Clustering greatly helps the placer to identify and pack together
specific cells at the same site (e.g. LUT+FF), or cells that are chained through
dedicated interconnections (e.g. CARRY CHAINS)
Signed-off-by: Alessandro Comodi <acomodi@antmicro.com>
Previous pseudo pips were the same cost as regular pips, but this is
definitely too fast, and meant that the router was prefering them.
Signed-off-by: Keith Rothman <537074+litghost@users.noreply.github.com>
This prevents the general router from routing through sites, which is
not legal in FPGA interchange.
Signed-off-by: Keith Rothman <537074+litghost@users.noreply.github.com>
- Finishes implementation of SiteArch::nameOfPip and SiteArch::nameOfWire
- Adds "explain_bel_status", which should be an exhaustive diagnostic
of the status of a BEL placement.
Signed-off-by: Keith Rothman <537074+litghost@users.noreply.github.com>
If get_net_type was called before the driver was placed, it could return
the wrong value.
Signed-off-by: Keith Rothman <537074+litghost@users.noreply.github.com>
For now just implements some inspection capabilities, and the site
router (for now) avoids inverted paths.
Signed-off-by: Keith Rothman <537074+litghost@users.noreply.github.com>
Currently the lookahead is disabled by default because of the time to
compute and RAM usage. However it does appear to work reasonably well
in testing. Further effort is required to lower RAM usage after initial
computation, and explore trade-off for cheaper time to compute.
Signed-off-by: Keith Rothman <537074+litghost@users.noreply.github.com>
The new site router should be robust to most situations, and isn't
significantly slower with the use of caching.
Signed-off-by: Keith Rothman <537074+litghost@users.noreply.github.com>
This is important for distiguishing valid pseudo pips in the FPGA
interchange arch. This also avoids a double or triple lookup of
pip->net map.
Signed-off-by: Keith Rothman <537074+litghost@users.noreply.github.com>
"nextpnr.h" is no longer the god header. Important improvements:
- Functions in log.h can be used without including
BaseCtx/Arch/Context. This means that log_X functions can be called
without included "nextpnr.h"
- NPNR_ASSERT can be used without including "nextpnr.h" by including
"nextpnr_assertions.h". This allows NPNR_ASSERT to be used safely in
any header file.
- Types defined in "archdefs.h" are now available without including
BaseCtx/Arch/Context. This means that utility classes that will be
used inside of BaseCtx/Arch/Context can be defined safely in a
self-contained header.
Signed-off-by: Keith Rothman <537074+litghost@users.noreply.github.com>
Fixes:
- Only use map constant pins during routing, and not during placement.
- Unmapped cell ports have no BEL pins.
- Fix SiteRouter congestion not taking into account initial expansion.
- Fix psuedo-site pip output.
Signed-off-by: Keith Rothman <537074+litghost@users.noreply.github.com>
This replaces the arch-specific DelayInfo structure with new DelayPair
(min/max only) and DelayQuad (min/max for both rise and fall) structures
that form part of common code.
This further reduces the amount of arch-specific code; and also provides
useful data structures for timing analysis which will need to delay
with pairs/quads of delays as it is improved.
While there may be a small performance cost to arches that didn't
separate the rise/fall cases (arches that aren't currently separating
the min/max cases just need to be fixed...) in DelayInfo, my expectation
is that inlining will mean this doesn't make much difference.
Signed-off-by: gatecat <gatecat@ds0.me>
