Hardware

Intel Create 2019 Event Reveals ‘Master Plan’ Involving Open-Source Software With High Performance Kernels For Ray Tracing

Intel used its SIGGRAPH 2019 event, which is popularly known as ‘Intel Create’, to offer some amazing insights into its journey. The CPU, GPU, PC, server and other high-end component maker gave some glimpses into its future. Moreover, Intel surprised the audience by reaffirming its commitment to the software side of the hardware business. The software aspect within the hardware that companies manufacture often lacks finesse, sophistication, and excellence. However, using the Intel Create platform, the company demonstrated that it had stayed true to the promises made long back.

The Intel SIGGRAPH 2019 event was particularly interesting because of the brand new kernels for the high-performance ray tracing technology. However, even beyond the same, Intel made some startling revelations and renewed its commitment to maintaining the open-source nature of its software. Incidentally, Intel also promised delivery of an open-source ray tracing engine. Intel is calling the same ‘OSPRay 2.0 Ray Tracing Engine’. Included in the presentation was a clear and committed Intel roadmap for the next two quarters of the current year. Moreover, the company offered an entirely new law that it is calling ‘Raja’s Law’ that bases itself on the still valid Moore’s Law but becomes valid for the future of computing power beyond the currently prevalent Bell’s Law.

Intel Sets 1000x Performance Advancement Goal?

Intel’s master plan for its high-end computing processors is quite fascinating. The company may have recently sold off its mobile modem business to Apple Inc. However, that does not mean Intel is lagging behind. In fact, the company is racing with its development roadmap to stay ahead of its main competitor, AMD. While AMD has made some serious progress with its processors for CPUs as well as GPUs, Intel isn’t trailing at all. In fact, if the company stays true to the committed roadmap, it could easily race ahead of AMD in the next two quarters of 2019. Speaking about the immediate future of Intel, Jim Jeffers, the Sr. Principal Engineer and Sr. Director, Advanced Rendering and Visualization at Intel Corporation, mentioned:

“The path to exascale computing and content creation has strong interdependencies and opportunities between them. The advancements we make in these areas can be readily shared across both domains to their mutual benefit. Our investment in software initiatives like oneAPI, plus advanced algorithm research and development, plays a profound role as we set out to ‘leave no transistor behind’.”

Intel’s six technology pillars and OneAPI initiative is quite well known in the tech industry. However, Intel has maintained the focus on creators. In fact, Intel’s Xenon ecosystem processors segment is quite popular among the graphics artists in Hollywood. Still, the company announced a wide variety of software tools that will help accelerate workflow. Interestingly, Intel claims it is pushing to achieve its self-set 1000x performance advancement goal. While the goal itself is quite ambiguous, Intel promises to deliver the same within the “next few years.”

Intel Furthers Open Source Software Commitment And Promises OSPRay 2.0 Ray Tracing Engine

At the Intel Create event, the company launched a new version of its high-performance ray tracing kernel as well as a new Open Image de-noising tool. Needless to add, both the tools complement each other, but it is more refreshing to see Intel actively participating in the open-source software development and tools adoption movement. Most of the deep learning and accelerated ray tracers trade high quality for speed. This results in a little extra noise. By offering a de-noising tool alongside the new kernel Intel is attempting to cover both the ends.

If that’s not encouraging enough, Intel is also planning to release OSPRay 2.0. The open-source ray tracing engine could arrive as early as in the next quarter of 2019. In fact, the engine could jointly launch with the eagerly awaited OneAPI. While the initial release of OSPRay 2.0 will have features like asynchronous rendering, the final iteration will include full path traced volumetric lighting and rendering support. Intel is promising the final version of the engine to arrive in the last quarter of 2019.

Pixar, one of Intel’s partners, talked about its collaboration with the company. Together they are working to get Renderman to enable 2x acceleration of the Open Shading Language by Intel Xeon processors (using AVX 512). The Disney-owned company also announced it plans to utilize and extend R&D for Renderman to the upcoming Intel Xe GPU architecture. The Intel Embree and Open Image De-noise in Cinema 4D R21 should deliver faster rendering times. Moreover, Pixar also confirmed a technical partnership for future CPU and GPU technology.

Following are some of the confirmed commitments and deliverables that Intel promised:

  • Embree 3.6 is Intel’s latest version of high-performance ray-tracing kernels which is now available.
  • Intel Open Image De-noise 1.0 which uses AI / deep learning to deliver quality images to speed ray-tracing applications’ rendering time is also available.
  • Intel OSPRay 2.0: The open-source, scalable ray-tracing engine that will incorporate Open Image De-noise 1.0 will be available in the fourth quarter of this year.
  • Intel Open Volume Kernel Library for volumetric rendering and OneApi too will be available before 2019 ends.

Moore’s Law Still Valid But Bell’s Law Being Succeeded By Raja’s Law Which Is Already Applicable?

Intel Create 2019 was mainly about the company’s contribution to the computing and development world. However, Jim Keller and Raja Koduri also spoke on the accelerated pace of the computing industry and how it has rapidly evolved over the years. Interestingly, the widely accepted Moore’s Law, which was created by a former Intel CEO, has been facing some challenges. But Intel strongly believes the law is in effect.

While explaining the validity of Moore’s law, Intel stressed about the Bell’s Law and discussed how the form of computing changes roughly after every 10 years or so. Incidentally, the law has barely been valid owing to the highly accelerated pace and rapid miniaturization of technology. Hence, Intel went ahead and coined a new law which is being called ‘Raja’s Law’.

Named after Intel’s Raja Koduri, the law states that the major macroarchitecture type changes roughly every 20 or so years. Moreover, the exact timeframe continues to shrink with every iteration as technology advancement speeds up. Intel mentioned how the computing world began with the single-core CPUs and slowly moved on to the multi-core, multi-threaded CPUs. Even these powerful CPUs are rapidly being taken over by GPUs, but they too would eventually struggle against discreet Artificial Intelligence. Incidentally, we recently witnessed Microsoft partnering with OpenAI precisely for these reasons.


Leave a Reply

Your email address will not be published.

Close