|
NCTU-GR 2.0 |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Authors: Wen-Hao Liu, Ke-Ren
Dai, Wei-Chun Kao, Yih-Lang Li Download The 64bit Binary of NCTU-GR 2.0 (lasted updated: 2013/07/01) Download ISPD08 Global
Routing Benchmarks Download ISPD11, DAC12,
and ICCAD12
Placement Benchmarks and Placement Solutions Contact: NCTUGR {at} gmail
{dot} com |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Update History 2013/05/15:
NCTU-GR 2.0 is available for download. 2013/07/01:
Add a parameter that users can control the maximum runtime of the R&R
stage in NCTU-GR 2.0. (Note: when this utility is used, the routing result
may become nondeterministic.) |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Introduction NCTU-GR 2.0 is a unified placement
and routing tool that includes the following four utilities, in which the regular mode of
NCTU-GR 2.0 is selected to be the evaluation tool in DAC12 and ICCAD12
placement contests.
Users can use following commands to perform the
utilities of NCTU-GR 2.0.
Input and Output Files of NCTU-GR 2.0: 1. <ISPD’08 format benchmark> is the global
routing benchmark released by ISPD 2008 global routing contest. 2. <Circuit.aux> is
the placement benchmark released by ISPD 2011, DAC 2012 and ICCAD 2012
placement contests. 3. <Placement_solution.pl> is the placement
solution for the placement benchmark released by the placement contests. 4. [GR result] is the output file of regular mode and
fast mode, which is a multi-layer global routing result. 5. [PL result] is the output file of routability optimizer, which is a legalized placement
result with better routability. 6. <Design.rule> is a
parameter file used in the format converter to set the basic design rules in
the LEF/DEF files. 7. <Parameter.set> is
a parameter file used in the regular mode, fast mode and routability
optimizer to control the solution quality and runtime. Please read README
file in the downloaded package of NCTU-GR 2.0 so that the user can learn how
to tune the parameters to control the behavior of NCTU-GR 2.0. If the user unzipped the
downloaded package of NCTU-GR 2.0, the user can see the following files, in which
NCTUgr is the execution binary; please put NCTUgr, PORT.dat, POST9.dat and POWV9.dat in the same
directory. The directory [Parameter_Files] contains
several parameter setting files that are recommended setting for NCTU-GR 2.0
with different objectives. Of course, users can modify the parameters in
these setting files or create their own parameter setting files. =FILE_LIST= NCTUgr PORT9.dat POST9.dat POWV9.dat README [Parameter_Files]
└
RegularDefault.set
//Recommended setting to perform the regular mode └
RegularHighQuality.set //Regular mode with this setting
file would get better routing result but spend longer runtime └
RegularDacContest.set //DAC12 placement contest adopts
the regular mode with this setting to evaluate placement solutions └
RegularIccadContest.set //ICCAD12 placement contest adopts the
regular mode with this setting to evaluate placement solutions └
FastDefault.set
//Recommended setting to perform the fast mode └
PlaceForGR.set
//Recommended setting to perform routability
optimizer to optimize global routability └
PlaceForDR.set
//Recommended setting to perform routability
optimizer to optimize detailed routability [DesignRule_Files]
└
design22nm.rule └
design28nm.rule
//Converting academic placement benchmarks into LEF/DEF format under
28nm basic design rules. └
design45nm.rule └
design90nm.rule |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Experiments The following experiments were performed on a quad-core 3.0 GHz
Intel Xeon-based server with E5450 CPU and 32GB memory. Placement benchmarks
and the placement solutions used in the following experiments are downloaded
from ISPD11
placement contest website. Note that, the regular mode and the fast mode of NCTU-GR 2.0 both
consist of the 2D routing stage and layer assignment stage, so the columns of
“Routing time” and “LA time” in the following tables denote the runtime of
the 2D routing stage and layer assignment stage, respectively. Regular Mode The
routing results in Table 1 are obtained by the following command. ./NCTUgr
REGULAR <Circuit.aux>
<Placement_solution.pl>
<Parameter.set> [GR result] In
this experiment, <Parameter.set> is RegularDefault.set that is
included in the downloaded package of NCTU-GR 2.0. Because
the overflow reduction of sb2 and sb4 is almost stuck after iteration 20th,
we specially set Maze_Routing_Iteration=20 in RegularDefault.set for sb2 and sb4 to reduce runtime.
The
routing results in Table 2 are obtained by the following command. ./NCTUgr
REGULAR <Circuit.aux>
<Placement_solution.pl>
<Parameter.set> [GR result] In
this experiment, <Parameter.set> is RegularHighQuality.set
that is included in the downloaded package of NCTU-GR 2.0. Because
the overflow reduction of sb2 and sb4 is almost stuck after iteration 120th,
we specially set Maze_Routing_Iteration=120 in RegularHighQuality.set for sb2 and sb4 to reduce runtime.
By
comparing Table 1 and Table2, we can find that the routing results in Tabl2
have shorter wirelength and fewer vias than that in Table 1. However, the regular mode
spends longer routing time and LA time to get the routing results in Table 2
compared that in Table 1, because the parameter settings in RegularDefault.set and RegularHighQuality.set
are different. To learn more details about how to tune parameters, the user
can compare the content in RegularDefault.set and RegularHighQuality.set. Fast Mode The
routing results in Table 3 are obtained by the following command. ./NCTUgr
FAST <Circuit.aux>
<Placement_solution.pl>
<Parameter.set> [GR result] In
this experiment, <Parameter.set> is FastDefault.set that is
included in the downloaded package of NCTU-GR 2.0. Because
the overflow reduction of sb2 and sb4 is almost stuck after iteration 20th,
we specially set Maze_Routing_Iteration=20 in FastDefault.set for sb2 and sb4 to reduce runtime.
The
fast mode uses the techniques presented in the following papers to speed up
the 2D routing, so the runtime of 2D routing in Table 3 is about 2.4X~14.5X
faster than that in Table1. W.-H. Liu, Y.-L. Li and C.-K Koh, "A Fast Maze-Free
Routing Congestion Estimator With Hybrid Unilateral Monotonic Routing," in Proc. ICCAD, 2012. Routability Optimizer The
experiment in Table 4 shows the effectiveness of the routability
optimizer in NCTU-GR 2.0, in which we use the following command to optimize
the placement solutions in ISPD11 placement contest and then to get the
optimized placement solutions. ./NCTUgr
PLACE <Circuit.aux>
PlaceForGR.set <Parameter.set> [PL result] Then,
we use the following command to route the optimized placement solutions to show
their global routing results in Table 4. ./NCTUgr
REGULAR RegularDefault.set <PL result> <Parameter.set> [GR result]
By
comparing Table 1 and Table 4, we can find that the routability
optimizer can let the routing results of sb2 and sb4 become overflow-free,
and the routing results in Table 4 have 2.7%~12.2% improvement on “WL+VIA”
compared to that in Table 1. Format Convertor Because
ISPD11 and DAC12 placement benchmarks cannot be easily fed into commercial
routers, NCTU-GR 2.0 provides a format convertor to translate their format
into LEF/DEF files so that placement-related researches can rely on the
format convertor to evaluate the routability of
their placement solutions by commercial routers. The user can use the format
convertor by the following commend. ./NCTUgr
CONVERT <Circuit.aux>
<Placement_solution.pl>
<Design.rule> [LEF/DEF files] We
use an example to illustrate how to utilize the format convertor and then
evaluate the placement solution by commercial router WROUTE. For example,
sb1.aux and sb1.pl are a placement benchmark and the corresponding placement
solution in the ISPD11 format, respectively. The user can use the following
command to get three files: sb1ForWroute.aux, sb1ForWroute.lef, and
sb1ForWroute.def. The basic wire spacing, wire width, and via spacing rules
are specified in sb1ForWroute.lef file based on 28nm technology node. ./NCTUgr
CONVERT sb1.aux sb1.pl design28nm.rule sb1ForWroute Next,
if SOC encounter have been installed in the user’s server, the following
commands can be used to perform WROUTE on the LEF/DEF files, and finally get
a detailed routing result.
encounter -nowin
loadLefFile sb1ForWroute.lef
loadDefFile sb1ForWroute.def
wroute -grouteGrid
existing
wroute -grouteGrid
existing -mode postRouteRepair The
above commands are just for reference. If the user is familiar with WROUTE’s
commands, the user also can use other commands to control the behavior of
WROUTE. If
the user wants to adopt the proposed routability
optimizer to optimize the placement solutions before performing WROUTE, using
PlaceForDR.set to be the parameter setting file for
the routability optimizer is recommended. Because,
the routability optimizer with PlaceForDR.set
can better optimize the local congestion. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Related
Publications to the utilities in NCTU-GR 2.0:
Notice: l
If
you invoke the regular mode in your publication, you could cite papers [3]
and [4]. l
If
you invoke the negotiation-based layer assignment in your publication, you
could cite paper [5]. l
If you invoke the fast
mode in your publication,
you could cite paper [6]. l
If you invoke the routability optimize in your publication, you could cite paper [7]. l
If
you invoke the format converter in your publication, you could cite paper
[8]. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
License
Terms READ THIS LICENSE AGREEMENT CAREFULLY BEFORE
USING THIS PRODUCT. BY USING THIS PRODUCT YOU INDICATE YOUR ACCEPTANCE OF THE
TERMS OF THE FOLLOWING AGREEMENT. THESE TERMS APPLY TO YOU AND ANY SUBSEQUENT
LICENSEE OF THIS PRODUCT. License Agreement for NCTU-GR 2.0 Copyright (c) 2013 by Wen-Hao Liu and Yih-Lang Li
("Authors") Redistribution, with or without
modification, are permitted provided that the following conditions are met:
THIS FREE SOFTWARE IS PROVIDED BY THE AUTHOR
"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR ANY
CONTRIBUTOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, EFFECTS
OF UNAUTHORIZED OR MALICIOUS NETWORK ACCESS; PROCUREMENT OF SUBSTITUTE GOODS
OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
