Can I use openCL to develop an FPGA Accelerator device for an mac application

[Altera_Forum]

Hello, everyone, 

 We have an application run on our iMacs and macBooks. It is X11 application, mainly written in C and Fortran language, running on Unix environment. what we want to do is to use FPGA to accelerate its image processing.   It seems that Altera openCL SDK and RTE only support Windows64 and Linux64. Is there any solution to use Altera openCL SDK to develop an accelerator running on mac? (Windows64 developing environment + running on mac is acceptable.)  Thanks!  Bests, Qingyuan Tian(QT)

[Altera_Forum]

The OpenCL PCI-E driver which is required to communicate with the FPGA when using OpenCL will very likely not compile on Mac. You can always try to modify and compile it manually, and it might or might not work in the end, but Altera/Intel does not officially support it. 

 Also you will need to compile your OpenCL host code against Altera's OpenCL runtime, so you will eventually need to install Altera's tools on your machine. They have added support for ICD driver recently, so it might be possible to compile the host cost with other OpenCL runtimes (AMD, Intel CPU runtime, Nvidia, etc.), but I was not successful when I tried to do it myself.

[Altera_Forum]

Altera makes 3th way -- from embedded ARMv7 in FPGA, booting from SD-card and running with Linux. 

If you may compile kernels from Win/Lin and write its to SD-flash, then may insert big PCIe board to Mac and send control manipulations through ARMs through PCIe, write data arrays to shared memory and read back results from it. However, I recommend NVIDIA GPUs -- more faster for image processing.

[Altera_Forum]

--- Quote Start ---  Altera makes 3th way -- from embedded ARMv7 in FPGA, booting from SD-card and running with Linux. If you may compile kernels from Win/Lin and write its to SD-flash, then may insert big PCIe board to Mac and send control manipulations through ARMs through PCIe, write data arrays to shared memory and read back results from it. However, I recommend NVIDIA GPUs -- more faster for image processing.  --- Quote End ---    That would only work for SoC FPGAs that have an ARM core. The majority of high-end FPGA boards only have a non-SoC FPGAs and will need to be connected to a standard x86 host to be usable with OpenCL.  And of course, if you are comparing high-end Nvidia GPUs to low-end SoC FPGAs, the GPU will be much faster for any workload, not just image processing.