If let's say the enumeration of inputs took several lines, then all
after the first one were ignored. Since the first line ended with a
comma, an error was generated when trying to use the resulting file.
Signed-off-by: YRabbit <rabbit@yrabbit.cyou>
prjtrellis documentation shows that EBR clock inputs have optional inverters.
The bram techmap outputs those parameters, and nextpnr consumes them. But for
whatever reason, Diamond doesn't include those parameters in its blackbox
models. This makes synth_lattice fail when targeting ECP5 with a design that
maps block RAMs if you include any pass that needs cells_bb_ecp5.v's definitions.
This change fixes up the ECP5 bram blackbox models at generation time, by
adding the missing parameters back in.
Signed-off-by: David Anderson <dave@natulte.net>
Primitives that are not planned for implementation for reasons of
belonging to old unsupported chips or representing composite complex IPs
rather than primitives are removed.
Also latches and large MUXes not planned for implementation.
Signed-off-by: YRabbit <rabbit@yrabbit.cyou>
- Techlib pmgens are now in relevant techlibs/*.
- `peepopt` pmgens are now in passes/opt.
- `test_pmgen` is still in passes/pmgen.
- Update `Makefile.inc` and `.gitignore` file(s) to match new `*_pm.h` location,
as well as the `#include`s.
- Change default `%_pm.h` make target to `techlibs/%_pm.h` and move it to the
top level Makefile.
- Update pmgen target to use `$(notdir $*)` (where `$*` is the part of the file
name that matched the '%' in the target) instead of `$(subst _pm.h,,$(notdir
$@))`.
`Const::size()` returns int, so change iterators that use it to `auto` instead of `size_t`.
For cases where size is being explicitly cast to `int`, use the wrapper that we already have instead: `Yosys::GetSize()`.
I'm working on build reproducibility of Fedora packages, and this patch fixes
an issue observed in test rebuilds: the timestamp was set to the actual time
of the build, making builds nonreproducible.
Other "Generated by" strings do not include a timestamp, so drop it here too.
Most of the word/coarse level cells have an assigned group and individual page.
The gate/fine level cells are all on one page.
Fix links to `cell_library.rst`.
New `help -dump-cells-json <file>` to dump cells list.
Add 'group' field to SimHelper class/struct with defaults to gate_other and word_other depending on source (simcells or simlib).
Add 'unary' group to unary operator cells for testing (based on internal cell library docs page).
- Drop `cell_code` and instead map code lookups to the `cell_help` dict.
- Add helper functions to struct for checking and getting the right cell.
- Add `CellType` for cell to `write_cell_rst` function declaration in
preparation for use in future.
- Iterate over `yosys_celltypes.cell_types` when exporting cell rst files,
reporting errors for any cells defined in `cell_types` but not
`cell_help_messages`.
Since `simcells.v` uses consistent formatting we can handle it specifically to help tidy up sphinx warnings about the truth tables, and instead chuck them in a code block which when printing to rst.
Also has the side effect that rst code blocks can be added manually with `//- ::` followed by a blank line.
Include Source file and line number in SimHelper struct, and use it for verilog code caption in rst dump.
Also reformat python string conversion to iterate over a list of fields instead of repeating code for each.
Allows for more expressive code when constructing help messages for cells.
Will also move extra logic in parsing help strings into the initial python parse instead of doing it in the C++ at export time.
The default mapping rules for division-like operations (div/divfloor/
mod/modfloor) invoke subtractions which can get mapped to carry chains
in FPGA flows. Optimizations across carry chains are weak, so in
practice this ends up too costly compared to implementing the division
purely in soft logic.
For this reason arrange for `techmap.v` ignoring division operations
under `-D NODIV`, and use this mode in `synth_quicklogic` to avoid carry
chains for divisions.
EMCU is a micro-processor based on ARM Cortex-M3 embedded in the
GW1NSR-4C chip used in the Tangnano4k board.
Signed-off-by: YRabbit <rabbit@yrabbit.cyou>