More async

A project log for Libre Gates

A Libre VHDL framework for the static and dynamic analysis of mapped, gate-level circuits for FPGA and ASIC. Design For Test or nothing !

yann-guidon-ygdesYann Guidon / YGDES 05/26/2023 at 11:0610 Comments

@SHAOS  made me discover the "C-element" in his log Trying other ways to make chips. I don't know why I hadn't heard about it yet. It's something to keep in mind for when I have resurrected the project and after I added the ALI. The tools have a setback for a few reasons, one of them being the need to support asynchronous logic and clock domain crossings... This way I'll be able to fully characterise the debug interface of the #YGREC8.


SHAOS wrote 05/26/2023 at 16:13 point

BTW I don't think C-gate is possible in VHDL level - at least not for FPGA where fully synchronous nature of the circuits is a must...

  Are you sure? yes | no

Yann Guidon / YGDES wrote 05/26/2023 at 19:24 point

Almost anything is possible in VHDL, if you can write a logic diagram for it. There is a logic loop but one can coerce a synthesis tool to accept it (i've done this before).

LibreGates should be able to process loops too (provided a "meta gate" is included) because I need to handle asynch circuits for the Y8's debugger.

  Are you sure? yes | no

SHAOS wrote 05/27/2023 at 03:36 point

it might be closely tight to specific chip and virtually unportable to other chip families or even other chips of the same family - I'm familiar with such approach when workability of synthesized bitstream depends on order of modules listed and phase of the Moon - you adds something minor and design stops working forever :)

  Are you sure? yes | no

Yann Guidon / YGDES wrote 05/27/2023 at 12:27 point

@SHAOS  gates are gates... I assemble them however I want :-)

Furthermore the asynchronous types are used only in certain parts of the debugging circuit of Y8. The portability issues are limited to only a tiny part of the system :-)

  Are you sure? yes | no

SHAOS wrote 05/27/2023 at 20:25 point

> gates are gates...

Will synthesizing tools optimize out some of your gates when convert your logical circuits to LUTs?

  Are you sure? yes | no

Yann Guidon / YGDES wrote 05/27/2023 at 21:25 point

@SHAOS  Synthesis tools will detect and complain about "logic loops".

There is a trick however : at a certainly low description level, with LibreGates, the use of fundamental gates for the netlist might bypass some checks that are done at a higher logic level.

Synthesisers are not meant to remove fundamental gates, particularly if they are manually defined. Otherwise there must at least be an error or warning message and the synthesiser risks changing the behaviour of the circuit, since the synthesiser might not know exactly what the loop is meant for.

Logic loops should be kept very small, usually 2 or 3 gates long. If there is a trouble, the equivalent function could/should be turned into a macrogate or something like that, in case the synthesiser is really uncooperative. Longer asynch loops should be avoided, of course.

As for LUT, this is something quite specific and it requires tool-specific intervention. I had no trouble doing manual set/reset latches with Microsemi's Libero. Using manual insertion of fundamental gates, it was easy to solve.

  Are you sure? yes | no

SHAOS wrote 05/28/2023 at 03:50 point

What do you mean by "fundamental gates"?
Low level library components for specific FPGA?

  Are you sure? yes | no

Yann Guidon / YGDES wrote 05/28/2023 at 04:04 point

@SHAOS  Yes,

this project started indeed as a replacement library for the A3P family, for which Actel has a whole catalog of "gates" that are recognised by the synthesiser and that map directly to the hardware.

My interest was in efficiency (speed, space) but without compromising the portability of my code, which could become trapped by their tools. So I wrote my own library, open sourced, derived from the public documentation. So I could write Open Source FPGA code that could always be used on other architectures. I can also use GHDL instead of the manufacturer's proprietary and limited toolset.

The rest if feature creep :-)

  Are you sure? yes | no

SHAOS wrote 05/28/2023 at 04:35 point

ok, then you need to try TinyTapeout or even directly apply to efabless' GP180 or SKY130 free shuttles to become "silicon proven" :)

  Are you sure? yes | no

SHAOS wrote 05/26/2023 at 16:11 point

I even did C-gate with hysteresis constructed out of discrete PMOS and NMOS transistors in my 1st chip in 2015 :)
And it was working pretty well ;)

  Are you sure? yes | no