Commit Graph

3980 Commits

Author SHA1 Message Date
Jannis Harder 57b4e16acd sim: Include $display output in JSON summary
This allows tools like SBY to capture the $display output independent
from anything else sim might log. Additionally it provides source and
hierarchy locations for everything printed.
2024-01-11 12:01:39 +01:00
N. Engelhardt 78541be4d8
Merge pull request #3971 from povik/equiv_simple-fixes
Fixes to `equiv_simple`
2023-12-18 16:31:02 +01:00
Jannis Harder cca12d9d9b
Merge pull request #4055 from povik/sim-hier-prints
sim: Print hierarchy for failed assertions
2023-12-11 16:55:36 +01:00
Jannis Harder 7b74caa5db peepopt: Fix padding for the peepopt_shiftmul_right pattern
The previous version could easily generate a large amount of padding
when the constant factor was significantly larger than the width of the
shift data input. This could lead to huge amounts of logic being
generated before then being optimized away at a huge performance and
memory cost.

Additionally and more critically, when the input width was not a
multiple of the constant factor, the input data was padded with 'x bits
to such a multiple before interspersing the 'x padding needed to align
the selectable windows to power-of-two offsets.

Such a final padding would not be correct for shifts besides $shiftx,
and the previous version did attempt to remove that final padding at the
end so that the native zero/sign/x-extension behavior of the shift cell
would be used, but since the last selectable window also got
power-of-two padding appended after the padding the code is trying to
remove got added, it did not actually fully remove it in some cases.

I changed the code to only add 'x padding between selectable windows,
leaving the last selectable window unpadded. This omits the need to add
final padding to a multiple of the constant factor in the first place.
In turn, that means the only 'x bits added are actually impossible to
select. As a side effect no padding is added when the constant factor is
equal to or larger than the width of the shift data input, also solving
the reported performance bug.

This fixes #4056
2023-12-06 18:35:44 +01:00
Martin Povišer 6581b5593c sim: Print hierarchy for failed assertions 2023-12-06 12:09:07 +01:00
Martin Povišer 16ea497d7c pmgen: Have a single make pattern
Remove duplicate %.pmg -> %_pm.h pattern. One of the duplicates overrode
the other, and in some conditions there were build races as to whether
the target directory for the generated header would exist. Instead have
a single rule which is properly generalized.
2023-12-05 18:30:13 +01:00
Martin Povišer d6566eb344 booth: Redo baseline architecture summation
Redo the summation logic: strive for some degree of balance on the
generated Wallace tree, emit an `$add` cell for the final summation.
2023-11-22 15:47:11 +01:00
Martin Povišer beb5cb55a5 booth: Expose `-lowpower` option 2023-11-22 15:29:59 +01:00
Martin Povišer 7005ea9411 booth: Revisit help 2023-11-22 15:29:59 +01:00
Martin Povišer 48b73be8c6 booth: Replace the default signed architecture
Generalize what was formerly the unsigned-only architecture to support
both signed and unsigned multiplication, use that as default, and set
aside the special low-power architecture that was formerly used for
signed multipliers.
2023-11-22 15:29:59 +01:00
Martin Povišer f50894d8bf booth: Drop extra decoder arguments 2023-11-22 15:29:54 +01:00
Martin Povišer 579f6bdc17 booth: Do not special-case bottom rows
Later on all the rows are cropped to the target size anyway, so there's
no harm in transitionally including extra top bits.
2023-11-22 15:12:15 +01:00
Martin Povišer da207cdce0 booth: Make less assumptions when aligning partial products 2023-11-22 15:12:15 +01:00
Martin Povišer 69e994ff75 booth: Clean unused FA index variable 2023-11-22 12:47:09 +01:00
Martin Povišer d8408b2350 booth: Move up signed quadrant 1 logic 2023-11-22 12:46:15 +01:00
Martin Povišer 8d33cc2fb6 booth: Refactor signed CPA 2023-11-22 12:46:15 +01:00
Martin Povišer 00e899f98d booth: Refactor signed multiplier full adders emission 2023-11-22 12:46:15 +01:00
N. Engelhardt 63cec22a0c
Merge pull request #3883 from phsauter/peepopt-shiftadd
peepopt: Add `shiftadd` pattern
2023-11-07 10:42:15 +01:00
phsauter 3618294bac peepopt: Add assert of consistent `shiftadd` data 2023-11-06 16:35:00 +01:00
Philippe Sauter b6df900bcc peepopt: Describe `shiftadd` rule in help message 2023-11-06 14:01:37 +01:00
phsauter 9ca57d9f13 peepopt: fix and refactor `shiftadd`
- moved all selection and filtering logic to the match block
- applied less-verbose code suggestions
- removed constraint on number of bits in shift-amount
- added check for possible wrap-arround in the operation
2023-11-06 14:01:37 +01:00
Philippe Sauter 72c6a01e67 peepopt: Add initial `shiftadd` pattern 2023-11-06 14:01:37 +01:00
N. Engelhardt 080da693d1 memory_libmap: update search order for attributes 2023-10-24 13:55:45 +02:00
N. Engelhardt 1b6d1e9419 memory_libmap: look for ram_style attributes on surrounding signals 2023-10-19 19:23:35 +02:00
Lofty d21c464ae4
Merge pull request #3946 from rmlarsen/toposort
Speed up TopoSort by 2.7-3.3x.
2023-10-17 13:00:18 +01:00
Claire Xen a4951a3a97
Merge pull request #3986 from povik/sim-ui-fixes
Slightly improve `sim` UI
2023-10-16 16:54:05 +02:00
N. Engelhardt a2f59cf911
Merge pull request #3990 from zeldin/deterministic_scc 2023-10-16 16:51:54 +02:00
Martin Povišer d6d1cc705e pmgen: Fix sample syntax 2023-10-16 14:19:15 +02:00
Martin Povišer 660be4a31e peepopt: Describe rules in help message 2023-10-16 14:19:15 +02:00
Martin Povišer 5c0c8251c3 peepopt: Remove broken `-generate` option 2023-10-16 14:19:10 +02:00
Martin Povišer aa9b86aeec peepopt: Add left-shift 'shiftmul' variant
Add a separate shiftmul pattern to match on left shifts which implement
demuxing. This mirrors the right shift pattern matcher but is probably
best kept separate instead of merging the two into a single matcher.
In any case the diff of the two matchers should be easily readable.
2023-10-16 13:52:38 +02:00
Martin Povišer 038a5e1ed4 peepopt: Support shift amounts zero-padded from below
The `opt_expr` pass running before `peepopt` can interfere with the
detection of a shiftmul pattern due to some of the bottom bits of the
shift amount being replaced with constant zero. Extend the detection to
cover those situations as well.
2023-10-16 13:52:06 +02:00
Martin Povišer dd1a8ae49a peepopt: Try to use original wires 2023-10-16 13:52:06 +02:00
Martin Povišer bd8a81a907 peepopt: Clean up 'shiftmul' a bit
No functional change intended.
2023-10-16 13:52:06 +02:00
Martin Povišer a0c3be3aae peepopt: Drop unused 'initbits' code
Drop code that was once used by the 'dffmux' pattern but now is unused
after that pattern has been obsoleted by the 'opt_dff' pass.
2023-10-16 13:52:06 +02:00
N. Engelhardt 3e22791810
Merge pull request #3975 from rmlarsen/optmerge 2023-10-09 17:05:19 +02:00
Marcus Comstedt 0ca39e233b scc: Use hashlib instead of STL for deterministic behaviour 2023-10-07 10:43:00 +02:00
Rasmus Munk Larsen 0a37c2a301 Fix translation bug: The old code really checks for the presense of a node, not an edge in glift and flatten.
Add back statement that inserts nodes in order in opt_expr.cc.
2023-10-05 17:01:42 -07:00
Martin Povišer c3fd88624a sim: Bail on processes
Instead of silently missimulating, error out when there are processes
found in the simulation hierarchy.
2023-10-05 19:25:17 +02:00
Martin Povišer a782b15aae sim: s/instanced/instantiated/ 2023-10-05 19:25:17 +02:00
Martin Povišer 6ac43e49bc sim: Change clocked read port suggestion to `memory_nordff`
`memory_nordff` has the advantage that it can be called just ahead of
the simulation step no matter whether the clocked read port has been
inferred or was explicitly instantiated in a flow.
2023-10-05 19:25:17 +02:00
Martin Povišer 0434f9d3d1 booth: Fix vacancy check when summing down result
In commit fedd12261 ("booth: Move away from explicit `Wire` pointers")
a bug was introduced when checking for vacant slots in arrays holding
some intermediate results. Non-wire SigBit values were taken to imply
a vacant slot, but actually a constant one can make its way into those
results, if the multiplier cell configuration is just right. Fix the
vacancy check to address the bug.
2023-10-04 23:21:40 +02:00
Rasmus Munk Larsen 57a2b4b0cd Explicitly use uint64_t as the type of fingerprint to avoid type mismatch with some compilers. 2023-10-03 15:02:02 -07:00
Rasmus Munk Larsen 8e0308b5e7 Revert changes to celltypes.h. Use dict instead of std::unordered_map and most hash function for uint64_t to hashlib.h to support this. 2023-10-03 14:25:59 -07:00
Martin Povišer 1f1f43edd9 equiv_simple: Fix seed handling in non-short mode 2023-10-03 13:05:42 +02:00
Martin Povišer dbf11da50a equiv_simple: Do not special-case flip-flop types in cone expansion
If there's an asynchronous flip-flop type, it will be caught by not
having a synchronous SAT model later on. Otherwise we can support all
flip-flops.
2023-10-03 13:05:42 +02:00
Martin Povišer 26644ea779 equiv_simple: Drop hollow conditional
All the listed flip-flop types would be known cells, so the extra part
of the conditional is without effect.
2023-10-03 13:05:42 +02:00
Rasmus Munk Larsen bce984fa60 Speed up OptMergePass by 1.7x.
The main speedup comes from swithing from using a SHA1 hash to std::hash<std::string>. There is no need to use an expensive cryptographic hash for fingerprinting in this context.
2023-10-02 15:57:18 -07:00
Rasmus Munk Larsen 1bbc12f389 Revert changes to techmap.cc. 2023-10-02 17:32:56 +01:00
Rasmus Munk Larsen 67f1700486 Revert formatting changes. 2023-10-02 17:32:56 +01:00