RSS feed
Posted on 27-2-2012 by Zhenyu Ye Tags: FPGA

FPGA20: Highlighting Significant Contributions from 20 Years of the International Symposium on Field-Programmable Gate Arrays (1992—2011)

HiPEAC internships 2012
Posted on 23-2-2012 by Cedric Nugteren Tags: internships

HiPEAC is announcing new internships for PhD students at companies throughout Europe:

We are pleased to announce the HiPEAC industrial PhD internships in 2012. Any PhD student who is a member of or affiliated with a Hipeac institution in Europe is eligible for these internships. The internship opportunities are provided by the Hipeac member companies in Europe. The duration of the Hipeac-funded internships is 4 months but companies are free to extend the internships at their cost beyond 4 months. The internships have to start and complete in 2012 (you will need to negotiate your start time with the company). The details of the Hipeac internship funding scheme (e.g. how much is the pay and what is covered and so on) can be found here.

As of today, there are 28 internships from 10 different companies. You can see all the internships here, and apply by clicking on each internship title.

Applications are due by March 29th.

Back to Thin-Core Massively Parallel Processors
Posted on 22-2-2012 by Cedric Nugteren Tags: GPU, perspective, parallel processing

December 2011’s issue of IEEE Computer has a nice overview article titled: Back to Thin-Core Massively Parallel Processors. The article gives a perspective on parallel processor history and relates this to current GPU trends. It also gives an idea of what we can see in the near future. The abstract is as follows:

“Examination of the innovations of the past three decades that brought chips to the point at which many-core processors are possible reveals that there are multiple roads ahead, and each is full of challenges.”

Through the Looking Glass: Trend Tracking for ISSCC 2012
Posted on 16-2-2012 by Zhenyu Ye Tags: architecture, solid-state circuits

Through the Looking Glass: Trend Tracking for ISSCC 2012 is an article in IEEE Solid-State Circuits Magazine March 2012. It beautifully puts together the latest trends in academia and industry. Are you still using the old figures in your presentation? It is time to update them.

Structured Parallel Programming: Patterns for Efficient Computation
Posted on 2-2-2012 by Zhenyu Ye Tags: GPU, book, algorithmic skeletons

This Parallel Book heavily uses Intel’s Cilk and Thread Building Block model as examples. Yet it is generic enough to apply their patterns in other programming models, e.g. OpenCL. The algorithmic patterns used in the book are summarized in their SuperComputing 2011 turorial (PDF).

Xilinx Kintex-7 KC705 and Virtex-7 VC707 Evaluation Kit
Posted on 25-1-2012 by Zhenyu Ye Tags: FPGA

The Xilinx Kintex-7 FPGA KC705 Evaluation Kit is now available for order, at the price of $1690, with 5 week factory lead time. The Xilinx Virtex-7 FPGA VC707 Evaluation Kit, at the price of $3495, is going to be available at avnet.

  • Kintex-7 KC705
  • Virtex-7 VC707
'CUDA Application Design and Development' book available
Posted on 12-1-2012 by Cedric Nugteren Tags: GPU, CUDA

Recently, a new CUDA book was released titled: CUDA Application Design and Development.

The book starts off with a number of tutorial-like chapters, but addresses more interesting features towards the end. For example:

  • Chapter 6: Functor-based example to write your own generic methods
  • Chapter 7: Multi-GPU, asynchronous kernel execution, and Unified Virtual Address (UVA) spaces
  • Chapter 8: CUDA for multi-core processors and GPU computing with high-level languages (e.g. Python, Java, R)
  • Chapter 9: Accelerate visualization and gaming applications
  • Chapter 10: MPI cluster computing

Sciencedirect allows you to browse the book in PDF form online:

Embedded programmable GPUs
Posted on 10-1-2012 by Cedric Nugteren Tags: GPU, SIMD, architecture, Tegra

Embedded GPUs programmable in OpenCL or CUDA are slowly starting to become available.

First of all, there is the new CUDA on ARM dev-kit, which is named ‘CARMA’. The board contains a Tegra 3 quad-core ARM and a Quadro 1000M Fermi GPU with 96 CUDA cores. The GPU is programmable in OpenCL and CUDA. Details are found on their website. The dev-kit is expected to be available in Q2 2012.

Second, there is a new version of the ZiiLABS ZMS board, the ZMS-40. This is a system on chip with a quad-core ARM and a 96 SIMD cores. The SIMD cores are programmable with OpenCL. More information can be found on the ZiiLABS website. The ZMS-40 SoC is available in a tablet running Android ICS.