Entries tagged as: Developers

AMD FirePro Tech Talks at SIGGRAPH 2013

Posted by Tony DeYoung on July 10, 2013

The AMD FirePro team will host two 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
Interactive decoding enables real-time adjustment in scene-referred linear light of color balance, contrast, and effective exposure of high-dynamic-range ACES.  Efficient coding/decoding using floating-point provides extra precision to support these features.

FirePro Technologies for Visual Computing
Wednesday, 24 July | 11:15 am-12:15 pm
Visual Computing is the combination of two key assets from AMD: compute and graphics. Visual Computing needs numerous specific features in graphics and compute in order to become a new experience in digital content creation, multimedia, or visual simulation. AMD FirePro Technology provides these key features, delivering the best experience in Visual Computing.

RGK geometric 3D kernel for CAD supports CPU multi-threading and GPU acceleration using OpenCL

Posted by Tony DeYoung on June 27, 2013

Geometric 3D kernels are software components designed to represent and precisely model physical objects in 3D on computers.  From these kernels, software companies develop CAD/CAM CAE end user applications. Examples include Parasolid, ACIS, CATIA CGM and ASM.

This paper describes RGK (Russian geometric kernel), a parallel computation kernal that has support for multi-threading (CPU) and GPU acceleration using OpenCL.  The RGK kernel has the full set of tools to create models using solid, surface, and wireframe modeling. Because it uses parallel computations, the application can work simultaneously with these different types of data within the same model.

On the GPU, RGK supports OpenCL to handle functions to compute mass-inertia characteristics of bodies, projections, searching model elements upon the given point, NURBS computations, etc.  New parallel algorithms can easily be introduced to keep up with any newly introduced hardware capabilities. The kernel also supports visualization techniques for models including model tessellation and mesh generator functions using the GPU.

RGK supports 32- and 64-bit architecture, and Windows and Linux platforms. It can be compiled with any С++ compiler that implements features of С++11 standard.

Below is an example video demonstrating geometric constraints in assembly and kinematics simulation

Tags: CAD, Developers

AMD FirePro Technology SDK offers repository of source code samples

Posted by Tony DeYoung on August 23, 2012

AMD has released the FirePro Technology SDK, which is a a repository of samples with complete source code intended for educational purpose. The base code is up to date with the latest specification and can be seen as a guideline for good coding practice, to achieve the best performance from your graphic card.

The first round of code samples includes:  OpenGL textured cube, OpenGL lighting, OpenGL advanced rendering techniques, OpenGL tessellation, OpenCL transfer compute overlap, OpenGL QuadBugger stereo rendering, OpenGL sparse texture, OpenGL frame lock and OpenGL 10-bit per component surfaces.

This code is released free of charge for use in derivative works, whether academic, commercial, or personal.

Tags: Developers

Fabric Engine new DCC app ‘Creation Platform’ to debut at SIGGRAPH 2012 in AMD FirePro booth #709

Posted by Tony DeYoung on July 31, 2012
At this years SIGGRAPH, Fabric Engine will publicly debut the new Creation Platform framework for building custom, high-performance graphics / DCC applications. Creation is built on top of Fabric’s multi-threading heterogeneous compute engine.

Creation Platform will be shown in the AMD FirePro booth (#709) at SIGGRAPH 2012 next week.

Key benefits of Creation include:

  • High Performance: Creation is built on top of the Fabric Core Execution Engine, a multi-threading engine designed to get exceptional performance out of both CPUs and GPUs.
  • Modularity: Creation provides a set of building blocks that can be connected together to build a wide range of different applications.
  • Extensibility: Creation’s extension system enables developers to incorporate existing code libraries into their applications. This enables Creation to incorporate things like Bullet Physics and streaming of motion capture data, as well as allowing for support of custom file types like Alembic and LIDAR.
  • Accessibility: The majority of Creation applications are built using Python and Qt. For the critical performance parts of an application, Creation uses a custom Kernel Language (KL), which a simple, high-level language. If a TD or developer is comfortable with Python, they will find KL easy to pick up.

Small scale clustered computing ecosystem broadens: AMD FirePro,  LS-DYNA, CAPS HMPP compiler

Posted by Tony DeYoung on July 12, 2012

AMD is very active in advancing performance computing (see recent award for DOE Extreme-Scale Computing Research and Development Program).  Part of these efforts entail broadening the small scale clustered computing ecosystem including recent optimizations for OpenCL acceleration on AMD FirePro GPUs and APUs for LS-DYNA simulation software and the HMPP directive-based compiler from CAPS.

LS-DYNA is a general-purpose finite element program capable of simulating complex real world problems. It is used by the automobile, aerospace, construction, military, manufacturing, and bioengineering industries to run simulations that are complex and compute-intensive like auto crashes and explosions. The beta version supports OpenCL acceleration on AMD FirePro graphics (general availability in 3rd quarter) .

CAPS Enterprise HMPP compiler now fully integrates data-parallel backend for OpenCL and adds support for AMD FirePro GPU and APU devices. CAPS technology helps users take advantage of AMD Accelerated Parallel Processing Math Libraries (APPML), the software libraries containing FFT and BLAS functions designed to run on AMD GPUs.  Initial performance benchmarks in functions such as image-processing using Sobel Filters show a 40X increase in performance. With this new release, compiler directives can be used to leverage AMD FirePro capabilities while preserving C or FORTRAN codes.

Article: How does Order Independent Transparency Accelerate Creo 2.0

Posted by Tony DeYoung on April 17, 2012

Order independent transparency or “OIT” in computer graphics programming terminology denotes any technique that can correctly render overlapping semi-transparent objects without having to sort them before they are being rendered. Rendering semi-transparent objects has always been a problem because the blending operation is order dependent: when a semi-transparent fragment is rendered, the underlying color (i.e. the background) is crucial for the final color to be correct. 

This article and accompanying white paper, explain how AMD implemented the OIT technique in PTC Creo Parametric 2.0. The technique uses standard ARB OpenGL 4.2 extensions like Atomic Counter and Shader Image load/store which enables access to texture buffer directly from the shader. The implementation allows for pixel accurate rendering of overlapping semi-transparent objects without having to sort them before they are being rendered, providing up to 9 or 10 times performance improvement when rendering transparency in Creo Parametric 2.0 versus blended rendering in PRO/Engineer Wildfire5.0.

AMD FirePro Development User Community: Share tips and best practices

Posted by Tony DeYoung on February 09, 2012

AMD launched a FirePro Development user community for end-users, developers and AMD experts to share tips and best practices about the FirePro Vx800 and Vx900 graphic cards. The newly updated community page greatly improves users’ ability to ask questions and get answers on any FirePro related topic:

For example, here’s an interesting post just answered:

I’m trying to solve a very specific problem. For this, I need to run 4 displays at 120Hz, and have the vsync for each trigger a quarter of a frame after the previous:
Display 0, start frames at 0s, 4/480s, 8/480s, ...
Display 1, start frames at 1/480s, 5/480s, 9/480s, ...
Display 2, start frames at 2/480s, 6/480s, 10/480s, ...
Display 3, start frames at 3/480s, 7/480s, 11/480s, ...

Now the question is, can I do this with one S400 and one or more graphic cards?

1: Yes, you can connect up to 4 GPUs to a single S400 board.
2: You can genlock multiple displays that are connected to a single GPU as long as the connected monitors are identical.
3: The accepted input range for the house sync is between 15 and 120 H

Top 10 Reasons to Attend the AMD Fusion 2012 Developer Summit

Posted by Tony DeYoung on January 24, 2012

The AMD Fusion12 Developer Summit runs June 11-14, 2012 in Bellevue, Washington. At the summit you'll learn how to better leverage heterogeneous computing to advance your projects, see the latest advancements in OpenCL, C++ AMP, and Heterogeneous System Architectures.


Top 10 Reasons to Attend the Summit.

10 Ten tracks reflecting the topic areas you care about. Tracks include heterogeneous computing, cloud computing, gaming and consumer graphics, and more.

9 Keynotes packed with insights and never-before-seen demonstrations. Witness senior leaders from across the industry share what's next, now.

8 OpenCL tips and information from some of the industry's best. Learn how to leverage OpenCL to help unleash the full potential of your developments faster and more efficiently than ever.

7 Training you'll put into practice the day you return to work. More than 90 sessions packed with deep, content-rich data. Plus, pre-conference tutorials and hands-on labs.

6 Unmatched access to AMD technologists. AMD technologists present keynotes, teach training sessions and are ready to talk technology at the social events. Working at AMD would be the only other way to get this level of access.

5 Learn what the industry is doing. Come see the major players at AFDS. They present keynotes and lead technical sessions. Attending AFDS helps you better understand the marketplace and where the ecosystem is putting their resources so you stay competitive.

4 Your next partnership could start here. With more than 1,200 technologists expected to attend, there's no shortage of opportunities for you to strengthen current partnerships, refresh old ones, and start collaborating with future partners.

3 First to know. First to market. Attending AFDS gives you details on AMD's product roadmap. Your chance to lead the market begins here.

2 Venture capitalists will be participating. If you've got the next big thing, AFDS is your chance to meet some of the venture capitalists that help start-ups start, and keep small companies growing.

1 This is the center of the heterogeneous computing universe. If you're ready to use heterogeneous computing to redefine what your projects can do, to reset what you think technology will be like a decade from now, and reshape what consumers expect in every device they own, the technology and technologists you seek are at AFDS.

Learn more and sign up at amd.com/afds

Download and share a PDF of this Top Ten List

OpenCL & the Future of Desktop High Performance Computing in CAD: On-Demand Webcast + Slides:

Posted by Tony DeYoung on November 18, 2011

The slides and webinar recording of ‘OpenCL & the Future of Desktop High Performance Computing in CAD’  is now available on-demand. DesignWorld and AMD hosted this heavily attended event.  It examines the details of CPUs and GPUs, explore their differences and similarities, and highlight the computing power they can provide. It will also look specifically at OpenCL, what it is, what it does, and how this new computing interface will change the way software developers create software and help end users fully realize the compute power contained within today’s modern desktop computers.

OpenCL v1.2 enables GPU partitioning, support for DSPs, enhanced OpenGL & DirectX integration

Posted by Tony DeYoung on November 15, 2011

The Khronos Group today announced the public release of OpenCL 1.2.  The new release adds many new features including the ability to partition a device (i.e. a GPU) into sub-devices to allocate resources to high priority/latency-sensitive tasks, or effectively use shared hardware resources such as a cache. It also adds integration with specialized or non-programmable hardware and associated firmware, such as video encoder/decoders and digital signal processor.

AMD has been the leader driving OpenCL as a cross-vendor, non-proprietary solution for accelerating applications on CPU, GPUs and APUs. Congratulations on the release of v1.2 which will offer better performance and more flexibility for heterogeneous computing and multi-core architectures.

Page 2 of 5 pages  <  1 2 3 4 >  Last »