Maybe you should clarify the question... are you going to use an FPGA anyways and just deciding whether to get one with integrated CPU cores? Or are you deciding between using an ARM vs. using an FPGA and implementing a soft processor on that? In that case, as long as you don't integrate any custom logic in the FPGA fabric, using an FPGA just to implement a soft processor is rather pointless as it will perform much worse than the ARM.

In case your deciding between FPGA with or without dedicated processor cores: I have to admit I'm a fan of FPGAs with integrated, dedicated ARM cores (the Zynq family from Xilinx, Intel, formerly Altera, has them in some of the devices of each product line).

Reasons:

- dedicated processors cores are way faster than soft processors like MicroBlaze or NIOS (at least by a factor of 6-7)

- you don't spend FPGA resources

- a more compact design flow (no adding and configuring of common peripherals)

The sole downside is that such devices usually are bit more expensive.

Dear Johannes Hiltscher ,

Thanks for the detailed answer, of curse I'am deciding between two FPGA techniques, one with hard processor and another with soft processor.

Using ARM processor alone is probably not useful in high speed applications, but with some logic, one can boost the operation.

Zenq FPGAs are Superb especially industrial kits from Trenz Inc., being small and powerful.

,

Regards,

Dear Hussein,

maybe a little addon: in case you already have FPGA hardware available, then a soft processor is a reasonable choice when you need programmatic control and the available hardware otherwise suits your needs.

In case you're planning to acquire a new board, embedded processor cores should be preferred IMHO. Unless there are other constraints, like resources vs. cost considerations (it may be that one, e.g., Zynq, doesn't offer sufficient resources so you have to go for the next bigger device which might be unavailable or cost much more...).

And yeah, Trenz has a nice portfolio. But I wouldn't say that an ARM is inadequate for high speed applications - this depends very much on the actual SoC and the task. E.g., nVidias' Tegras are quite powerful. If you make use of NEON you can even get quite some power from a RaspberryPi (I ran a lane detection algorithm on an RPi 3 at >50Hz).

Best regards,

Johannes

Dear Hussein M. H. Al-Rikabi

If you are looking for a good performance I recommend you take a look at the Zynq Ultrascale + families. The Ultra96 board costs almost half of the ZedBoard, and has more resources. The features of the FPGA device of this board can be found at https://www.xilinx.com/products/silicon-devices/soc/zynq-ultrascale-mpsoc.html Zynq UltraScale + EG EG devices feature a quad-core ARM® Cortex-A53 platform running up to 1.5GHz. Combined with dual-core Cortex-R5 real-time processors, a Mali-400 MP2 graphics processing unit, and 16nm FinFET + programmable logic.