Select the search type
 
  • Site
  • Web
Search
You are here:   Home
  |  Login
Get More Insight

Hybrid DSP provide hardware and software tools for research and development groups and organizations.  We allow you to:
  • Gather and generate more data through use of the highest speed digitizers and waveform generators (ADC/DAC)
  • Store more data faster through tried and tested SSD and HDDs in various RAID configurations
  • Process more data on FPGA, CPU and GPU
CUDAfy .NET – General Purpose GPU on .NET

Hybrid DSP’s CUDAfy .NET tools and libraries allow easy development of high performance GPGPU applications completely from the Microsoft .NET framework. 

Write GPU Code in .NET

Modern graphics cards provide the potential of massive speed increase over CPUs for non-graphics related intensive numeric operations. Many large data set operations such as matrices can see a 100x or more speed up. CUDAfy allows .NET developers to easily create complex applications that split processing cleanly between host and GPU.  New OpenCL support allows CUDAfy to run on AMD GPUs and Intel CPUs and most applications will need no modification to take advantage of this new feature.  CUDAfy is the easiest way by far to use the CUDA programming model on any platform. 

Download CUDAfy



News
08 May 2013: An article called "Why use FPGAs for data acquisition systems?" has been posted in the new Articles section of this website.  It's a good question considering that FPGAs are relatively tough to work with.  But the sheer performance and relatively low price of new FPGA daughter cards makes the extra effort worthwhile. 

07 May 2013: Version 1.22 of CUDAfy is the first official release with support for OpenCL.  There are a bunch of bug fixes and other new features like integer intrinsics and code insertion.

23 Apr 2013: Benchmarking OpenCL and CUDA using CUDAfy on GTX Titan - an article from John Michael Hauck where he tried CUDAfy on GTX Titan in both CUDA and OpenCL mode. He also compares against the .NET parallel library on a 8-core Xeon CPU. An entertained read. 

17 Apr 2013: A range of high performance digitizers and waveform generators (ADCs and DACs) are now featured on this website. When used with supported Xilinx FPGA cards and PCI-Express drivers fast, low latency transfers can be made to and from NVIDIA GPUs using our GPUDirect implementation.

16 Apr 2013: V1.21 Beta is supplied with an installer that also checks your system is correctly configured for OpenCL and/or CUDA targets.

04 Mar 2013: V1.20 Beta with new OpenCL support is available on the download page or codeplex.  For many applications you only need to change a singe enum to target an AMD GPU or Intel CPU.  Remember to install the relevant OpenCL SDK from AMD or Intel! Have fun and let us know how you get on.

25 Jan 2013: CUDAfy featured on Dotnet Rocks. Carl and Richard interview Evan Hauck. Listen to the show here. Interview is at about 5 minutes from start.

22 Jan 2013: NVIDIA GPUDirect support is available for 4DSP FPGA and I/O cards (ADCs, DACs, frame grabbers).  Using GPUDirect improves bandwidth and reduces latency between GPUs and the 4DSP cards.  Email or phone for more info.

14 Jan 2013: CUDAfy featured at CodeMash 2013 with John and Evan Hauck.  

13 Jan 2013: We're working on OpenCL support and so far it's looking good. This will allow CUDAfy applications to work on AMD GPUs as well as Intel and AMD CPUs.  Little or no change is required to current CUDAfy apps. If you'd like to beta-test then please get in touch. 

16 Nov 2012: Binomial Option Pricing sample available - This finance sample project is based on the NVIDIA CUDA C sample of the same name. When compared to the CUDA C code it demonstrates just how easy writing CUDAfy applications is. A beta of CUDAfy V1.13 is included (allows non-public methods and members to be CUDAfied). CUDA 5.0 and a GPU with compute 1.3 or greater is required. http://cudafy.codeplex.com/documentation