Khronos announced the provisional release of OpenCL 2.0. OpenCL provides an open, royalty-free standard mechanism for software to tap into the computational power of graphics chip. AMD has been working with many ISVs to implement OpenCL in their software and take advantage of FirePro Compute + Graphics capabilities. Several DCC partners are showing at SIGGRAPH 2013 including Adobe, Autodesk and Optis.
One of the complexities of using the GPU and OpenCL for general-purpose software is the need to write software that runs on both the GPU and CPU. Programmers have to figure out how to coordinate the GPU and CPU components and to use data in memory. The new OpenCL 2.0 helps solve this by letting programmers share complex data structures in shared virtual memory rather than rely on the slower and more work-intensive method of transferring data from one component to another. OpenCL 2.0 also can schedule new computing tasks using Dynamic Parallelism, rather than waiting for the CPU.
Below is a list of the major enhancements in OpenCL 2.0:
Shared Virtual Memory
Host and device kernels can directly share complex, pointer-containing data structures such as trees and linked lists, providing significant programming flexibility and eliminating costly data transfers between host and devices.
Device kernels can enqueue kernels to the same device with no host interaction, enabling flexible work scheduling paradigms and avoiding the need to transfer execution control and data between the device and host, often significantly offloading host processor bottlenecks.
Generic Address Space
Functions can be written without specifying a named address space for arguments, especially useful for those arguments that are declared to be a pointer to a type, eliminating the need for multiple functions to be written for each named address space used in an application.
Improved image support including sRGB images and 3D image writes, the ability for kernels to read from and write to the same image, and the creation of OpenCL images from a mip-mapped or a multi-sampled OpenGL texture for improved OpenGL interop.
A subset of C11 atomics and synchronization operations to enable assignments in one work-item to be visible to other work-items in a work-group, across work-groups executing on a device or for sharing data between the OpenCL device and host.
Pipes are memory objects that store data organized as a FIFO and OpenCL 2.0 provides built-in functions for kernels to read from or write to a pipe, providing straightforward programming of pipe data structures that can be highly optimized by OpenCL implementers.
Android Installable Client Driver Extension
Enables OpenCL implementations to be discovered and loaded as a shared object on Android systems.
Join the HSA Foundation BoF at SIGGRAPH on July 24th at 1pm at the Anaheim Convention Center ( Room 202 B) to hear Phil Rogers, President of the HSA (Heterogeneous System Architecture) Foundation and AMD Fellow, present on the state of the HSA Foundation. Don’t miss out on this opportunity to learn about the HSA Foundation’s work to date in making heterogeneous computing natural, pervasive and fun and participate in a Q&A with foundation members. For more information, contact admin [at] standards.hsafoundation.com.
The AMD FirePro booth #521 at this years SIGGRAPH 2013 is going to be packed with demos this year demonstrating the power of the AMD FirePro W series. Check out the GPU-accelerated demos from software industry leaders as well as new software firms. There are also several cutting edge display and projection hardwares worth checking out.
What to look for at a glance:
6 Sharp 32" 4K Ultra-HD displays driven by a single AMD FirePro W9000 card: 9252 x 4320 resolution = 40 million pixels at 60fps
OpenCL real-time 12-bit color processing at 4K running on the first high frame rate 4K DLP projector (Christie D4K2560) driven by a FirePro W9000. Gary Demos will be presenting.
Windows 8.1 (Blue) tiled resources using partially resident textures in DirectX 11.2 running on on a Dell T5600 and AMD FirePro W7000
AMD Radeon Sky Series for Cloud Gaming
OpenCL-accelerated OPTIS THEIA-RT for real-time physics-based ray-tracing running across 6 30"displays using AMD FirePro W9000
Mitusbishi 6 46" LCD display wall across using Playout Composer running on a Dell T5600 and AMD FirePro W9000
OpenCL-accelerated facial tracking with Mixamo Face+ running on on a Dell T5600 and AMD FirePro W7000
WebGL-based virtual globe rendering by Cesium running on a Dell T5600 and AMD FirePro W7000
Autodesk 3ds max 2013 running on a Dell T5600 and AMD FirePro W8000
OpenCL-accelerated MayaBullet physics simulation plug-in for Autodesk Maya 2013 running on a Dell T5600 and AMD FirePro W8000
OpenCL-accelerated Adobe Premiere Pro CC for real-time video production in HD and 4K running on a Dell T5600 and AMD FirePro W7000
Enhanced OpenCL and OpenGL support in Adobe PhotoShop CC for 10-bit high dynamic range imagery and accelerated special effects running on a Dell M6700 PremierColor w/ FirePro M6000 Mobility Pro Graphics
AMD OpenCL Developer Tools demos running on a Dell T5600 and AMD FirePro W8000
Also be sure to check out the two AMD FirePro tech talks at SIGGRAPH 2013.
Real-Time ACES Decoding and AMD FirePro GPU-Enabled RRT/ODT Processing at 4K Tuesday, 23 July | 11:15 am-12:15 pm
FirePro Technologies for Visual Computing Wednesday, 24 July | 11:15 am-12:15 pm
AMD and Adobe recently announced expanded OpenCL support for Adobe Premiere Pro CC at NAB 2013, including multi-stream and mixed format 6X accelerated workflows with real-time effects, color grading and finishing with hardware-accelerated rendering to the chosen final destination format.
Premiere Pro CC: Today AMD announced support for additional OpenCL-accelerated features with the announcement of Adobe Premiere Pro CC including OpenCL hardware-powered rendering with Adobe Media Encoder CC for Adobe Premiere Pro CC projects and multi-GPU support. This enables users to queue up multiple Adobe Premiere Pro CC projects to be rendered using all available GPU horsepower in the background. There are now over 40 GPU-accelerated features using OpenCL in Premiere Pro CC.
PhotoShop CC: New and enhanced Adobe Photoshop CC features have also been added to the dozens of existing open hardware accelerated features in Adobe Photoshop CC:
Adobe's all-new Smart Sharpen filter leverages OpenCL and AMD graphics technology to sharpen images faster1 while minimizing halo effects on distinct edges and providing precise noise reduction control
The enhanced Blur Gallery harnesses the power of OpenCL and AMD graphics technology to deliver mouse-down previews quickly, accurately and with high on-screen resolution of up to a million pixels. High quality instant on-screen and in-context feedback allows artists to work more efficiently without unnecessary disruptions to the creative workflow. Non-destructive hardware-accelerated Blur Gallery effects are applied as Smart Objects to create a desired look for images and even video within Photoshop CC
Performance enhancements have also been made to the Liquify effect so that creative professionals can now push, pull, rotate, reflect, pucker, or bloat selected image areas to achieve unique artistic effects with greater speed2 while working with even the largest images
Both Adobe Photoshop CC and Adobe Premiere Pro CC fully support advanced display technologies such as 4K and AMD Eyefinity multi-display technology.
Desktop Engineering has published an article on CPU vs GPU computing for engineering and CAD. The article discusses using the Open Standard OpenCL programming framework, as well as Nvidia’s proprietary Cuda option. OpenCL can be used to give an application access to a GPU for non-graphical computing , such as engineering data analysis or simulation. Importantly offers the ability to use both CPUs and GPUs in combination.
AMD has integrated OpenCL as its programming framework for its FirePro family of GPUs, as well as its CPU and APU offerings. As noted by BSN AMD’s focus on OpenCL puts them in a competitive advantage against Nvidia as many of the compute programs out there tend to run faster on AMD hardware if they are optimized correctly.
This PDF article on "The OpenCL Revolution" describes how 10 leading DCC, CAD and simulation software applications tap OpenCL running on AMD Firepro GPUs to accelerate their applications and enable new functionality.
The summary takeaways:
OpenCL on the GPU speeds up complex tasks
OpenCL opens up new ways of working, in some cases cutting computation time form days to minutes
OpenCL is hardware-agnostic
OpenCL enables write once, run anywhere accelerated code
OpenCL cuts development time
Mobile developer back OpenCL
AMD FirePro GPUs fully support the latest versions of OpenCL and AMD actively works with software developers to enable their applications for OpenCL/GPU acceleration
Some of the leading CAD, DCC and simulation tools currently use OpenCL
Key tasks accelerated
Speed Boost over CPU
Executing MATLAB code
Simulation of particulate materials
Blend Shape deformation
Side Effects Software
Fire and smoke simulation
Painting and image processing
Video editing and processing
Vratis announced a collaboration with the AMD FirePro team to bring OpenCL acceleration to Vratis’ SpeedIT technology for OpenFOAM and physical simulations.
The library can be used in a wide spectrum of domains such as computational fluid dynamics (CFD), electromagnetics, thermodynamics, materials, acoustics, computer vision and graphics, robotics, semiconductor devices and structural engineering.
The implementation of OpenCL support will bring OpenFOAM to the wider range of users and provide the first steps towards acceleration of CFD on OpenCL-enabled devices.
From the press release: “With SpeedIT, Vratis has demonstrated its ability to develop products that are capable of efficiently utilizing the massively parallel compute capability of today’s graphics processors,” said David Cummings, senior director of AMD Professional Graphics. “We want to support their efforts to move their leading-edge SpeedIT product into an OpenCL-based environment so that a much larger user base has access to a tremendous tool for computational fluid dynamic research and simulation.”
The AMD 2013 Developer Summit has been announced. The Developer Summit is the evolution of the very successful AFDS held for the past two years in Bellevue WA. Last year there were more than 150 heterogeneous computing technical sessions along with keynote presentations by Microsoft, ARM, Adobe, and Cloudera as well as AMD.
This year the summit moves to San Jose CA and the dates change from mid-June to November 11-14. AMD is accepting presentation proposals until March 15, 2013. Topics can be on any of the following:
FireUser.com is a community resource for CAD, visualization, 3D, video and engineering professionals to learn about the latest acceleration and display technologies and news with a focus on the AMD FirePro workstation graphics line.