public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* ICE : HP g77 look unrolling
@ 1998-09-11 3:27 Peter Bienstman
1999-01-15 1:54 ` Jeffrey A Law
0 siblings, 1 reply; 3+ messages in thread
From: Peter Bienstman @ 1998-09-11 3:27 UTC (permalink / raw)
To: egcs-bugs
[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1331 bytes --]
Hello,
The bug I reported a while ago on a pre-1.1 snapshot is still present in the
1.1 release :
gcc -v
Reading specs from
/users/opt/pbienst/bin/gnu/lib/gcc-lib/hppa1.1-hp-hpux9.05/egcs-2.91.57/spec
s
gcc version egcs-2.91.57 19980901 (egcs-1.1 release)
When compiling the file zgelsx.f from the LAPACK distribution (see
attachment), I get an ICE when combining -O2 and -funroll-loops :
g77 -O2 -funroll-loops zgelsx.f
zgelsx.f: In subroutine `zgelsx':
zgelsx.f:352: internal error--unrecognizable insn:
(insn 2932 2936 2931 (set (reg:DF 1152)
(mem/s:DF (plus:SI (reg:SI 1098)
(const_int 16)))) -1 (insn_list 2874 (insn_list 2884
(insn_list 2888 (insn_list 2892(insn_list 2896 (insn_list 2906 (insn_list
2910 (insn_list 2912 (insn_list 2914 (insn_list 2930 (nil))))))))))) (nil))
/users/opt/pbienst/egcs-19980824/gcc/toplev.c:1360: Internal compiler error
in function fatal_insn
Compiling without -funroll-loops works fine, as does compiling with -O1.
Greetings,
----------------------------------
Peter Bienstman
INTEC / University of Gent
St.-Pietersnieuwstraat 41
B-9000 Gent - Belgium                       Â
Email: Peter.Bienstman@rug.ac.be
Tel : +32 9 264 3445
Fax : +32 9 264 3593
----------------------------------
[-- Attachment #2: zgelsx.f --]
[-- Type: text/plain, Size: 11313 bytes --]
SUBROUTINE ZGELSX( M, N, NRHS, A, LDA, B, LDB, JPVT, RCOND, RANK,
$ WORK, RWORK, INFO )
*
* -- LAPACK driver routine (version 2.0) --
* Univ. of Tennessee, Univ. of California Berkeley, NAG Ltd.,
* Courant Institute, Argonne National Lab, and Rice University
* September 30, 1994
*
* .. Scalar Arguments ..
INTEGER INFO, LDA, LDB, M, N, NRHS, RANK
DOUBLE PRECISION RCOND
* ..
* .. Array Arguments ..
INTEGER JPVT( * )
DOUBLE PRECISION RWORK( * )
COMPLEX*16 A( LDA, * ), B( LDB, * ), WORK( * )
* ..
*
* Purpose
* =======
*
* ZGELSX computes the minimum-norm solution to a complex linear least
* squares problem:
* minimize || A * X - B ||
* using a complete orthogonal factorization of A. A is an M-by-N
* matrix which may be rank-deficient.
*
* Several right hand side vectors b and solution vectors x can be
* handled in a single call; they are stored as the columns of the
* M-by-NRHS right hand side matrix B and the N-by-NRHS solution
* matrix X.
*
* The routine first computes a QR factorization with column pivoting:
* A * P = Q * [ R11 R12 ]
* [ 0 R22 ]
* with R11 defined as the largest leading submatrix whose estimated
* condition number is less than 1/RCOND. The order of R11, RANK,
* is the effective rank of A.
*
* Then, R22 is considered to be negligible, and R12 is annihilated
* by unitary transformations from the right, arriving at the
* complete orthogonal factorization:
* A * P = Q * [ T11 0 ] * Z
* [ 0 0 ]
* The minimum-norm solution is then
* X = P * Z' [ inv(T11)*Q1'*B ]
* [ 0 ]
* where Q1 consists of the first RANK columns of Q.
*
* Arguments
* =========
*
* M (input) INTEGER
* The number of rows of the matrix A. M >= 0.
*
* N (input) INTEGER
* The number of columns of the matrix A. N >= 0.
*
* NRHS (input) INTEGER
* The number of right hand sides, i.e., the number of
* columns of matrices B and X. NRHS >= 0.
*
* A (input/output) COMPLEX*16 array, dimension (LDA,N)
* On entry, the M-by-N matrix A.
* On exit, A has been overwritten by details of its
* complete orthogonal factorization.
*
* LDA (input) INTEGER
* The leading dimension of the array A. LDA >= max(1,M).
*
* B (input/output) COMPLEX*16 array, dimension (LDB,NRHS)
* On entry, the M-by-NRHS right hand side matrix B.
* On exit, the N-by-NRHS solution matrix X.
* If m >= n and RANK = n, the residual sum-of-squares for
* the solution in the i-th column is given by the sum of
* squares of elements N+1:M in that column.
*
* LDB (input) INTEGER
* The leading dimension of the array B. LDB >= max(1,M,N).
*
* JPVT (input/output) INTEGER array, dimension (N)
* On entry, if JPVT(i) .ne. 0, the i-th column of A is an
* initial column, otherwise it is a free column. Before
* the QR factorization of A, all initial columns are
* permuted to the leading positions; only the remaining
* free columns are moved as a result of column pivoting
* during the factorization.
* On exit, if JPVT(i) = k, then the i-th column of A*P
* was the k-th column of A.
*
* RCOND (input) DOUBLE PRECISION
* RCOND is used to determine the effective rank of A, which
* is defined as the order of the largest leading triangular
* submatrix R11 in the QR factorization with pivoting of A,
* whose estimated condition number < 1/RCOND.
*
* RANK (output) INTEGER
* The effective rank of A, i.e., the order of the submatrix
* R11. This is the same as the order of the submatrix T11
* in the complete orthogonal factorization of A.
*
* WORK (workspace) COMPLEX*16 array, dimension
* (min(M,N) + max( N, 2*min(M,N)+NRHS )),
*
* RWORK (workspace) DOUBLE PRECISION array, dimension (2*N)
*
* INFO (output) INTEGER
* = 0: successful exit
* < 0: if INFO = -i, the i-th argument had an illegal value
*
* =====================================================================
*
* .. Parameters ..
INTEGER IMAX, IMIN
PARAMETER ( IMAX = 1, IMIN = 2 )
DOUBLE PRECISION ZERO, ONE, DONE, NTDONE
PARAMETER ( ZERO = 0.0D+0, ONE = 1.0D+0, DONE = ZERO,
$ NTDONE = ONE )
COMPLEX*16 CZERO, CONE
PARAMETER ( CZERO = ( 0.0D+0, 0.0D+0 ),
$ CONE = ( 1.0D+0, 0.0D+0 ) )
* ..
* .. Local Scalars ..
INTEGER I, IASCL, IBSCL, ISMAX, ISMIN, J, K, MN
DOUBLE PRECISION ANRM, BIGNUM, BNRM, SMAX, SMAXPR, SMIN, SMINPR,
$ SMLNUM
COMPLEX*16 C1, C2, S1, S2, T1, T2
* ..
* .. External Subroutines ..
EXTERNAL DLABAD, XERBLA, ZGEQPF, ZLAIC1, ZLASCL, ZLASET,
$ ZLATZM, ZTRSM, ZTZRQF, ZUNM2R
* ..
* .. External Functions ..
DOUBLE PRECISION DLAMCH, ZLANGE
EXTERNAL DLAMCH, ZLANGE
* ..
* .. Intrinsic Functions ..
INTRINSIC ABS, DCONJG, MAX, MIN
* ..
* .. Executable Statements ..
*
MN = MIN( M, N )
ISMIN = MN + 1
ISMAX = 2*MN + 1
*
* Test the input arguments.
*
INFO = 0
IF( M.LT.0 ) THEN
INFO = -1
ELSE IF( N.LT.0 ) THEN
INFO = -2
ELSE IF( NRHS.LT.0 ) THEN
INFO = -3
ELSE IF( LDA.LT.MAX( 1, M ) ) THEN
INFO = -5
ELSE IF( LDB.LT.MAX( 1, M, N ) ) THEN
INFO = -7
END IF
*
IF( INFO.NE.0 ) THEN
CALL XERBLA( 'ZGELSX', -INFO )
RETURN
END IF
*
* Quick return if possible
*
IF( MIN( M, N, NRHS ).EQ.0 ) THEN
RANK = 0
RETURN
END IF
*
* Get machine parameters
*
SMLNUM = DLAMCH( 'S' ) / DLAMCH( 'P' )
BIGNUM = ONE / SMLNUM
CALL DLABAD( SMLNUM, BIGNUM )
*
* Scale A, B if max elements outside range [SMLNUM,BIGNUM]
*
ANRM = ZLANGE( 'M', M, N, A, LDA, RWORK )
IASCL = 0
IF( ANRM.GT.ZERO .AND. ANRM.LT.SMLNUM ) THEN
*
* Scale matrix norm up to SMLNUM
*
CALL ZLASCL( 'G', 0, 0, ANRM, SMLNUM, M, N, A, LDA, INFO )
IASCL = 1
ELSE IF( ANRM.GT.BIGNUM ) THEN
*
* Scale matrix norm down to BIGNUM
*
CALL ZLASCL( 'G', 0, 0, ANRM, BIGNUM, M, N, A, LDA, INFO )
IASCL = 2
ELSE IF( ANRM.EQ.ZERO ) THEN
*
* Matrix all zero. Return zero solution.
*
CALL ZLASET( 'F', MAX( M, N ), NRHS, CZERO, CZERO, B, LDB )
RANK = 0
GO TO 100
END IF
*
BNRM = ZLANGE( 'M', M, NRHS, B, LDB, RWORK )
IBSCL = 0
IF( BNRM.GT.ZERO .AND. BNRM.LT.SMLNUM ) THEN
*
* Scale matrix norm up to SMLNUM
*
CALL ZLASCL( 'G', 0, 0, BNRM, SMLNUM, M, NRHS, B, LDB, INFO )
IBSCL = 1
ELSE IF( BNRM.GT.BIGNUM ) THEN
*
* Scale matrix norm down to BIGNUM
*
CALL ZLASCL( 'G', 0, 0, BNRM, BIGNUM, M, NRHS, B, LDB, INFO )
IBSCL = 2
END IF
*
* Compute QR factorization with column pivoting of A:
* A * P = Q * R
*
CALL ZGEQPF( M, N, A, LDA, JPVT, WORK( 1 ), WORK( MN+1 ), RWORK,
$ INFO )
*
* complex workspace MN+N. Real workspace 2*N. Details of Householder
* rotations stored in WORK(1:MN).
*
* Determine RANK using incremental condition estimation
*
WORK( ISMIN ) = CONE
WORK( ISMAX ) = CONE
SMAX = ABS( A( 1, 1 ) )
SMIN = SMAX
IF( ABS( A( 1, 1 ) ).EQ.ZERO ) THEN
RANK = 0
CALL ZLASET( 'F', MAX( M, N ), NRHS, CZERO, CZERO, B, LDB )
GO TO 100
ELSE
RANK = 1
END IF
*
10 CONTINUE
IF( RANK.LT.MN ) THEN
I = RANK + 1
CALL ZLAIC1( IMIN, RANK, WORK( ISMIN ), SMIN, A( 1, I ),
$ A( I, I ), SMINPR, S1, C1 )
CALL ZLAIC1( IMAX, RANK, WORK( ISMAX ), SMAX, A( 1, I ),
$ A( I, I ), SMAXPR, S2, C2 )
*
IF( SMAXPR*RCOND.LE.SMINPR ) THEN
DO 20 I = 1, RANK
WORK( ISMIN+I-1 ) = S1*WORK( ISMIN+I-1 )
WORK( ISMAX+I-1 ) = S2*WORK( ISMAX+I-1 )
20 CONTINUE
WORK( ISMIN+RANK ) = C1
WORK( ISMAX+RANK ) = C2
SMIN = SMINPR
SMAX = SMAXPR
RANK = RANK + 1
GO TO 10
END IF
END IF
*
* Logically partition R = [ R11 R12 ]
* [ 0 R22 ]
* where R11 = R(1:RANK,1:RANK)
*
* [R11,R12] = [ T11, 0 ] * Y
*
IF( RANK.LT.N )
$ CALL ZTZRQF( RANK, N, A, LDA, WORK( MN+1 ), INFO )
*
* Details of Householder rotations stored in WORK(MN+1:2*MN)
*
* B(1:M,1:NRHS) := Q' * B(1:M,1:NRHS)
*
CALL ZUNM2R( 'Left', 'Conjugate transpose', M, NRHS, MN, A, LDA,
$ WORK( 1 ), B, LDB, WORK( 2*MN+1 ), INFO )
*
* workspace NRHS
*
* B(1:RANK,1:NRHS) := inv(T11) * B(1:RANK,1:NRHS)
*
CALL ZTRSM( 'Left', 'Upper', 'No transpose', 'Non-unit', RANK,
$ NRHS, CONE, A, LDA, B, LDB )
*
DO 40 I = RANK + 1, N
DO 30 J = 1, NRHS
B( I, J ) = CZERO
30 CONTINUE
40 CONTINUE
*
* B(1:N,1:NRHS) := Y' * B(1:N,1:NRHS)
*
IF( RANK.LT.N ) THEN
DO 50 I = 1, RANK
CALL ZLATZM( 'Left', N-RANK+1, NRHS, A( I, RANK+1 ), LDA,
$ DCONJG( WORK( MN+I ) ), B( I, 1 ),
$ B( RANK+1, 1 ), LDB, WORK( 2*MN+1 ) )
50 CONTINUE
END IF
*
* workspace NRHS
*
* B(1:N,1:NRHS) := P * B(1:N,1:NRHS)
*
DO 90 J = 1, NRHS
DO 60 I = 1, N
WORK( 2*MN+I ) = NTDONE
60 CONTINUE
DO 80 I = 1, N
IF( WORK( 2*MN+I ).EQ.NTDONE ) THEN
IF( JPVT( I ).NE.I ) THEN
K = I
T1 = B( K, J )
T2 = B( JPVT( K ), J )
70 CONTINUE
B( JPVT( K ), J ) = T1
WORK( 2*MN+K ) = DONE
T1 = T2
K = JPVT( K )
T2 = B( JPVT( K ), J )
IF( JPVT( K ).NE.I )
$ GO TO 70
B( I, J ) = T1
WORK( 2*MN+K ) = DONE
END IF
END IF
80 CONTINUE
90 CONTINUE
*
* Undo scaling
*
IF( IASCL.EQ.1 ) THEN
CALL ZLASCL( 'G', 0, 0, ANRM, SMLNUM, N, NRHS, B, LDB, INFO )
CALL ZLASCL( 'U', 0, 0, SMLNUM, ANRM, RANK, RANK, A, LDA,
$ INFO )
ELSE IF( IASCL.EQ.2 ) THEN
CALL ZLASCL( 'G', 0, 0, ANRM, BIGNUM, N, NRHS, B, LDB, INFO )
CALL ZLASCL( 'U', 0, 0, BIGNUM, ANRM, RANK, RANK, A, LDA,
$ INFO )
END IF
IF( IBSCL.EQ.1 ) THEN
CALL ZLASCL( 'G', 0, 0, SMLNUM, BNRM, N, NRHS, B, LDB, INFO )
ELSE IF( IBSCL.EQ.2 ) THEN
CALL ZLASCL( 'G', 0, 0, BIGNUM, BNRM, N, NRHS, B, LDB, INFO )
END IF
*
100 CONTINUE
*
RETURN
*
* End of ZGELSX
*
END
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: ICE : HP g77 look unrolling
1998-09-11 3:27 ICE : HP g77 look unrolling Peter Bienstman
@ 1999-01-15 1:54 ` Jeffrey A Law
0 siblings, 0 replies; 3+ messages in thread
From: Jeffrey A Law @ 1999-01-15 1:54 UTC (permalink / raw)
To: Peter Bienstman; +Cc: egcs-bugs
In message <001501bddd74$9da4a5c0$0454c19d@pcoptomc1.intec.rug.ac.be>you writ
e:
> This is a multi-part message in MIME format.
>
> ------=_NextPart_000_0016_01BDDD7C.FF690DC0
> Content-Type: text/plain;
> charset="iso-8859-1"
> Content-Transfer-Encoding: 8bit
>
> Hello,
>
> The bug I reported a while ago on a pre-1.1 snapshot is still present in th
> e
> 1.1 release :
>
> gcc -v
>
> Reading specs from
> /users/opt/pbienst/bin/gnu/lib/gcc-lib/hppa1.1-hp-hpux9.05/egcs-2.91.57/spe
> c
> s
> gcc version egcs-2.91.57 19980901 (egcs-1.1 release)
>
> When compiling the file zgelsx.f from the LAPACK distribution (see
> attachment), I get an ICE when combining -O2 and -funroll-loops :
>
>
> g77 -O2 -funroll-loops zgelsx.f
> zgelsx.f: In subroutine `zgelsx':
> zgelsx.f:352: internal error--unrecognizable insn:
> (insn 2932 2936 2931 (set (reg:DF 1152)
> (mem/s:DF (plus:SI (reg:SI 1098)
> (const_int 16)))) -1 (insn_list 2874 (insn_list 2884
> (insn_list 2888 (insn_list 2892(insn_list 2896 (insn_list 2906 (insn_list
> 2910 (insn_list 2912 (insn_list 2914 (insn_list 2930 (nil))))))))))) (nil))
> /users/opt/pbienst/egcs-19980824/gcc/toplev.c:1360: Internal compiler error
> in function fatal_insn
Thanks. I've checked in a fix for this bug.
jeff
^ permalink raw reply [flat|nested] 3+ messages in thread
* ICE : HP g77 look unrolling
@ 1998-08-31 4:40 Peter Bienstman
0 siblings, 0 replies; 3+ messages in thread
From: Peter Bienstman @ 1998-08-31 4:40 UTC (permalink / raw)
To: egcs-bugs
[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1314 bytes --]
Hello,
I work on the following platform :
Reading specs from
/users/opt/pbienst/lib/gcc-lib/hppa1.1-hp-hpux9.05/egcs-2.91.55/specs
gcc version egcs-2.91.55 19980824 (gcc2 ss-980609 experimental)
When compiling the file zgelsx.f from the LAPACK distribution (see
attachment), I get an ICE when combining -O2 and -funroll-loops :
opt0:/users/opt/pbienst/LAPACK/SRC$g77 -O2 -funroll-loops zgelsx.f
zgelsx.f: In subroutine `zgelsx':
zgelsx.f:352: internal error--unrecognizable insn:
(insn 2932 2936 2931 (set (reg:DF 1152)
(mem/s:DF (plus:SI (reg:SI 1098)
(const_int 16)))) -1 (insn_list 2874 (insn_list 2884
(insn_list 2888 (insn_list 2892 (insn_list 28
96 (insn_list 2906 (insn_list 2910 (insn_list 2912 (insn_list 2914
(insn_list 2930 (nil)))))))))))
(nil))
/users/opt/pbienst/egcs-19980824/gcc/toplev.c:1360: Internal compiler error
in function fatal_insn
Compiling without -funroll-loops works fine, as does compiling without -O2.
Greetings,
----------------------------------
Peter Bienstman
INTEC / University of Gent
St.-Pietersnieuwstraat 41
B-9000 Gent - Belgium                       Â
Email: Peter.Bienstman@rug.ac.be
Tel : +32 9 264 3445
Fax : +32 9 264 3593
----------------------------------
[-- Attachment #2: zgelsx.f --]
[-- Type: text/plain, Size: 11313 bytes --]
SUBROUTINE ZGELSX( M, N, NRHS, A, LDA, B, LDB, JPVT, RCOND, RANK,
$ WORK, RWORK, INFO )
*
* -- LAPACK driver routine (version 2.0) --
* Univ. of Tennessee, Univ. of California Berkeley, NAG Ltd.,
* Courant Institute, Argonne National Lab, and Rice University
* September 30, 1994
*
* .. Scalar Arguments ..
INTEGER INFO, LDA, LDB, M, N, NRHS, RANK
DOUBLE PRECISION RCOND
* ..
* .. Array Arguments ..
INTEGER JPVT( * )
DOUBLE PRECISION RWORK( * )
COMPLEX*16 A( LDA, * ), B( LDB, * ), WORK( * )
* ..
*
* Purpose
* =======
*
* ZGELSX computes the minimum-norm solution to a complex linear least
* squares problem:
* minimize || A * X - B ||
* using a complete orthogonal factorization of A. A is an M-by-N
* matrix which may be rank-deficient.
*
* Several right hand side vectors b and solution vectors x can be
* handled in a single call; they are stored as the columns of the
* M-by-NRHS right hand side matrix B and the N-by-NRHS solution
* matrix X.
*
* The routine first computes a QR factorization with column pivoting:
* A * P = Q * [ R11 R12 ]
* [ 0 R22 ]
* with R11 defined as the largest leading submatrix whose estimated
* condition number is less than 1/RCOND. The order of R11, RANK,
* is the effective rank of A.
*
* Then, R22 is considered to be negligible, and R12 is annihilated
* by unitary transformations from the right, arriving at the
* complete orthogonal factorization:
* A * P = Q * [ T11 0 ] * Z
* [ 0 0 ]
* The minimum-norm solution is then
* X = P * Z' [ inv(T11)*Q1'*B ]
* [ 0 ]
* where Q1 consists of the first RANK columns of Q.
*
* Arguments
* =========
*
* M (input) INTEGER
* The number of rows of the matrix A. M >= 0.
*
* N (input) INTEGER
* The number of columns of the matrix A. N >= 0.
*
* NRHS (input) INTEGER
* The number of right hand sides, i.e., the number of
* columns of matrices B and X. NRHS >= 0.
*
* A (input/output) COMPLEX*16 array, dimension (LDA,N)
* On entry, the M-by-N matrix A.
* On exit, A has been overwritten by details of its
* complete orthogonal factorization.
*
* LDA (input) INTEGER
* The leading dimension of the array A. LDA >= max(1,M).
*
* B (input/output) COMPLEX*16 array, dimension (LDB,NRHS)
* On entry, the M-by-NRHS right hand side matrix B.
* On exit, the N-by-NRHS solution matrix X.
* If m >= n and RANK = n, the residual sum-of-squares for
* the solution in the i-th column is given by the sum of
* squares of elements N+1:M in that column.
*
* LDB (input) INTEGER
* The leading dimension of the array B. LDB >= max(1,M,N).
*
* JPVT (input/output) INTEGER array, dimension (N)
* On entry, if JPVT(i) .ne. 0, the i-th column of A is an
* initial column, otherwise it is a free column. Before
* the QR factorization of A, all initial columns are
* permuted to the leading positions; only the remaining
* free columns are moved as a result of column pivoting
* during the factorization.
* On exit, if JPVT(i) = k, then the i-th column of A*P
* was the k-th column of A.
*
* RCOND (input) DOUBLE PRECISION
* RCOND is used to determine the effective rank of A, which
* is defined as the order of the largest leading triangular
* submatrix R11 in the QR factorization with pivoting of A,
* whose estimated condition number < 1/RCOND.
*
* RANK (output) INTEGER
* The effective rank of A, i.e., the order of the submatrix
* R11. This is the same as the order of the submatrix T11
* in the complete orthogonal factorization of A.
*
* WORK (workspace) COMPLEX*16 array, dimension
* (min(M,N) + max( N, 2*min(M,N)+NRHS )),
*
* RWORK (workspace) DOUBLE PRECISION array, dimension (2*N)
*
* INFO (output) INTEGER
* = 0: successful exit
* < 0: if INFO = -i, the i-th argument had an illegal value
*
* =====================================================================
*
* .. Parameters ..
INTEGER IMAX, IMIN
PARAMETER ( IMAX = 1, IMIN = 2 )
DOUBLE PRECISION ZERO, ONE, DONE, NTDONE
PARAMETER ( ZERO = 0.0D+0, ONE = 1.0D+0, DONE = ZERO,
$ NTDONE = ONE )
COMPLEX*16 CZERO, CONE
PARAMETER ( CZERO = ( 0.0D+0, 0.0D+0 ),
$ CONE = ( 1.0D+0, 0.0D+0 ) )
* ..
* .. Local Scalars ..
INTEGER I, IASCL, IBSCL, ISMAX, ISMIN, J, K, MN
DOUBLE PRECISION ANRM, BIGNUM, BNRM, SMAX, SMAXPR, SMIN, SMINPR,
$ SMLNUM
COMPLEX*16 C1, C2, S1, S2, T1, T2
* ..
* .. External Subroutines ..
EXTERNAL DLABAD, XERBLA, ZGEQPF, ZLAIC1, ZLASCL, ZLASET,
$ ZLATZM, ZTRSM, ZTZRQF, ZUNM2R
* ..
* .. External Functions ..
DOUBLE PRECISION DLAMCH, ZLANGE
EXTERNAL DLAMCH, ZLANGE
* ..
* .. Intrinsic Functions ..
INTRINSIC ABS, DCONJG, MAX, MIN
* ..
* .. Executable Statements ..
*
MN = MIN( M, N )
ISMIN = MN + 1
ISMAX = 2*MN + 1
*
* Test the input arguments.
*
INFO = 0
IF( M.LT.0 ) THEN
INFO = -1
ELSE IF( N.LT.0 ) THEN
INFO = -2
ELSE IF( NRHS.LT.0 ) THEN
INFO = -3
ELSE IF( LDA.LT.MAX( 1, M ) ) THEN
INFO = -5
ELSE IF( LDB.LT.MAX( 1, M, N ) ) THEN
INFO = -7
END IF
*
IF( INFO.NE.0 ) THEN
CALL XERBLA( 'ZGELSX', -INFO )
RETURN
END IF
*
* Quick return if possible
*
IF( MIN( M, N, NRHS ).EQ.0 ) THEN
RANK = 0
RETURN
END IF
*
* Get machine parameters
*
SMLNUM = DLAMCH( 'S' ) / DLAMCH( 'P' )
BIGNUM = ONE / SMLNUM
CALL DLABAD( SMLNUM, BIGNUM )
*
* Scale A, B if max elements outside range [SMLNUM,BIGNUM]
*
ANRM = ZLANGE( 'M', M, N, A, LDA, RWORK )
IASCL = 0
IF( ANRM.GT.ZERO .AND. ANRM.LT.SMLNUM ) THEN
*
* Scale matrix norm up to SMLNUM
*
CALL ZLASCL( 'G', 0, 0, ANRM, SMLNUM, M, N, A, LDA, INFO )
IASCL = 1
ELSE IF( ANRM.GT.BIGNUM ) THEN
*
* Scale matrix norm down to BIGNUM
*
CALL ZLASCL( 'G', 0, 0, ANRM, BIGNUM, M, N, A, LDA, INFO )
IASCL = 2
ELSE IF( ANRM.EQ.ZERO ) THEN
*
* Matrix all zero. Return zero solution.
*
CALL ZLASET( 'F', MAX( M, N ), NRHS, CZERO, CZERO, B, LDB )
RANK = 0
GO TO 100
END IF
*
BNRM = ZLANGE( 'M', M, NRHS, B, LDB, RWORK )
IBSCL = 0
IF( BNRM.GT.ZERO .AND. BNRM.LT.SMLNUM ) THEN
*
* Scale matrix norm up to SMLNUM
*
CALL ZLASCL( 'G', 0, 0, BNRM, SMLNUM, M, NRHS, B, LDB, INFO )
IBSCL = 1
ELSE IF( BNRM.GT.BIGNUM ) THEN
*
* Scale matrix norm down to BIGNUM
*
CALL ZLASCL( 'G', 0, 0, BNRM, BIGNUM, M, NRHS, B, LDB, INFO )
IBSCL = 2
END IF
*
* Compute QR factorization with column pivoting of A:
* A * P = Q * R
*
CALL ZGEQPF( M, N, A, LDA, JPVT, WORK( 1 ), WORK( MN+1 ), RWORK,
$ INFO )
*
* complex workspace MN+N. Real workspace 2*N. Details of Householder
* rotations stored in WORK(1:MN).
*
* Determine RANK using incremental condition estimation
*
WORK( ISMIN ) = CONE
WORK( ISMAX ) = CONE
SMAX = ABS( A( 1, 1 ) )
SMIN = SMAX
IF( ABS( A( 1, 1 ) ).EQ.ZERO ) THEN
RANK = 0
CALL ZLASET( 'F', MAX( M, N ), NRHS, CZERO, CZERO, B, LDB )
GO TO 100
ELSE
RANK = 1
END IF
*
10 CONTINUE
IF( RANK.LT.MN ) THEN
I = RANK + 1
CALL ZLAIC1( IMIN, RANK, WORK( ISMIN ), SMIN, A( 1, I ),
$ A( I, I ), SMINPR, S1, C1 )
CALL ZLAIC1( IMAX, RANK, WORK( ISMAX ), SMAX, A( 1, I ),
$ A( I, I ), SMAXPR, S2, C2 )
*
IF( SMAXPR*RCOND.LE.SMINPR ) THEN
DO 20 I = 1, RANK
WORK( ISMIN+I-1 ) = S1*WORK( ISMIN+I-1 )
WORK( ISMAX+I-1 ) = S2*WORK( ISMAX+I-1 )
20 CONTINUE
WORK( ISMIN+RANK ) = C1
WORK( ISMAX+RANK ) = C2
SMIN = SMINPR
SMAX = SMAXPR
RANK = RANK + 1
GO TO 10
END IF
END IF
*
* Logically partition R = [ R11 R12 ]
* [ 0 R22 ]
* where R11 = R(1:RANK,1:RANK)
*
* [R11,R12] = [ T11, 0 ] * Y
*
IF( RANK.LT.N )
$ CALL ZTZRQF( RANK, N, A, LDA, WORK( MN+1 ), INFO )
*
* Details of Householder rotations stored in WORK(MN+1:2*MN)
*
* B(1:M,1:NRHS) := Q' * B(1:M,1:NRHS)
*
CALL ZUNM2R( 'Left', 'Conjugate transpose', M, NRHS, MN, A, LDA,
$ WORK( 1 ), B, LDB, WORK( 2*MN+1 ), INFO )
*
* workspace NRHS
*
* B(1:RANK,1:NRHS) := inv(T11) * B(1:RANK,1:NRHS)
*
CALL ZTRSM( 'Left', 'Upper', 'No transpose', 'Non-unit', RANK,
$ NRHS, CONE, A, LDA, B, LDB )
*
DO 40 I = RANK + 1, N
DO 30 J = 1, NRHS
B( I, J ) = CZERO
30 CONTINUE
40 CONTINUE
*
* B(1:N,1:NRHS) := Y' * B(1:N,1:NRHS)
*
IF( RANK.LT.N ) THEN
DO 50 I = 1, RANK
CALL ZLATZM( 'Left', N-RANK+1, NRHS, A( I, RANK+1 ), LDA,
$ DCONJG( WORK( MN+I ) ), B( I, 1 ),
$ B( RANK+1, 1 ), LDB, WORK( 2*MN+1 ) )
50 CONTINUE
END IF
*
* workspace NRHS
*
* B(1:N,1:NRHS) := P * B(1:N,1:NRHS)
*
DO 90 J = 1, NRHS
DO 60 I = 1, N
WORK( 2*MN+I ) = NTDONE
60 CONTINUE
DO 80 I = 1, N
IF( WORK( 2*MN+I ).EQ.NTDONE ) THEN
IF( JPVT( I ).NE.I ) THEN
K = I
T1 = B( K, J )
T2 = B( JPVT( K ), J )
70 CONTINUE
B( JPVT( K ), J ) = T1
WORK( 2*MN+K ) = DONE
T1 = T2
K = JPVT( K )
T2 = B( JPVT( K ), J )
IF( JPVT( K ).NE.I )
$ GO TO 70
B( I, J ) = T1
WORK( 2*MN+K ) = DONE
END IF
END IF
80 CONTINUE
90 CONTINUE
*
* Undo scaling
*
IF( IASCL.EQ.1 ) THEN
CALL ZLASCL( 'G', 0, 0, ANRM, SMLNUM, N, NRHS, B, LDB, INFO )
CALL ZLASCL( 'U', 0, 0, SMLNUM, ANRM, RANK, RANK, A, LDA,
$ INFO )
ELSE IF( IASCL.EQ.2 ) THEN
CALL ZLASCL( 'G', 0, 0, ANRM, BIGNUM, N, NRHS, B, LDB, INFO )
CALL ZLASCL( 'U', 0, 0, BIGNUM, ANRM, RANK, RANK, A, LDA,
$ INFO )
END IF
IF( IBSCL.EQ.1 ) THEN
CALL ZLASCL( 'G', 0, 0, SMLNUM, BNRM, N, NRHS, B, LDB, INFO )
ELSE IF( IBSCL.EQ.2 ) THEN
CALL ZLASCL( 'G', 0, 0, BIGNUM, BNRM, N, NRHS, B, LDB, INFO )
END IF
*
100 CONTINUE
*
RETURN
*
* End of ZGELSX
*
END
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~1999-01-15 1:54 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
1998-09-11 3:27 ICE : HP g77 look unrolling Peter Bienstman
1999-01-15 1:54 ` Jeffrey A Law
-- strict thread matches above, loose matches on Subject: below --
1998-08-31 4:40 Peter Bienstman
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).