{"id":23683,"date":"2019-12-16T08:00:05","date_gmt":"2019-12-16T16:00:05","guid":{"rendered":"https:\/\/insidebigdata.com\/?p=23683"},"modified":"2020-01-03T08:59:21","modified_gmt":"2020-01-03T16:59:21","slug":"heterogeneous-computing-programming-oneapi-and-data-parallel-c","status":"publish","type":"post","link":"https:\/\/insidebigdata.com\/2019\/12\/16\/heterogeneous-computing-programming-oneapi-and-data-parallel-c\/","title":{"rendered":"Heterogeneous Computing Programming: oneAPI and Data Parallel C++"},"content":{"rendered":"\n

Sponsored Post<\/em><\/p>\n\n\n\n

What you missed at the Intel Developer Conference, and how to catch-up today<\/em><\/p>\n\n\n\n

By James Reinders<\/p>\n\n\n\n

In the interests of full disclosure \u2026 I must admit that I became sold on DPC++ after Intel approached me (as a consultant \u2013 3 years retired from Intel) asking if I\u2019d help with a book on DPC++. I was skeptical at first, but the more I learned, the more I wanted to help. Many months later, with the opening day of Intel\u2019s HPC Developers conference (ahead of SuperComputing 2019 in Denver), I sat in the audience with everyone else anxious to see Intel make oneAPI and DPC++ public. I was also invited to meet some customers, and Intel engineers, afterwards to provide feedback.<\/p>\n\n\n\n

I\u2019d like to share why I see so much potential in what Intel is announcing \u2014 and most importantly: <\/em>how you can try it out yourself and decide for yourself. By the way, the book I mentioned is a work in progress and you can download the first four chapters of Data Parallel C++.  I\u2019d love to hear your feedback!<\/p>\n\n\n\n

\"\"<\/a><\/figure>\n\n\n\n

Imagine a world where\nmultiarchitecture programming is a given \u2013 and software developer tools are\nalways multiarchitecture: compilers, debuggers, libraries, and optimization\ntools. That can become a reality with Intel\u2019s unveiling of the oneAPI open industry\ninitiative, including a cloud platform to test them on.<\/p>\n\n\n\n

Intel offers strong evidence of its commitment\nto this era of heterogeneous computing, by aligning its hardware and software\nroadmaps to strongly support this world order\u2014where multiarchitecture\nprogramming is the norm. In addition to its continuing delivery of new\ncapabilities on Intel\u00ae Xeon\u00ae Scalable processors and powerful Intel Stratix\u00ae\nFPGAs, Intel disclosed more about the new Intel Xe<\/sup>\narchitecture for HPC.<\/p>\n\n\n\n

This is definitely good news. Read on\nto learn how you<\/em> can check it out\nyourself.<\/p>\n\n\n\n

Background:\noneAPI and DPC++<\/strong><\/p>\n\n\n\n

The oneAPI\nproject creates a specification for supporting programmers of heterogeneous\nsystems, in a way that is open and aims to make the most out of every compute\ndevice within a system \u2014 no unequal treatment (Raja said \u201cno transistor left\nbehind\u201d and \u201cno programmer left behind\u201d). \nThe project outlines support for both direct<\/em> programming and API<\/em>\nprogramming. Intel promises to deliver product (in beta now) that delivers a\nunified language and libraries with the ability to yield full native code\nperformance across a range of hardware, including CPUs, GPUs, FPGAs and AI\naccelerators. <\/p>\n\n\n\n

The oneAPI project is a game changer for software developers \u2014 in that it declares that the norm for programming should be a unified programming model that simplifies development for workloads across diverse architectures. Gone should be the days of proprietary solutions specific to one vendor\u2019s hardware. Gone should be the days when a library, or a tool, can limit its visibility and usefulness to a single type of compute device.<\/p>\n\n\n\n

What may be truly different about Intel in its approach to heterogeneous computing is that it seeks to be both open and performant \u2014 which would set it apart from other approaches available today.<\/p>\n\n\n\n

Imagine a world where libraries and\nlanguages are expected to support a range of hardware \u2014 a world where the\nprogrammer can program a range of hardware using a unified language and a\nsingle code base — but still have access to full native code performance!<\/p>\n\n\n\n

The vision does not stop at Intel\nhardware either \u2014 it is open, including both the oneAPI specification, and the Khronos Group\u2019s SYCL specifications\nupon which DPC++ is built. The openness means that anyone can build their own\nimplementations of oneAPI to support\ntheir solutions and products. The specifications for oneAPI are labelled\npre-1.0 (and they look that way \u2013 so feedback looks likely to be useful for the\nspec), with Intel encouraging community feedback. <\/p>\n\n\n\n

Data Parallel C++ Builds on C++ the Right Way<\/strong><\/p>\n\n\n\n

At the heart of the specifications, tools, toolkits, and libraries is the Data Parallel C++ compiler. Data Parallel C++ will feel very familiar to any C++ programmer, and has all the productivity and performance you would expect with C++. Data Parallel C++ is full C++, extended for data parallelism by the Khronos SYCL specifications, with some Intel extensions that aim to provide functionality not yet in C++ or SYCL. <\/p>\n\n\n\n

Data Parallel C++ is a kernel-based\napproach to programming, so it will feel familiar to those who know CUDA,\nOpenCL, or SYCL.  DPC++ is a compliant\nSYCL compiler, with extensions. Like SYCL and OpenCL, DPC++ is not a\nproprietary solution.  <\/p>\n\n\n\n

Intel\u2019s implementation of oneAPI is designed to ensure a seamless journey\nfor developers as they migrate current applications or build new ones to take\nadvantage of different architectures. It is compatible and interoperable with\nother standards including Fortran, OpenMP and MPI, so developers don\u2019t lose\ntheir existing coding investments.<\/p>\n\n\n\n

Build DPC++ Completely from Open Source<\/strong><\/p>\n\n\n\n

Get the DPC++ compiler prebuilt with installers or build\nthe DPC++ compiler completely from source, it is your choice.<\/p>\n\n\n\n

You\u2019ll want the oneAPI tools and toolkits from Intel which is also the easy way to get the DPC++ compiler as well. It is also very easy to build from the open source repository (I\u2019ve done it successfully on several Linux and Windows systems by following the instructions at GetStartedWithSYCLCompiler.md<\/a>). All the examples in the training videos, and in the DPC++ book, will compile and work with either the Intel product DPC++ compiler, or the open source compiler built from the GitHub repo \u2014 on Linux and Windows.<\/p>\n\n\n\n

How to Try it Out: oneAPI Beta is Free, and Easy to Try<\/strong><\/p>\n\n\n\n

On Intel\u2019s oneAPI product website<\/a> for software developers you will find more information on the toolkits and their component tools. The oneAPI site for the specifications, and the oneAPI community is http:\/\/oneapi.com<\/a>.<\/p>\n\n\n\n

This is a 1-2-3 to learn more (all of\nthis is free \u2013 no hidden costs):<\/p>\n\n\n\n