Skip to content
Projects
Groups
Snippets
Help
This project
Loading...
Sign in / Register
Toggle navigation
A
abc
Overview
Overview
Details
Activity
Cycle Analytics
Repository
Repository
Files
Commits
Branches
Tags
Contributors
Graph
Compare
Charts
Issues
0
Issues
0
List
Board
Labels
Milestones
Merge Requests
0
Merge Requests
0
CI / CD
CI / CD
Pipelines
Jobs
Schedules
Charts
Wiki
Wiki
Snippets
Snippets
Members
Members
Collapse sidebar
Close sidebar
Activity
Graph
Charts
Create a new issue
Jobs
Commits
Issue Boards
Open sidebar
lvzhengyang
abc
Commits
0e0f2e64
Commit
0e0f2e64
authored
Sep 30, 2015
by
Alan Mishchenko
Browse files
Options
Browse Files
Download
Email Patches
Plain Diff
Naive LUT packing algorithm (command &pack).
parent
10c31c65
Show whitespace changes
Inline
Side-by-side
Showing
4 changed files
with
311 additions
and
0 deletions
+311
-0
abclib.dsp
+4
-0
src/aig/gia/giaPack.c
+207
-0
src/aig/gia/module.make
+1
-0
src/base/abci/abc.c
+99
-0
No files found.
abclib.dsp
View file @
0e0f2e64
...
...
@@ -4251,6 +4251,10 @@ SOURCE=.\src\aig\gia\giaOf.c
# End Source File
# Begin Source File
SOURCE=.\src\aig\gia\giaPack.c
# End Source File
# Begin Source File
SOURCE=.\src\aig\gia\giaPat.c
# End Source File
# Begin Source File
...
...
src/aig/gia/giaPack.c
0 → 100644
View file @
0e0f2e64
/**CFile****************************************************************
FileName [giaPack.c]
SystemName [ABC: Logic synthesis and verification system.]
PackageName [Scalable AIG package.]
Synopsis [LUT packing.]
Author [Alan Mishchenko]
Affiliation [UC Berkeley]
Date [Ver. 1.0. Started - June 20, 2005.]
Revision [$Id: giaPack.c,v 1.00 2005/06/20 00:00:00 alanmi Exp $]
***********************************************************************/
#include "gia.h"
ABC_NAMESPACE_IMPL_START
////////////////////////////////////////////////////////////////////////
/// DECLARATIONS ///
////////////////////////////////////////////////////////////////////////
////////////////////////////////////////////////////////////////////////
/// FUNCTION DEFINITIONS ///
////////////////////////////////////////////////////////////////////////
/**Function*************************************************************
Synopsis [Collects LUT nodes in the increasing order of distance from COs.]
Description []
SideEffects []
SeeAlso []
***********************************************************************/
Vec_Int_t
*
Gia_ManLutCollect2
(
Gia_Man_t
*
p
)
{
Gia_Obj_t
*
pObj
;
Vec_Int_t
*
vOrder
;
int
i
,
k
,
Id
,
iFan
;
vOrder
=
Vec_IntAlloc
(
Gia_ManLutNum
(
p
)
);
Gia_ManIncrementTravId
(
p
);
Gia_ManForEachCoDriver
(
p
,
pObj
,
i
)
{
if
(
!
Gia_ObjIsAnd
(
pObj
)
)
continue
;
Id
=
Gia_ObjId
(
p
,
pObj
);
assert
(
Gia_ObjIsLut
(
p
,
Id
)
);
if
(
Gia_ObjIsTravIdCurrentId
(
p
,
Id
)
)
continue
;
Gia_ObjSetTravIdCurrentId
(
p
,
Id
);
Vec_IntPush
(
vOrder
,
Id
);
}
Vec_IntForEachEntry
(
vOrder
,
Id
,
i
)
{
Gia_LutForEachFanin
(
p
,
Id
,
iFan
,
k
)
{
if
(
!
Gia_ObjIsAnd
(
Gia_ManObj
(
p
,
iFan
))
)
continue
;
assert
(
Gia_ObjIsLut
(
p
,
iFan
)
);
if
(
Gia_ObjIsTravIdCurrentId
(
p
,
iFan
)
)
continue
;
Gia_ObjSetTravIdCurrentId
(
p
,
iFan
);
Vec_IntPush
(
vOrder
,
iFan
);
}
}
assert
(
Vec_IntCap
(
vOrder
)
==
16
||
Vec_IntSize
(
vOrder
)
==
Vec_IntCap
(
vOrder
)
);
Vec_IntReverseOrder
(
vOrder
);
return
vOrder
;
}
Vec_Int_t
*
Gia_ManLutCollect
(
Gia_Man_t
*
p
)
{
Vec_Int_t
*
vLuts
,
*
vDist
,
*
vOrder
;
int
i
,
k
,
Id
,
iFan
,
*
pPerm
;
// collect LUTs
vLuts
=
Vec_IntAlloc
(
Gia_ManAndNum
(
p
)
);
Gia_ManForEachLut
(
p
,
Id
)
Vec_IntPush
(
vLuts
,
Id
);
// propagate distance
vDist
=
Vec_IntStart
(
Gia_ManObjNum
(
p
)
);
Gia_ManForEachCoDriverId
(
p
,
Id
,
i
)
Vec_IntWriteEntry
(
vDist
,
Id
,
1
);
Vec_IntForEachEntryReverse
(
vLuts
,
Id
,
i
)
{
int
Dist
=
1
+
Vec_IntEntry
(
vDist
,
Id
);
Gia_LutForEachFanin
(
p
,
Id
,
iFan
,
k
)
Vec_IntUpdateEntry
(
vDist
,
iFan
,
Dist
);
}
// sort LUTs by distance
k
=
0
;
Vec_IntForEachEntry
(
vLuts
,
Id
,
i
)
Vec_IntWriteEntry
(
vDist
,
k
++
,
-
Vec_IntEntry
(
vDist
,
Id
)
);
Vec_IntShrink
(
vDist
,
k
);
pPerm
=
Abc_MergeSortCost
(
Vec_IntArray
(
vDist
),
Vec_IntSize
(
vDist
)
);
// collect
vOrder
=
Vec_IntAlloc
(
Vec_IntSize
(
vLuts
)
);
for
(
i
=
0
;
i
<
Vec_IntSize
(
vLuts
);
i
++
)
Vec_IntPush
(
vOrder
,
Vec_IntEntry
(
vLuts
,
pPerm
[
i
])
);
Vec_IntFree
(
vDist
);
Vec_IntFree
(
vLuts
);
ABC_FREE
(
pPerm
);
return
vOrder
;
}
/**Function*************************************************************
Synopsis [LUT packing algorithm.]
Description []
SideEffects []
SeeAlso []
***********************************************************************/
void
Gia_ManLutPacking
(
Gia_Man_t
*
p
,
int
nBlockSize
,
int
DelayRoute
,
int
DelayDir
,
int
fVerbose
)
{
int
Delays
[
32
],
Perm
[
32
];
Vec_Int_t
*
vPacking
,
*
vStarts
;
Vec_Int_t
*
vOrder
=
Gia_ManLutCollect
(
p
);
Vec_Int_t
*
vDelay
=
Vec_IntStart
(
Gia_ManObjNum
(
p
)
);
Vec_Int_t
*
vBlock
=
Vec_IntStart
(
Gia_ManObjNum
(
p
)
);
Vec_Int_t
*
vBSize
=
Vec_IntAlloc
(
2
*
Vec_IntSize
(
vOrder
)
/
nBlockSize
);
int
i
,
k
,
Id
,
iFan
,
nSize
,
iBlock
,
Delay
,
DelayMax
=
0
;
// create blocks
Vec_IntForEachEntry
(
vOrder
,
Id
,
i
)
{
nSize
=
Gia_ObjLutSize
(
p
,
Id
);
assert
(
nSize
<=
32
);
Gia_LutForEachFanin
(
p
,
Id
,
iFan
,
k
)
{
Delays
[
k
]
=
Vec_IntEntry
(
vDelay
,
iFan
);
Perm
[
k
]
=
iFan
;
}
Vec_IntSelectSortCost2Reverse
(
Perm
,
nSize
,
Delays
);
assert
(
nSize
<
2
||
Delays
[
0
]
>=
Delays
[
nSize
-
1
]
);
assert
(
Delays
[
0
]
>=
0
&&
Delays
[
nSize
-
1
]
>=
0
);
// check if we can reduce delay by adding it to the same bin as the latest one
iBlock
=
Vec_IntEntry
(
vBlock
,
Perm
[
0
]
);
if
(
Delays
[
0
]
>
0
&&
Delays
[
0
]
>
Delays
[
1
]
&&
Vec_IntEntry
(
vBSize
,
iBlock
)
<
nBlockSize
)
{
Delay
=
Delays
[
0
]
+
DelayDir
;
Vec_IntWriteEntry
(
vBlock
,
Id
,
iBlock
);
Vec_IntAddToEntry
(
vBSize
,
iBlock
,
1
);
}
else
// clean new block
{
Delay
=
Delays
[
0
]
+
DelayRoute
;
Vec_IntWriteEntry
(
vBlock
,
Id
,
Vec_IntSize
(
vBSize
)
);
Vec_IntPush
(
vBSize
,
1
);
}
// calculate delay
for
(
k
=
1
;
k
<
nSize
;
k
++
)
Delay
=
Abc_MaxInt
(
Delay
,
Delays
[
k
]
+
DelayRoute
);
Vec_IntWriteEntry
(
vDelay
,
Id
,
Delay
);
DelayMax
=
Abc_MaxInt
(
DelayMax
,
Delay
);
}
assert
(
Vec_IntSum
(
vBSize
)
==
Vec_IntSize
(
vOrder
)
);
// create packing info
vPacking
=
Vec_IntAlloc
(
Vec_IntSize
(
vBSize
)
+
Vec_IntSize
(
vOrder
)
+
1
);
Vec_IntPush
(
vPacking
,
Vec_IntSize
(
vBSize
)
);
// create starting places for each block
vStarts
=
Vec_IntAlloc
(
Vec_IntSize
(
vBSize
)
);
Vec_IntForEachEntry
(
vBSize
,
nSize
,
i
)
{
Vec_IntPush
(
vPacking
,
nSize
);
Vec_IntPush
(
vStarts
,
Vec_IntSize
(
vPacking
)
);
Vec_IntFillExtra
(
vPacking
,
Vec_IntSize
(
vPacking
)
+
nSize
,
-
1
);
}
assert
(
Vec_IntCap
(
vPacking
)
==
16
||
Vec_IntSize
(
vPacking
)
==
Vec_IntCap
(
vPacking
)
);
// collect LUTs from the block
Vec_IntForEachEntryReverse
(
vOrder
,
Id
,
i
)
{
int
Block
=
Vec_IntEntry
(
vBlock
,
Id
);
int
Start
=
Vec_IntEntry
(
vStarts
,
Block
);
assert
(
Vec_IntEntry
(
vPacking
,
Start
)
==
-
1
);
Vec_IntWriteEntry
(
vPacking
,
Start
,
Id
);
Vec_IntAddToEntry
(
vStarts
,
Block
,
1
);
}
assert
(
Vec_IntCountEntry
(
vPacking
,
-
1
)
==
0
);
// cleanup
Vec_IntFree
(
vOrder
);
Vec_IntFree
(
vDelay
);
Vec_IntFree
(
vBlock
);
Vec_IntFree
(
vBSize
);
Vec_IntFree
(
vStarts
);
Vec_IntFreeP
(
&
p
->
vPacking
);
p
->
vPacking
=
vPacking
;
printf
(
"Global delay = %d.
\n
"
,
DelayMax
);
}
////////////////////////////////////////////////////////////////////////
/// END OF FILE ///
////////////////////////////////////////////////////////////////////////
ABC_NAMESPACE_IMPL_END
src/aig/gia/module.make
View file @
0e0f2e64
...
...
@@ -47,6 +47,7 @@ SRC += src/aig/gia/giaAig.c \
src/aig/gia/giaMuxes.c
\
src/aig/gia/giaNf.c
\
src/aig/gia/giaOf.c
\
src/aig/gia/giaPack.c
\
src/aig/gia/giaPat.c
\
src/aig/gia/giaPf.c
\
src/aig/gia/giaQbf.c
\
...
...
src/base/abci/abc.c
View file @
0e0f2e64
...
...
@@ -414,6 +414,7 @@ static int Abc_CommandAbc9Lf ( Abc_Frame_t * pAbc, int argc, cha
static
int
Abc_CommandAbc9Mf
(
Abc_Frame_t
*
pAbc
,
int
argc
,
char
**
argv
);
static
int
Abc_CommandAbc9Nf
(
Abc_Frame_t
*
pAbc
,
int
argc
,
char
**
argv
);
static
int
Abc_CommandAbc9Of
(
Abc_Frame_t
*
pAbc
,
int
argc
,
char
**
argv
);
static
int
Abc_CommandAbc9Pack
(
Abc_Frame_t
*
pAbc
,
int
argc
,
char
**
argv
);
static
int
Abc_CommandAbc9Unmap
(
Abc_Frame_t
*
pAbc
,
int
argc
,
char
**
argv
);
static
int
Abc_CommandAbc9Struct
(
Abc_Frame_t
*
pAbc
,
int
argc
,
char
**
argv
);
static
int
Abc_CommandAbc9Trace
(
Abc_Frame_t
*
pAbc
,
int
argc
,
char
**
argv
);
...
...
@@ -1034,6 +1035,7 @@ void Abc_Init( Abc_Frame_t * pAbc )
Cmd_CommandAdd
(
pAbc
,
"ABC9"
,
"&mf"
,
Abc_CommandAbc9Mf
,
0
);
Cmd_CommandAdd
(
pAbc
,
"ABC9"
,
"&nf"
,
Abc_CommandAbc9Nf
,
0
);
Cmd_CommandAdd
(
pAbc
,
"ABC9"
,
"&of"
,
Abc_CommandAbc9Of
,
0
);
Cmd_CommandAdd
(
pAbc
,
"ABC9"
,
"&pack"
,
Abc_CommandAbc9Pack
,
0
);
Cmd_CommandAdd
(
pAbc
,
"ABC9"
,
"&unmap"
,
Abc_CommandAbc9Unmap
,
0
);
Cmd_CommandAdd
(
pAbc
,
"ABC9"
,
"&struct"
,
Abc_CommandAbc9Struct
,
0
);
Cmd_CommandAdd
(
pAbc
,
"ABC9"
,
"&trace"
,
Abc_CommandAbc9Trace
,
0
);
...
...
@@ -34257,6 +34259,103 @@ usage:
SeeAlso []
***********************************************************************/
int
Abc_CommandAbc9Pack
(
Abc_Frame_t
*
pAbc
,
int
argc
,
char
**
argv
)
{
extern
void
Gia_ManLutPacking
(
Gia_Man_t
*
p
,
int
nBlock
,
int
DelayRoute
,
int
DelayDir
,
int
fVerbose
);
int
c
,
nBlock
=
2
,
DelayRoute
=
10
,
DelayDir
=
2
,
fVerbose
=
0
;
Extra_UtilGetoptReset
();
while
(
(
c
=
Extra_UtilGetopt
(
argc
,
argv
,
"NRDvh"
)
)
!=
EOF
)
{
switch
(
c
)
{
case
'N'
:
if
(
globalUtilOptind
>=
argc
)
{
Abc_Print
(
-
1
,
"Command line switch
\"
-N
\"
should be followed by a positive integer.
\n
"
);
goto
usage
;
}
nBlock
=
atoi
(
argv
[
globalUtilOptind
]);
globalUtilOptind
++
;
if
(
nBlock
<
2
)
{
Abc_Print
(
-
1
,
"LUT block size (%d) should be more than 1.
\n
"
,
nBlock
);
goto
usage
;
}
break
;
case
'R'
:
if
(
globalUtilOptind
>=
argc
)
{
Abc_Print
(
-
1
,
"Command line switch
\"
-N
\"
should be followed by a positive integer.
\n
"
);
goto
usage
;
}
DelayRoute
=
atoi
(
argv
[
globalUtilOptind
]);
globalUtilOptind
++
;
if
(
DelayRoute
<=
0
)
{
Abc_Print
(
-
1
,
"Rounting delay (%d) should be more than 0.
\n
"
,
DelayRoute
);
goto
usage
;
}
break
;
case
'D'
:
if
(
globalUtilOptind
>=
argc
)
{
Abc_Print
(
-
1
,
"Command line switch
\"
-D
\"
should be followed by a positive integer.
\n
"
);
goto
usage
;
}
DelayDir
=
atoi
(
argv
[
globalUtilOptind
]);
globalUtilOptind
++
;
if
(
DelayDir
<=
0
)
{
Abc_Print
(
-
1
,
"Direct delay (%d) should be more than 0.
\n
"
,
DelayRoute
);
goto
usage
;
}
break
;
case
'v'
:
fVerbose
^=
1
;
break
;
case
'h'
:
default:
goto
usage
;
}
}
if
(
pAbc
->
pGia
==
NULL
)
{
Abc_Print
(
-
1
,
"Empty GIA network.
\n
"
);
return
1
;
}
if
(
!
Gia_ManHasMapping
(
pAbc
->
pGia
)
)
{
Abc_Print
(
-
1
,
"Current AIG has no mapping. Run
\"
&if
\"
.
\n
"
);
return
1
;
}
if
(
Gia_ManLutSizeMax
(
pAbc
->
pGia
)
>
6
)
Abc_Print
(
0
,
"Current AIG has mapping into %d-LUTs.
\n
"
,
Gia_ManLutSizeMax
(
pAbc
->
pGia
)
);
Gia_ManLutPacking
(
pAbc
->
pGia
,
nBlock
,
DelayRoute
,
DelayDir
,
fVerbose
);
return
0
;
usage:
Abc_Print
(
-
2
,
"usage: &pack [-NRD num] [-vh]
\n
"
);
Abc_Print
(
-
2
,
"
\t
performs packing for the LUT mapped network
\n
"
);
Abc_Print
(
-
2
,
"
\t
-N num : the number of LUTs in the block [default = %d]
\n
"
,
nBlock
);
Abc_Print
(
-
2
,
"
\t
-R num : the routable delay of a LUT [default = %d]
\n
"
,
DelayRoute
);
Abc_Print
(
-
2
,
"
\t
-D num : the direct (non-routable) delay of a LUT [default = %d]
\n
"
,
DelayDir
);
Abc_Print
(
-
2
,
"
\t
-v : toggles verbose output [default = %s]
\n
"
,
fVerbose
?
"yes"
:
"no"
);
Abc_Print
(
-
2
,
"
\t
-h : prints the command usage
\n
"
);
return
1
;
}
/**Function*************************************************************
Synopsis []
Description []
SideEffects []
SeeAlso []
***********************************************************************/
int
Abc_CommandAbc9Unmap
(
Abc_Frame_t
*
pAbc
,
int
argc
,
char
**
argv
)
{
extern
void
Gia_ManTestStruct
(
Gia_Man_t
*
p
);
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment