU), and the Intel SDK for OpenCL Applications (CPU and Intel HD Graphics). libconfig for processing configuration files.Nature of problem:Describing the dynamics of electrons under intense laser fields in atoms or molecules.Solution method:Solving the discretised system through finite difference using the Lanczos, Runge–Kutta, and Taylor methods.Restrictions:The example problem which is implemented is under the single active electron approximation.Unusual features:Focused on GPGPU acceleration through OpenCL.Running time:The running time depends on the nature of the pulse. This can vary from seconds to tens of minutes.