Blackmagic Forum

Hi. I've had an interest for decades in designing a microprocessor, and have been getting closer to working out a good theoretical solution to incorporate all types of processing. I've got a curiosity to find a functional specification tool, that can output the specification as C, (cpu code, GP-GPU, Open CL preferable) FPGA, ASIC, and whatever else to experiment.

I'm also looking at a better target than FPGA, as it tends to have poor price and energy to performance ratios. ASIC is too much, and nobody seems to have figured a low cost way to do samples. They used to use gate arrays years ago, which were cheaper, and less performance, but still a quantity product. I remember they were moving to techniques to wire a gate array after delivery, but I can't seem to find anything substantial out there.

Anybody know good functional description tools which can output code and designs, and faster alternative to FPGA?

Thanks.

I don't have personal experience with it, but have you checked out Mathworks? I believe it can output code in C or Fortran, as well as VHDL.

As for a faster alternative to FPGA, so far as I know the only alternative would be an ASIC. There used to be an organization called MOSIS that would make low-volume, low-cost, prototypes; you had to do the design yourself, though.

I'm a little confused, though: when you say, "They used to use gate arrays," that makes it sound as though you think an FPGA is not a gate array. You do know that FPGA stands for Field Programmable Gate Array, don't you?

Thanks for the heads up there Lain.

FPGA is a different beast to old gate array, which is a goer if you have enough numbers.

FPGA tends to be slow and power hungry. I knew a guy that would do world performance leading processors on gate array, but it is like mist trying to get an FPGA to emulate a performance processor, or ultra low energy performance. He currently does world leading low energy (at release, with an old large node process) CPU as an alternative to FPGA. But his processors are strange and not right for my purposes.

Wayne Steven wrote:But his processors are strange and not right for my purposes.

What are you trying to build then?

As per OP, general purpose computing architecture that can act as low energy replacement to most things. The issue is to get the design concepts and ISA right first. Then use it for most things. I've been around some good people in this, and am building upon the branch of work they were in,
which goes back to an early computing model in the 1950's. Very fine balancing to efficiently get things done. With the current physical handicaps it remains a development curiosity until implemented. I wanted to implement it into an magnetic fpga, but looks like unless I develop one it would still take a while longer again for one to come out. I wanted to use an fpga technology that worked 10x the speed 15 years ago, but that disapeared into the military, as did two other lens designs I was waiting to use after that. Since then I got sicker.