Leadtek Research Inc., known globally for extreme visual graphics technology development, is pleased to announce the cooperation with CAPS to be its reseller in China, ensuring Leadtek customers to enjoy the best GPU computing experience possible. Leadtek extends its offering with the distribution of HMPP and related services. CAPS offers services and consult supported by specialists.
CAPS provides end users of such systems with a breakthrough technology enabling to port legacy applications onto hybrid systems in a record time. An efficient and portable hybrid application is automatically produced from the source code of the application, annotated with directives or programs.
HMPP™ enables application developers to reduce the development time required for accelerating legacy code by taking advantage of the massively parallel CUDA architecture of NVIDIA’s GPUs.
Living up to its reputation for reliability, quality and performance based on unrelenting R&D, Leadtek mission is to integrate high-end computing solution and bring the best platform to independent software suppliers. Based on the present mature channel system Leadtek would continue to develop and plow industries customers deeply. Consequently the key challenge confusing enterprises in HPC fields will get completed, which is how to increase the developing efficiency and performance. From a macro perspective, it is bound to raise the technology level to such an extent on an industry scale in China.
The value resulting from this powerful cooperation will be out of expectation, bringing an innovative and leading edge technology to end users, CAPSexpands into Asia Pacific with the opening of new offices in Shanghai. This new local presence in China will enable CAPS to provide clients with local market expertise and to strengthen ties with its reseller network throughout the Asia market.
"This partnership with a global player like Leadtek is one important step in our internationalization strategy”, declares Benoît Raoult, CAPS APAC VP and General Manager “It contributes to CAPS willing to ease and secure manycore computing adoption among the industry. This worldwide strategy will be adapted locally by Leadtek and its network among China ensuring great customer experiences."
“This agreement gives us the opportunity to bring into play both of advantages in hardware and applications”, declares Bor Tsang, Leadtek (Shanghai) COO& General Manager, “CAPS Software solutions fully complement the tools and services Leadtek provides. Not only will the enterprises, but also our customers benefit from this new technology and solution”.
PathScale Inc., an industry leader in delivering high performance AMD64 and Intel64 compilers, today announced its new PathScale ENZO Compiler Suite will support the NVIDIA GPUs using the HMPP directive-based programming model originally developed by CAPS. Today also marks the start to CAPS and PathScale jointly working on advancing the HMPP directives as a new open standard. The new ENZO Compiler Suite is available for testing by selected customers and will be generally available later this summer.
PathScale has been delivering its EKO compilers for Fortran, C, and C++ tailored to meet the needs of customers demanding the highest level of performance from their applications. “Our new ENZO compilers not only bring performance to another level by exploiting the NVIDIA GPU's, but also allows incremental changes without sacrificing maximum performance” according to Christopher Bergström, Chief Technical Officer at PathScale. “By jointly working together with CAPS on making HMPP directives an open standard we make possible a new evolution in the gpgpu programming model that has been long overdue.”
The HMPP directives have been designed and used for more than 2 years by major HPC leaders. As a high level of abstraction for programming GPUs in scientific applications, HMPP preserves customers’ software assets by keeping applications portable. The HMPP directive-based programming model offers a powerful syntax to efficiently offload computations on hardware accelerators and to optimize data movement.
“PathScale’s choice to implement the HMPP directive programming model in their ENZO compiler suite supports CAPS commitment in delivering to users an open standard solution that will secure their investment”, declares François Bodin, Chief Technical Officer at CAPS.
CAPS is pleased to announce the availability of an OpenCL code generator within the just released 2.3 version of its HMPP directive-based hybrid compiler. Also, the CUDA back-end generator has been enhanced with Fermi capabilities and this new release brings support for more native compilers with Intel ifort/icc, GNU gcc/gfortran and PGI pgcc/pgfort compilers, enabling developers to freely use their favorite compiler with HMPP 2.3.
Based on GPU programming and tuning directives, HMPP offers an incremental programming model that allows developers with different levels of expertise to fully exploit GPU hardware accelerators in their legacy code.
As an emerging open programming standard, OpenCL back-end expands the portfolio of targets supported by HMPP to the AMD ATI GPUs. The OpenCL version of HMPP fully supports AMD and NVIDIA GPU compute processors, bringing to users a wider set of hybrid platforms they can execute their applications on. Recently released, the NVIDIA Tesla 200-series GPUs based on the ‘Fermi’ codename new CUDA architecture is also supported by HMPP 2.3.
“The addition of this OpenCL back-end to our existing NVIDIA CUDA back-end is a major milestone in HMPP development that gives users another powerful standard programming option,” comments Stéphane Bihan, Sales manager at CAPS, “We really look forward to demonstrating it at ISC’10 in Hamburg this week.”
CS, designer, integrator and operator of mission critical systems, and CAPS, a leading global provider of compiler technologies and engineering services for parallel hybrid computing, announce the signing of a reseller agreement.
According to this agreement, CS mission is to promote and support HMPP™ (Heterogeneous Multicore Parallel Programming), a development tools suite allowing software applications to be ported, maintained and deployed on various hybrid systems configurations. This solution makes the most of the computing power of clusters based hybrid processor GPU (Graphics Processing Unit).
CAPS will benefit from CS expertise in high performance simulation, which includes:
System and application support, with access to powerful computing resources and migration, optimization and parallelization of computer coding,
Consulting & training: consulting services, audits, engineering and design consultancy, benchmarking, technology assessments, training in digital design.
“CS willingness to expand its solutions portfolio and expertise in high computing and simulation makes it a partner of choice to develop the French market. In addition, we complement each other in terms of solutions proposed, markets and customers. This partnership looks very promising.” comments Denis Gerrer, CAPS Sales Manager.
“With the rise of simulation needs and hybrid systems deployment, this partnership complements our offerings in computing and high performance simulation in order to speed up the phases of design and products development.” concludes Philippe Barabinot, CS Director of Business Development and Partnerships at Systems Simulation Activity.
CAPS entreprise expands into Asia Pacific with the opening of new offices in Shanghai, China.
Bolstered by its success in Europe with major companies in energy, oil & gas, defense and research, CAPS has started its internationalization in 2009 by signing new partnerships with American and Asian High Performance Computing actors.
Founded in April 2002, CAPS has been developing for more than 7 years manycore programming tools. CAPS is now a leading global provider of compiler technologies and engineering services for parallel hybrid computing.
This new office opening in Shanghai reinforces its local presence in APAC first initiated in 2009 by new reselling agreements with supercomputing specialists in Japan and Taïwan. Several organizations, computing centers, universities and industrial companies are currently evaluating HMPP™, CAPS hybrid compiler.
“ This new Shanghai office opening reaffirms CAPS commitment to this rapidly expanding market. We are proud to deliver our APAC customer the same level of services than in Europe”, declares Benoît Raoult, CAPS APAC VP and General Manager, “Sharing our GPU computing experience is our goal when setting up in China”.
There is undeniably a growing interest in the APAC region for hybrid computing and CAPS is well positioned to provide developers with the skills and technologies they need to be successful. This new local presence in China will enable CAPS to provide clients with local market expertise and to strengthen ties with its reseller network throughout the Asia market.
CAPS entreprise is accelerating its international deployment with a new distribution agreement with JCC-Gimmick dedicated to the promotion and sales of its flagship product HMPP™ (Heterogeneous Multicore Parallel Programming) in Japan.
GPU Computing is one of the key breaking news of year 2009. The use of hybrid systems, e.g. computer systems mixing CPUs & GPUs to accelerate applications has spawned from research to industry over the past 2 years. This year numerous testimonials, ranging from research centers to commercial and government organizations, have shown the speed up that can be gained through the use of these systems. Most of the leading system providers have included hybrid systems, from desk side supercomputers to very large systems, in their product portfolio.
CAPS entreprise provides end users of such systems with a breakthrough technology enabling to port legacy applications onto hybrid systems in a record time. An efficient and portable hybrid application is automatically produced from the source code of the application, annotated with directives or pragmas. CAPS entreprise’s customers benefit from a pragmatic and elegant solution to the porting of their legacy applications.
JCC-Gimmick is a growing player in the area of supercomputing, providing expertise and training. The company extends its offering with the distribution of HMPP and related services.
According to this agreement, JCC-Gimmick mission is to ensure HMPP promotion, commercialization, deployment and support on Japanese territory.
«In the development of CAPS entreprise in APAC, Japan is an important step. This agreement with JCC-Gimmick as a partner to promote HMPP on the Japanese market illustrates CAPS’ international fast deployment strategy. I am glad to have JCC-Gimmick as a partner to promote our products », says Benoît Raoult, CAPS Manager for APAC sales and partnerships. « This partnership creates value for the customers of HMPP in Japan, bringing an innovative and leading edge technology to end users, supported by the expertise in parallel programming of a growing reputation specialist like JCC-Gimmick».
«This new agreement gives us the opportunity to add to our HPC market offerings a major player in manycore programming», declares Y. Watanabe san, JCC-Gimmick Executive General Manager, «CAPS Software solutions fully complement the tools and services JCC-Gimmick provides. They offer new technology opportunities for our customers and complement our expertise».
CAPS, leading global provider of compiler technologies and engineering services for parallel hybrid computing, has announced that Oak Ridge National Laboratory (ORNL) will use CAPS’ HMPP compiler to leverage the computing power of a graphics processing unit (GPU)-based hybrid cluster.
As a world leader in high-performance computing (HPC), ORNL is preparing the future of next-generation petascale computing with machines that will combine general-purpose central processing unit (CPU) cores with hundreds of GPU cores, which can deliver tremendous performance. In this context, ORNL have selected HMPP to enable the programming of high-performing parallel GPU/CPU hybrid applications.
“We are very glad ORNL chose HMPP for the future of their parallel hybrid application development. During the past two years, HMPP has been used by major European HPC leaders and has been able to achieve a level of maturity that makes it a tool of choice.” said Stéphane Bihan, director of CAPS Global Business Development, “Having ORNL select HMPP is a commitment from the industry for a portable, compiler-based solution in manycore computing”.
Based on a set of directives for programming and tuning GPU-accelerated applications, HMPP is a C and Fortran source-to-source compiler that gives developers a high level of abstraction for programming GPUs in scientific applications. HMPP offers developers an incremental way of programming from minimal expertise to advanced and expert. HMPP works with standard compilers and hardware vendor tools to create the application binary.
“We like the way HMPP addresses manycore programming: in addition to GPU programming directives that efficiently define and optimize computations offloaded in GPUs, the tuning directives give us control over the fine-tuning of the GPU-accelerated kernels,” said Richard Graham, group leader for the Applications Performance Tools group within ORNL’s Computer Science and Mathematics Division. “Also, by letting us use our standard compilers, HMPP really seamlessly integrates in our development environment.”
GENCI awards to CAPS entreprise a contract enabling the research & development community to benefit from its expertise and tools to port large scale scientific applications onto hybrid systems.
CAPS entreprise has been awarded a research & development contract by GENCI (www.genci.fr) to enable scientific users to seamlessly port large scale applications onto hybrid systems available from their workstations to the national supercomputers located in CCRT (Computing Center for Research and Technology) and CINES (France's National Computer Center for Higher Education).
The users of computer resources that GENCI made available to the French and European research and engineering communities have notably access to Europe’s largest hybrid supercomputer installed in CCRT, provided by BULL and including NVIDIA Tesla GPUs. From research to industry, the numerical simulations performed with CCRT and CINES support are addressing a number of scientific and societal domains: design of aerospace vehicles and propulsion engines, safety of nuclear power plants, medical image processing, earthquake prediction, nanomaterials, climate modeling and astrophysics …
CAPS entreprise provides to end users of such systems a breakthrough technology enabling to port and optimize applications in a record time. An efficient and portable hybrid application is automatically produced from the source code of the application, annotated with directives or pragmas. CAPS entreprise’s customers therefore benefit from a pragmatic and elegant solution to the porting of their legacy applications.
« Thanks to this contract, CAPS entreprise will provide to the engineering and scientific community with its HMPP™ software product, training and expertise to port and run large scale applications onto our hybrid systems. We are pleased to work with CAPS to speed up the Time-to-Production for new applications, broadening the spectrum of scientific fields that benefits from the power of hybrid systems”, mentioned Catherine Rivière, CEO of GENCI.
A dedicated Call for Proposals has been recently proposed by GENCI to the French scientific community so that a number of applications that would benefit from the power offered by hybrid systems will be ported with the help of CAPS experts and the use of its flagship product HMPP. CAPS will also train engineers and researchers on the programming of manycore systems with a specific emphasis on GPU programming.
« We are proud of this award by GENCI, a leading European organization promoting the use and large deployment of HPC resources. This R&D contract and our participation with CEA/GENCI on a PRACE (Partnership for Advanced Computing in Europe) prototype is the recognition that CAPS is the leading European independent supplier of tools and expertise in the manycore area and that HMPP is a unique solution whose value proposal to end users, OEMs & Integrators is well accepted.
CAPS will enlarge through this contract the set of large scale applications ported on hybrid systems using HMPP and develop new key product features that will be announced later this year, enforcing HMPP’s technology leadership », said Jean-Marc Talbot, CAPS President & CEO.
CAPS entreprise the industry leader in development tools for high performance applications is pleased to announce that it is working closely with the NVIDIA teams to make HMPP compiler fully support the new NVIDIA Fermi architecture.
“With several key features such as a true cache hierarchy, concurrent thread execution and ECC, Fermi brings real breakthrough in GPU computing that we are very excited to exploit within our HMPP directive-based CUDA compiler.” declares Dr François Bodin, CAPS Chief Technical Officer.
By generating CUDA code from C and Fortran, HMPP offers a high level abstraction of hybrid programming that fully leverages the computing power of NVIDA Tesla processors. HMPP enables developers to incrementally write hybrid applications, simply from accelerating regions of code, letting HMPP manage all the optimizations such as CPU-GPU data movements, to advanced tuning CUDA kernels. One of the key benefits of HMPP is to ensure portability and performance of CUDA applications across new versions of NVIDIA processors.
“HMPP enables application developers to reduce the development time required for accelerating legacy code by taking advantage of the massively parallel CUDA architecture of NVIDIA’s GPUs”, explains Andy Keane, general manager, Tesla GPU computing solutions at NVIDIA. “NVIDIA is collaborating with CAPS to enable them to build a powerful data parallel code generator within HMPP for CUDA GPUs based on the new Fermi architecture.”
CAPS a leader in development tools for parallel hybrid computing is pleased to announce the release of a new HMPP CAL/IL code generator to accelerate applications using AMD graphics processors.
With this new back-end, CAPS broadens its portfolio of supported hardware accelerators allowing customers to address different hybrid systems with one single directive-based programming model.
While HMPP was already partially supporting AMD graphics products offering a portable interface to integrate hand written CAL/IL kernels in Fortran or C legacy applications, the new CAL/IL code generator makes HMPP fully automatic in generating AMD accelerated applications, reducing substantially development time.
“AMD is proud to work with technology partners, like CAPS, to deliver a robust development ecosystem for ATI Stream technology, designed to enable developers to take full advantage of the immense processing capabilities available in today’s heterogeneous architectures,” says Patricia Harrell, director of stream computing at AMD. “With the release of its CAL/IL back-end for the HMPP development platform, CAPS continues to strengthen the ecosystem for ATI Stream technology, by enabling developers to leverage both the CPU and the GPU in their system using familiar programming interfaces such as C and Fortran."
“We are pleased to add the support of AMD/ATI GPU within HMPP. This will allow customers to benefit from the performance of this processor. Together with the AMD stream computing software tools, HMPP brings a real added value by allowing developers to quickly leverage the tremendous power of AMD GPU processors for their applications, while preserving their software assets”, declares Jean-Marc TALBOT, CAPS CEO.
The official release date is planned in December this year and CAPS will be demonstrating HMPP CAL/IL at SuperComputing, in Portland, Oregon in from 16th to 19th of November, booth number 1100.
CAPS to announce CAL/IL and OpenCL back-end plus Windows support for its HMPP™ hybrid compiler
CAPS, the pioneer in directive-based compiler technology with its HMPP flagship workbench, is pleased to announce the release of two new AMD CAL/IL and OpenCL back-end code generators, plus support for Windows operating system, thus expanding the field for new HMPP adopters.
While CAL/IL leverages the performance of AMD Stream processors, OpenCL perfectly matches HMPP code generation. OpenCL brings a low level API that enables to fine tune the programming of parallel processors from various hardware vendors coupled to general purpose CPU processors. HMPP positions itself as a high level abstraction for manycore programming, thus removing developers need for a deep knowledge of the underlying hardware architecture.
In addition to Linux x86_64, HMPP is opening to a wider developers’ community with a Windows x86_64 release running with Visual Studio.
“By complementing our existing HMPP CUDA™ compiler with two new back-ends, HMPP now fully addresses GPUs, the today’s most adopted parallel compute co-processors. From the past two years, we have seen a real move of the industry toward these new technologies and CAPS is committed to provide developers with the software tools they need to enjoy the ride of manycore programming.” declares Laurent Bertaux, CAPS COO and HMPP Product Manager.
CAL/IL back-end will be available in December, OpenCL and Windows support are planned to be released Q1 2010. CAPS will be demonstrating HMPP CAL/IL at SuperComputing, Portland, Oregon, from 16th to 19th of November, booth number 1100.
CAPS to announce reseller agreement with ParaTools to distribute and support HMPP in North America CAPS entreprise is starting its international deployment with a first distribution agreement with ParaTools, Inc. dedicated to the promotion and sales of its flagship product HMPP™ (Heterogeneous Multicore Parallel Programming) in North America.
GPU Computing is one of the key breaking news of year 2009. The use of hybrid systems, e.g. computer systems mixing CPUs & GPUs to accelerate applications has spawned from research to industry over the past 2 years. This year numerous testimonials, ranging from research centers to commercial and government organizations, have shown the speed up that can be gained through the use of these systems. Most of the leading system providers have included hybrid systems, from desk side supercomputers to very large systems, in their product portfolio.
CAPS entreprise provides end users of such systems with a breakthrough technology enabling to port legacy applications onto hybrid systems in a record time. An efficient and portable hybrid application is automatically produced from the source code of the application, annotated with directives or pragmas. CAPS entreprise’s customers benefit from a pragmatic and elegant solution to the porting of their legacy applications.
ParaTools, Inc. is a growing player in the area of supercomputing, providing consulting and value added services in the field of performance analysis of parallel applications using tools like TAU and Vampir. The company extends its offering with the distribution of HMPP and related services.
According to this agreement, ParaTools mission is to ensure HMPP promotion, commercialization, deployment and support on North American territory, focusing on academic and government organizations.
« This agreement is the first step of the international deployment of CAPS and HMPP that will be followed by other announcements in the coming months. I am glad to have ParaTools, Inc. as a partner to promote HMPP on the North American market. », says Jean-Marc Talbot, CAPS President & CEO. « This partnership creates value for the customers of HMPP in the United States, bringing an innovative and leading edge technology to end users, supported by the expertise in parallel programming of a reputed specialist like ParaTools, Inc. Their skills in performance analysis of C,
C++, Fortran programs will complement the added value services that they will provide around HMPP».
« This new agreement gives us the opportunity to add to our parallel tool offerings a major player in manycore programming», declares Dr. Allen Malony, ParaTools CEO, « CAPS Software solutions fully complement the performance analysis tools and services ParaTools provides. They offer new technology opportunities for our customers and complement our expertise».
CAPS is proud to announce the launch of HMPP New Generation (HMPP 2.x). HMPP 2.x is a follow-on to HMPP First Generation, rolled out last fall. It offers a host of new features and enhancements that make HMPP more robust, easy-to-use, flexible and performance-boosting than ever.
Last October, after more than 5 years of advanced research and building on customer experience feedback, CAPS released HMPP First Generation, a complete development tool suite for high performance applications.
Based on C and FORTRAN directives, HMPP Workbench provides developers with powerful data-parallel code generators that efficiently leverage the computing power of stream processors while drastically reducing development time. HMPP Workbench is based on OpenMP™-like directive extensions, used to build hardware accelerated variants of functions to be offloaded in hardware accelerators such as NVIDIA® Tesla™ (or any CUDA™ compatible hardware) and AMD FireStream™.
The added-value for HMPP users is in massive development time savings, while still ensuring application portability and interoperability. HMPP™ enhances the value of software assets by keeping them independent from both hardware platforms and commercial software.
HMPP 2.x allows users to pipeline computations in multi-GPU systems and makes better use of asynchronous hardware features to build even better performing GPU accelerated applications.
HMPP 2.x fully supports AMD FireStream hardware with a CAL/IL code generator. The addition of an OpenCL code generator is another major milestone planned for the second half of this year that will give HMPP developers another powerful standard programming option.
“Our beta test customers are thrilled with the new HMPP features. Our new system makes software development so much easier, with new directive syntax error correction. Moving one step further in code optimization, our R&D teams have designed new HMPP directives that have fully proved their performance potential.” explains Laurent Bertaux, CAPS Managing Director. “This new HMPP generation will enable CAPS to address more hardware platforms and new operating systems.” Stay tuned to HMPP hot spots this fall!
CAPS Entreprise, the industry leader in development tools for high performance applications with its HMPP™ (Hybrid Multicore Parallel Programming) workbench, announces the launch of its CAPS Compute Lab, a first and exclusive EMEA solution center for hybrid computing with both BULL and NVIDIA partners.
As the industry is shifting toward hybrid computation technologies, the need for a solution enabling to rapidly and successfully evaluate the performance of software applications has become crucial.
By offering a remote access to a complete hardware/software platform along with CAPS high level expertise in manycore programming, CAPS Compute Lab provides a concrete and efficient answer to this need.
“All the HPC community is aware of the extraordinary performance potential that GPU-accelerated hardware systems combining NVIDIA Tesla with x86 can bring. There are already a few software applications in production but the adoption pace is quickly growing. Many industrial accounts have started to look at hybrid systems as a mean to speed up their strategic applications.”, explains Jean-Marc Talbot, President & CEO of CAPS entreprise, “However, evaluating the impact of these technologies on the application side requires some effort in terms of acquiring a hardware prototype, installing and getting used to the software environment, porting applications, etc. CAPS Compute Lab initiative aims at accelerating the evaluation step by bringing a ready-to-use cluster along with all the software programming environment and CAPS expertise”.
The CAPS Compute Lab Linux cluster combines 20 BULL NovaScale® R422 E2 compute nodes with 10 NVIDIA® Tesla™ S1070 servers for a peak performance of 42 Teraflops. This system comes with all the software components: CAPS HMPP hybrid programming workbench and NVIDIA CUDA™ environment.
" Bull is very proud to supply the cluster for the Compute Lab launched by our long-time partner CAPS. Combining Bull R422 E2 servers and NVIDIA Tesla S1070 systems is a very popular solution already adopted by some of our largest customers. CAPS' initiative will help hybrid computing gain momentum, by offering developers an easy and fast way to evaluate and port their applications on a fully operational hybrid system." declares Fabio Gallo, Vice President, High Performance Computing at Bull.
CAPS enables its customers to experiment their industrial applications on their hybrid platforms before production deployment, obtaining significant performance gain. Today, CAPS Compute Lab makes this evaluation affordable by removing this key barrier to technology adoption represented by the acquisition cost and time of hybrid platforms.
“Our goal is to bring the transformative power of GPU Computing to every commercial, industrial and research organization which can benefit from its massively parallel processing capabilities,” adds Walter Mundt-Blum, Vice President of Sales for EMEA & India - Professional Solutions Group. “The launch of their CAPS Compute Lab represents an important step towards that goal by making evaluation much more straightforward and thereby removing a key barrier to the technology’s adoption.”
Strengthening relationships with both BULL and NVIDIA provided CAPS with a unique opportunity to build this GPU computing experimentation offer equipped with state-of-the-art technology that fully matches developers’ requests.
All customers who purchase an evaluation package on CAPS Compute Lab will be eligible for a 2 months HMPP evaluation license, 400 computation hours and days for support, consulting and/or training.
For more information about CAPS Compute lab, please contact compute@caps-entreprise.com.
CAPS to release hardware-specific code generators within its HMPP™ Workbench
CAPS is pleased to announce its HMPP™ Workbench now integrates hardware-specific code generators that make the workbench a complete hybrid compiler solution.
“After more than 5 years of advanced research and getting the most of its customers experience feedbacks, CAPS has managed to release a complete hybrid compiler toolchain with powerful dataparallel code generators that fully leverage the computing power of today’s stream processors while drastically reducing development time.”, says Laurent Bertaux, CAPS Managing Director.
HMPP™ is based on OpenMP™-like directive extensions to build hardware accelerated variants of functions to be offloaded in hardware accelerators such as NVIDIA® Tesla™ (or any CUDA™ compatible hardware), AMD FireStream™ and SSE compute units.
By insulating the accelerated versions of functions from the legacy code, the HMPP™ standard and portable programming interface makes the link between general-purpose and hardware-specific programming. As an open platform, HMPP™ still allows developers to fine tune their hardware-specific kernels.
While preserving the application source from hardware-specific coding, HMPP™ applications dynamically distribute their computations on the different cores, thus adapting their execution to the platform configuration.
“The added-value for the HMPP™ users is the huge development efforts they save while ensuring their application portability and interoperability. HMPP™ allows them to protect their software investment by keeping them independent from the hardware platforms”, says Stéphane Bihan, CAPS Global Business Development Director, “As a complete and automatic solution for programming and deploying parallel hybrid applications, we are confident that HMPP™ will be largely adopted by the community of HPC developers.”
HMPP™ will be officially demonstrated at SC08, November, 2008 in Austin, Texas.
CAPS entreprise announced today that its HMPP™ Multicore Development Workbench now supports the AMD FireStreamTM 9170 stream processors and ATI RadeonTM HD 3870 Graphics using AMD Stream SDK’s Brook+.
“With the HMPP™ Workbench using AMD Stream SDK’s Brook+, CAPS provides developers with unique standard programming tools that keep their applications portable and interoperable with their evolutive hybrid platforms built with the AMD FireStreamTM 9170 stream processor and ATI RadeonTM HD 3870 graphics,” said Laurent Bertaux, President of CAPS entreprise.
CAPS HMPP™ Workbench is based on a set of compiler directives and comes with development tools and a runtime to simplify the use of hardware acceleration for conventional general purpose applications. The aim of HMPP™ is to integrate the use of hardware accelerators rather than porting applications to make use of them.
While the application control flow runs on the main CPU, the computational intensive parts are integrated as additional software plugins and are dynamically managed by HMPP™. They are written using AMD Stream SDK’s Brook+.
“We are proud to add CAPS’s HMPP Workbench to our growing AMD Stream Computing software ecosystem,” said Patricia Harrell, Director of AMD Stream Computing. “By adding support for AMD Stream SDK in its HMPP Workbench, CAPS helps enable its users to accelerate their applications easily with AMD Stream processors while retaining the implementation portability developers have come to demand.”
