Added a mini type system for Dfg using DfgDataType to replace Dfg's use
of AstNodeDType. This is much more restricted and represents only the
types Dfg can handle in a canonical form. This will be needed when
adding more support for unpacked arrays and maybe unpacked structs one
day.
Also added an internal type checker for DfgGraphs which encodes all the
assumptions the code makes about type relationships in the graph. Run
this in a few places with --debug-check. Fix resulting fallout.
Large scale refactoring to simplify some of the more obtuse internals of
DFG. Remove multiple redundant internal APIs, simplify representation of
variables, fix potential unsoundness in circular decomposition. No
functional change intended.
The previous algorithm was designed to handle the general case where a
full control flow path predicate is required to select which value to
use when synthesizing control flow join point in an always block.
Here we add a better algorithm that tries to use the predicate of
the closest dominating branch if the branch paths dominate the joining
paths. This is almost universally true in synthesizable logic (RTLMeter
has no exceptions), however there are cases where this is not
applicable, for which we fall back on the previous generic algorithm.
Overall this significantly simplifies the synthesized Dfg graphs and
enables further optimization.
This patch adds DfgLogic, which is a vertex that represents a whole,
arbitrarily complex combinational AstAlways or AstAssignW in the
DfgGraph.
Implementing this requires computing the variables live at entry to the
AstAlways (variables read by the block), so there is a new
ControlFlowGraph data structure and a classical data-flow analysis based
live variable analysis to do that at the variable level (as opposed to
bit/element level).
The actual CFG construction and live variable analysis is best effort,
and might fail for currently unhandled constructs or data types. This
can be extended later.
V3DfgAstToDfg is changed to convert the Ast into an initial DfgGraph
containing only DfgLogic, DfgVertexSplice and DfgVertexVar vertices.
The DfgLogic are then subsequently synthesized into primitive operations
by the new V3DfgSynthesize pass, which is a combination of the old
V3DfgAstToDfg conversion and new code to handle AstAlways blocks with
complex flow control.
V3DfgSynthesize by default will synthesize roughly the same constructs
as V3DfgAstToDfg used to handle before, plus any logic that is part of a
combinational cycle within the DfgGraph. This enables breaking up these
cycles, for which there are extensions to V3DfgBreakCycles in this patch
as well. V3DfgSynthesize will then delete all non synthesized or non
synthesizable DfgLogic vertices and the rest of the Dfg pipeline is
identical, with minor changes to adjust for the changed representation.
Because with this change we can now eliminate many more UNOPTFLAT, DFG
has been disabled in all the tests that specifically target testing the
scheduling and reporting of circular combinational logic.
Store all flags in a DfgVertexVar relating to the underlying
AstVar/AstVarScope stored via AstNode::user1(). user2/user3/user4 are
then usable by DFG algorithms as needed.
When we had a `{a, b} = ...`, and the DFG conversion of `a = ...`
succeeded, but `b = ...` failed, we still used to include `a = ...` in
the DFG, which then caused a spurious multi-driver error for `a` on
a subsequent DFG pass, as the original `{a, b} = ...` was still present
in the Ast, but we also had the extra `a = ...` from converting out of
DFG on the previous pass.
In this patch we only convert assignments with a concatenation on the
LHS, if all target LValues can be converted into DFG.
This is the proper fix for #4231
SystemC variables are fairly special (they can only be assigned to/from,
but not otherwise participate in expressions), which complicates some
DFG code. These variables only ever appear as port on the top level
wrapper, so excluding them from DFG does not make us loose any
optimizations, but simplifies internals.
Previously DFG was limited to having a Sel, or an ArraySel potentially
under a Concat on the LHS of combinational assignments. Other forms or
combinations were not representable in the graph.
This adds support for arbitrary combinations of the above by
combining DfgSplicePacked and DfgSpliceArray vertices introduced in
#6176. In particular, Sel(ArraySel(VarRef,_),_) enables a lot more code
to be represented in DFG.
This is mostly a refactoring, but also enables handling some more
UNOPTFLAT, when the variable is only partially assigned in the cycle.
Previously the way partial assignments to variables were handled were
through the DfgVerexVar types themselves, which kept track of all
drivers. This has been replaced by DfgVertexSplice (which always drives
a DfgVeretexVar), and all DfgVertexVar now only have a single source,
either a DfgVertexSplice, if partially assigned, or an arbitrary
DfgVertex when updated as a whole.
DFG could remove forceable signals by replacing them with their
in-design driver. This is a bit of a pain to prevent, and ideally the
forcing transform should happen before DFG, but implementing it there is
a pain due to having to rewrite ports based on direction. This is an
attempted fix in DFG. More cases might remain.
In order to avoid unexpected breakage on multi-driven variables, we
resolve in DFG construction by using only the first driver encountered.
Also issues the MULTIDRIVEN error for these signals.
Apart from the representational changes below, this patch renames
AstNodeMath to AstNodeExpr, and AstCMath to AstCExpr.
Now every expression (i.e.: those AstNodes that represent a [possibly
void] value, with value being interpreted in a very general sense) has
AstNodeExpr as a super class. This necessitates the introduction of an
AstStmtExpr, which represents an expression in statement position, e.g :
'foo();' would be represented as AstStmtExpr(AstCCall(foo)). In exchange
we can get rid of isStatement() in AstNodeStmt, which now really always
represent a statement
Peak memory consumption and verilation speed are not measurably changed.
Partial step towards #3420
AstSel is a ternary node, but the 'widthp' is always constant and is
hence redundant, and 'lsbp' is very often constant. As AstSel is fairly
common, we special case as a DfgSel for the constant 'lsbp', and as
'DfgMux` for the non-constant 'lsbp'.
Use the same style, and reuse the bulk of astgen to generate DfgVertex
related code. In particular allow for easier definition of custom
DfgVertex sub-types that do not directly correspond to an AstNode
sub-type. Also introduces specific names for the fixed arity vertices.
No functional change intended.
Added DfgVertexVariadic to represent DFG vetices with a varying number
of source operands. Converted DfgVar to be a variadic vertex, with each
driver corresponding to a fixed range of bits in the packed variable.
This allows us to handle AstSel on the LHS of assignments. Also added
support for AstConcat on the LHS by selecting into the RHS as
appropriate.
This improves OpenTitan ST speed by ~13%
Added a new data-flow graph (DFG) based combinational logic optimizer.
The capabilities of this covers a combination of V3Const and V3Gate, but
is also more capable of transforming combinational logic into simplified
forms and more.
This entail adding a new internal representation, `DfgGraph`, and
appropriate `astToDfg` and `dfgToAst` conversion functions. The graph
represents some of the combinational equations (~continuous assignments)
in a module, and for the duration of the DFG passes, it takes over the
role of AstModule. A bulk of the Dfg vertices represent expressions.
These vertex classes, and the corresponding conversions to/from AST are
mostly auto-generated by astgen, together with a DfgVVisitor that can be
used for dynamic dispatch based on vertex (operation) types.
The resulting combinational logic graph (a `DfgGraph`) is then optimized
in various ways. Currently we perform common sub-expression elimination,
variable inlining, and some specific peephole optimizations, but there
is scope for more optimizations in the future using the same
representation. The optimizer is run directly before and after inlining.
The pre inline pass can operate on smaller graphs and hence converges
faster, but still has a chance of substantially reducing the size of the
logic on some designs, making inlining both faster and less memory
intensive. The post inline pass can then optimize across the inlined
module boundaries. No optimization is performed across a module
boundary.
For debugging purposes, each peephole optimization can be disabled
individually via the -fno-dfg-peepnole-<OPT> option, where <OPT> is one
of the optimizations listed in V3DfgPeephole.h, for example
-fno-dfg-peephole-remove-not-not.
The peephole patterns currently implemented were mostly picked based on
the design that inspired this work, and on that design the optimizations
yields ~30% single threaded speedup, and ~50% speedup on 4 threads. As
you can imagine not having to haul around redundant combinational
networks in the rest of the compilation pipeline also helps with memory
consumption, and up to 30% peak memory usage of Verilator was observed
on the same design.
Gains on other arbitrary designs are smaller (and can be improved by
analyzing those designs). For example OpenTitan gains between 1-15%
speedup depending on build type.
Geza Lore
Wilson Snyder
github action
Mariusz Glebocki
Adrien Le Masle
Jevin Sweval
HungMingWu