Automated synthesis of a limited-switch-dynamic logic (LSDL) circuits
Original Publication Date: 2008-Mar-28
Included in the Prior Art Database: 2008-Mar-28
The invention describes a synthesis flow for a new circuit family comprised of latched dynamic logic. The flow utilizes a set of novel functional decomposition techniques used in implementing our technology mapper.
Automated synthesis of a limited -switch-dynamic logic (LSDL) circuits
1. INTRODUCTION AND MOTIVATION
As CMOS scaling continues to diminish frequency gains, the
are faced with new challenges in producing high frequency low
circuits. Dynamic latched circuits offer new design style for
frequency microprocessor circuits while keeping their transistor
smaller that in those in CMOS.
This invention describes technology mapper for latch bounded
relies on a functional decomposition techniques that was
explore technology-specific constraints of the domino latched
logic (Figure 1). The decomposition relies on symbolic
functional choices that are generated implicitly using binary
diagrams (BDDs). The core computation relies on the symbolic
of recursive decomposition.
2. DECOMPOSITION FOR A CELL LIBRARY
A restriction on the height, width and the number of latched
trees within a cell imposes a constraint on the type of logic
components that this library can implement. To determine logical
feasibility of a cell to implement a given function we therefore
a procedure that tries to find a possible implementation of a
that is consistent with the constraints of a latched dynamic
applied procedure utilizes newly developed symbolic functional
Referring to Figure 2, to implement of a logic piece using a cell
with two dynamic trees we need a way of breaking a function into
sub-functions that are composed with a simple gate such as NAND.
is achieved through application of recursive bi-decomposition
form, where can be an arbitrary 2-input Boolean function (Figure
decomposition is not unique and varies depending on the selection
and subsets. This implies that depending on their particular
selection the height and the width constraints on the dynamic
and may or may not get satisfied. The problem is first solved
by first finding symbolic encoding of all feasible subsets, and
identifying those for which constraints on and are satisfiable.
2.2 Decomposition Choices
Our method uses notion of parameterization, that is applied
bi-decomposition of a function.
To formulate the problem we introduce an operator for the
quantification of variables, such as universal or existential
quantification. The operator uses a set of auxiliary variables
parameterize original symbolic statement of a problem, which are
to guide decisions on whether its corresponding variables are
abstracted or not during quantification. An assignment to the
induces a subset of quantified out variables. To encode the
the assignments to the variables on the quantification we use the
if-then-else operator. Using this operator variable encodes an