Академический Документы
Профессиональный Документы
Культура Документы
The Cadence® Virtuoso® Characterization Suite delivers the industry’s most comprehensive and
robust solution for the characterization and validation of your foundation IP—from standard cells,
I/Os, and complex multi-bit cells, to memories and mixed-signal blocks. Its patented Inside View
technology delivers better correlation to silicon by improving library throughput and ensuring
timing, power, noise, and statistical coverage of your IP.
Pro
ex
Std
mixed-signal blocks. The suite achieves • Speed, accuracy, capacity (hybrid partitioning)
View
C ust
both accuracy and high speed through • Modeling (CCS, ECSM, power, noise)
y
or
om
MM
Bl e
ock M
s
View approach—patented technology Variation and validation
for generating and optimizing Lib on • Accuracy, modeling (AOCV, SOCV, LVF)
r ar y Valid ati
• Completeness, integration (Tempus, Voltus,
characterization stimulus—coupled
ETS/EPS, PT/PT-PX), ease of use
with a parallel processing capability
that takes advantage of enterprise-
wide compute resources (Figure 1). Figure 1: The Virtuoso Characterization Suite with Inside View technology
The solution includes the Virtuoso®
Liberate™ Characterization Solution,
Virtuoso Liberate LV Library Validation Benefits • Support for larger macro blocks,
Solution, Virtuoso Variety Statistical such as memories and custom
• Ultra-fast cell library characterization
Characterization Solution, Virtuoso mixed-signal blocks, employing
for standard cells and complex I/Os
Liberate MX Memory Characterization a unique “hybrid partitioning”
Solution, and Virtuoso Liberate • Automatic pre-characterization technology to optimize runtime
AMS Mixed-Signal Characterization of each cell using the Inside View
• Ultra-fast throughput to complete
Solution. transistor-level circuit analysis
library validation overnight on
technology to learn all internal logic
The Virtuoso Characterization Suite a small number of multi-core
states and enable automatic vector
also integrates with the Spectre® computers
generation
Circuit Simulator, the industry- • Variation-aware timing model
standard SPICE simulator, delivering • Complex cell characterization for
creation accounting for process
even greater throughput than when low-power and/or high-speed
variations (systematic and random)
used with standalone third-party designs
for any set of correlated or
simulators (Figure 2). uncorrelated process parameters
Virtuoso Characterization Suite
MMSIM
Spectre
tom IC Design oso
Characteriz
Cus ati
so Pla rtu on
uo alog
Prototyp
in t Vi van
ced Model
in
An g Spectre APS Spectre RF Spectre XPS AMS Designer Ad g
fo
rt
Vi
rm
Ve
Accurate, high- Analog and RF-IC High-speed, high- Mixed-signal Com
n lls pl
r performance Ce
g
SIM Integration
ig noise analysis verification at all
capacity full-chip
ustom Routin
Pre
Pro
ify
ex
s
De
Std
simulatioin abstraction levels
cision Layout
cess Variation
I/O
Correct by variation analysis Inside
s
Construction View
C ust
M an
y
en
or
ll C
om
MM
uf
ac Spectre Infrastructure
m
le Bl e
Fu
tur
e Imp ock
s M
Pa n
Common infrastructure, advanced simulation database,
ras ig versatile front-end parser, robust device library n
i t i c - A a re D es Lib
ra ry Vali d ati o
w
Spectre Applications
Comprehensive coverage, integrated with Virtuoso, LEA, Liberate,
Allegro, and Encounter Timing System technologies
Figure 2: A complete solution for custom simulation and characterization, with integrated Spectre technology
Virtuoso Liberate Virtuoso Liberate LV Virtuoso Variety Virtuoso Liberate MX Virtuoso Liberate AMS
www.cadence.com 2
Virtuoso Characterization Suite
Model generation
www.cadence.com 3
Virtuoso Characterization Suite
www.cadence.com 4
Virtuoso Characterization Suite
www.cadence.com 5
Virtuoso Characterization Suite
Liberate MX Solution
The Cadence Virtuoso Liberate MX
Memory Characterization Solution
Monte Carlo reference (ns) extends the ultra-fast standard cell and
I/O library characterization capabilities
Figure 10: Process parameter variation
of the Liberate solution to cover large
are supported. Custom SSTA formats can validated to be highly accurate against memory cores. Macro blocks require
be easily supported using a Tcl API to the traditional Monte Carlo simulations additional pre-analysis steps to make fast
characterization database. (Figure 10). and accurate characterization feasible.
www.cadence.com 6
Virtuoso Characterization Suite
the generation of timing constraints and Static partitions are based solely on circuit terization capabilities of the Liberate
modeling of current source models for topology and are used for fully digital AMS solution to cover large mixed-signal
timing, power, and noise. sections of the macro. A typical static macros such as phase-locked loops (PLLs),
partition includes all of the channel- data converters (ADCs, DACs), SerDes,
Memories and large custom macros
connected logic components on the path high-speed transceivers, and I/Os. Macro
comprise a large percentage of silicon
between a primary input and a first-level blocks require additional pre-analysis
area on most chips and, consequently,
flip-flop, including all the required clock- steps to make fast and accurate character-
can often be major contributors to chip
generation circuitry. ization feasible.
performance and power consumption. To
validate a design’s electrical performance, Dynamic partitions are derived from a Leveraging a network of distributed CPUs
it is essential to have a highly accurate full-block transistor-level simulation using and utilizing the “hybrid partitioning”
electrical model for each macro equivalent a pattern sequence or a high-level truth- technology for optimizing character-
in accuracy of the electrical models used table description. From the simulation, an ization runtime, mixed-signal macros
for standard cells and I/Os. active circuit snapshot corresponding to can be characterized quickly and easily
the worst-case timing can be extracted. with the same accuracy and methods as
Using pre-packaged models provided by
Dynamic partitioning is useful for charac- standard cells, including the generation
an IP provider or memory compiler may
terizing timing arcs that contain analog of timing constraints and modeling of
not provide sufficient accuracy, especially
circuitry, such as the clock-to-output-data current source models for timing, power,
as the exact context of the macro is not
arcs in embedded memories. and noise.
known until it is placed on the chip. The
model may be too pessimistic, causing Once a macro has been partitioned into To validate a design’s electrical perfor-
overruns in schedule and increased usage sub-blocks comprising typically a few mance, it is essential to have a highly
of larger or leakier cells to close timing. It hundred transistors, and armed with accurate electrical model for each mixed-
is also common to operate a macro block the worst-case vector set required to signal macro equivalent in accuracy to the
at a lower voltage to save power. To get characterize each arc of each partition, electrical models used for standard cells
an accurate electrical model that reflects the Liberate MX solution submits each and I/Os. It is also common to operate a
the exact usage of the macro, a design- partition for characterization using the macro block at a lower voltage to save
specific and/or instance-specific macro integrated Spectre circuit simulator, power. To get an accurate electrical
block characterization is required. Spectre APS, or an external simulator such model that reflects the exact usage of the
as FineSim, HSpice, or Eldo. macro, a design-specific and/or instance-
Bi-modal view characterization
specific macro block characterization is
Library generation
The Liberate MX solution uses both a required.
full-block view and partitioned sub-block Leveraging the same characterization
Hybrid partitioning
views to characterize large macro blocks techniques and the same command
efficiently and accurately. The full-block options as the Liberate solution, arcs from The Liberate AMS solution uses a “hybrid
view is used to characterize power and to each Liberate MX partition are charac- partitioning” technology comprised of a
drive the creation of sub-block partitions. terized across a distributed network of full-block view and partitioned sub-block
Typically a FastSPICE simulator, such as computers utilizing all available CPUs. views to characterize large mixed-signal
Spectre XPS, UltraSim Full-Chip Simulator, The distribution can be controlled using a macro blocks efficiently and accurately.
FineSim Pro, or CustomSim, is used for job management system. Current source The full-block view is used to charac-
full-block circuit analysis, followed by models (CCS/ECSM) for timing, power, terize power and to drive the creation of
an accurate SPICE simulator such as the and noise can also be generated. sub-block partitions. Typically, a designer
Spectre circuit simulator, Spectre APS, will use a FastSPICE simulator such as
Timing constraints are calculated using
FineSim, HSpice, or Eldo for accurate Spectre XPS, UltraSim Full-Chip Simulator,
the same bi-sectional method as used for
characterization of sub-blocks. FineSim Pro, or CustomSim for full-block
standard cells or, even more efficiently, as
circuit analysis followed by an accurate
The Liberate MX solution optionally a difference in clock and data-path delay.
SPICE simulator such as the Spectre circuit
supports a single-block view character-
After characterization, all the charac- simulator for accurate characterization of
ization using only a FastSPICE simulator.
terized library data for each partition is sub-blocks.
Circuit partitioning assembled and compressed into a single
To partition each macro, the Liberate
output library representing the macro in
To partition each macro, the Liberate AMS solution first determines internal
Liberty format.
MX solution first determines internal measurement points via propagation of
measurement points via propagation of clocks signals and recognition of internal
Liberate AMS Solution storage elements (Figure 12). It uses two
clocks signals and recognition of internal
storage elements (Figure 11). It uses two The Cadence Virtuoso Liberate AMS distinct techniques for partitioning: static
distinct techniques for partitioning: static Mixed-Signal Solution extends the ultra- and dynamic.
and dynamic. fast standard cell and I/O library charac-
www.cadence.com 7
Virtuoso Characterization Suite
Autoprobing
TCL Boundary Modeling User Decks
using/reusing Virtuoso Analog Design
Spice DYNAMIC ANALYSIS
Environment (ADE/ADE-XL) testbenches
Netlist(s) and setup to drive the characterization
Vector Generation
flow. Thus, no major changes to the
implementation/verification flows are
required and analog circuit designers are
Timing Partitioning Power Characterization able to quickly move from circuit design
validation into library generation.
Block-Level Spectre XPS Liberate AMS Clients
Block-Level
Power/Leakage
Library generation
Spectre XPS
Leveraging the same characterization
Liberate AMS Clients
Partition-Level Timing techniques and the same command
.ldb
Spectre/Spectre APS options as the Liberate solution, arcs from
each Liberate AMS partition are charac-
Model Generation terized across a distributed network of
computers utilizing all available CPUs.
The distribution can be controlled using a
.LIB job management system. Current source
nldm, nlpm, ccs,
ccsn, ecsm ecsmn models (CCS/ECSM) for timing, power,
and noise can also be generated.
Figure 12: Mixed-signal macro characterization flow Timing constraints are calculated using
the same bi-sectional method as used for
Static partitions are based solely on circuit input-data-to-internal-clock and internal- standard cells or, even more efficiently, as
topology and are used for fully digital clock-to-output-data arcs in high-speed a difference in clock and data path delay.
sections of the macro. A typical static mixed-signal macros.
After characterization, all the charac-
partition includes all of the channel-
Once a macro has been partitioned into terized library data for each partition is
connected logic components on the path
sub-blocks comprising typically a few assembled and compressed into a single
between a primary input and a first-level
hundred transistors, and armed with output library representing the macro in
flip-flop, including all the required clock-
the worst-case vector set required to Liberty format.
generation circuitry.
characterize each arc of each partition,
Dynamic partitions are derived from a full- the Liberate AMS solution submits each
block transistor-level FastSPICE simulation partition for characterization using the
using a pattern sequence, or a high-level integrated Spectre circuit simulator,
truth-table description, or a testbench. Spectre APS, or an external simulator such
From the simulation, an active circuit as FineSim, HSpice, or Eldo.
snapshot corresponding to the worst-case
The Liberate AMS solution optionally
timing can be extracted. Dynamic parti-
supports a single-block view character-
tioning is useful for characterizing timing
ization using only a FastSPICE simulator.
arcs that contain analog circuitry, such as
Cadence Design Systems enables global electronic design innovation and plays an essential role in the
creation of today’s electronics. Customers use Cadence software, hardware, IP, and expertise to design
and verify today’s mobile, cloud and connectivity applications. www.cadence.com
© 2014 Cadence Design Systems, Inc. All rights reserved worldwide. Cadence, the Cadence logo, Encounter, Incisive, Spectre, and Virtuoso are
registered trademarks and Liberate and Tempus are trademarks of Cadence Design Systems, Inc. in the United States and other countries. All other
trademarks are properties of their respective holders. 2964 08/14 SA/DM/PDF