Data Types
interface	fft3d

type	fft_scratch_pool_type

type	fft_scratch_sizes

type	fft_scratch_type

Functions/Subroutines
subroutine, public	init_fft (fftlib, alltoall, fftsg_sizes, pool_limit, wisdom_file, plan_style)
	...

subroutine, public	finalize_fft (para_env, wisdom_file)
	does whatever is needed to finalize the current fft setup

subroutine, public	fft_radix_operations (radix_in, radix_out, operation)
	Determine the allowed lengths of FFT's '''.

subroutine, public	fft_fw1d (n, m, trans, zin, zout, scale, stat)
	Performs m 1-D forward FFT-s of size n.

subroutine, public	x_to_yz (sb, group, my_pos, p2p, yzp, nray, bo, tb, fft_scratch)
	...

subroutine, public	yz_to_x (tb, group, my_pos, p2p, yzp, nray, bo, sb, fft_scratch)
	...

subroutine, public	yz_to_xz (sb, group, dims, my_pos, p2p, yzp, nray, bo, tb, fft_scratch)
	...

subroutine, public	xz_to_yz (sb, group, dims, my_pos, p2p, yzp, nray, bo, tb, fft_scratch)
	...

subroutine, public	cube_transpose_1 (cin, boin, boout, sout, fft_scratch)
	...

subroutine, public	cube_transpose_2 (cin, boin, boout, sout, fft_scratch)
	...

subroutine, public	get_fft_scratch (fft_scratch, tf_type, n, fft_sizes)
	...

subroutine, public	release_fft_scratch (fft_scratch)
	...

Variables
integer, save	init_fft_pool = 0

integer, save	tick_fft_pool = 0

integer, save	fft_pool_scratch_limit = 15

type(fft_scratch_pool_type), pointer, save	fft_scratch_first

integer, parameter, public	fwfft = +1

integer, parameter, public	bwfft = -1

integer, parameter, public	fft_radix_closest = 493

integer, parameter, public	fft_radix_next = 494

integer, parameter, public	fft_radix_next_odd = 497

integer, save, public	fft_type = 0

integer, save, public	fft_plan_style = 1

Detailed Description

History: JGH (30-Nov-2000): ESSL FFT Library added JGH (05-Jan-2001): Added SGI library FFT JGH (14-Jan-2001): Added parallel 3d FFT JGH (10-Feb-2006): New interface type JGH (31-Mar-2008): Remove local allocates and reshapes (performance) Possible problems can be related with setting arrays not to zero Some interfaces could be further simplified by avoiding an initial copy. However, this assumes contiguous arrays IAB (15-Oct-2008): Moved mp_cart_sub calls out of cube_tranpose_* and into fft_scratch type, reducing number of calls dramatically IAB (05-Dec-2008): Moved all other non-essential MPI calls into scratch type IAB (09-Jan-2009): Added fft_plan_type to store FFT data, including cached FFTW plans IAB (13-Feb-2009): Extended plan caching to serial 3D FFT (fft3d_s) IAB (09-Oct-2009): Added OpenMP directives to parallel 3D FFT (c) The Numerical Algorithms Group (NAG) Ltd, 2008-2009 on behalf of the HECToR project

Author: JGH

Function/Subroutine Documentation

◆ init_fft()

subroutine, public fft_tools::init_fft	(	character(len=*), intent(in)	fftlib,
		logical, intent(in)	alltoall,
		logical, intent(in)	fftsg_sizes,
		integer, intent(in)	pool_limit,
		character(len=*), intent(in)	wisdom_file,
		integer, intent(in)	plan_style
	)

...

Parameters

fftlib	...
alltoall	...
fftsg_sizes	...
pool_limit	...
wisdom_file	...
plan_style	...

Author: JGH

Definition at line 182 of file fft_tools.F.

Here is the call graph for this function:

Here is the caller graph for this function:

◆ finalize_fft()

subroutine, public fft_tools::finalize_fft	(	class(mp_comm_type)	para_env,
		character(len=*), intent(in)	wisdom_file
	)

does whatever is needed to finalize the current fft setup

Parameters

para_env	...
wisdom_file	...

History: 10.2007 created [Joost VandeVondele]

Definition at line 214 of file fft_tools.F.

Here is the call graph for this function:

Here is the caller graph for this function:

◆ fft_radix_operations()

subroutine, public fft_tools::fft_radix_operations	(	integer, intent(in)	radix_in,
		integer, intent(out)	radix_out,
		integer, intent(in)	operation
	)

Determine the allowed lengths of FFT's '''.

Parameters

radix_in	...
radix_out	...
operation	...

History: new library structure (JGH)

Author: Ari Seitsonen

Definition at line 237 of file fft_tools.F.

Here is the call graph for this function:

Here is the caller graph for this function:

◆ fft_fw1d()

subroutine, public fft_tools::fft_fw1d	(	integer, intent(in)	n,
		integer, intent(in)	m,
		logical, intent(in)	trans,
		complex(kind=dp), dimension(*), intent(inout)	zin,
		complex(kind=dp), dimension(*), intent(inout)	zout,
		real(kind=dp), intent(in)	scale,
		integer, intent(out)	stat
	)

Performs m 1-D forward FFT-s of size n.

Parameters

n	size of the FFT
m	number of FFT-s
trans	shape of input and output arrays: [n x m] if it is FALSE, [m x n] if it is TRUE
zin	input array
zout	output array
scale	scaling factor
stat	status of the operation, non-zero code indicates an error

Definition at line 329 of file fft_tools.F.

Here is the call graph for this function:

Here is the caller graph for this function:

◆ x_to_yz()

subroutine, public fft_tools::x_to_yz	(	complex(kind=dp), dimension(:, :), intent(in), contiguous	sb,
		class(mp_comm_type), intent(in)	group,
		integer, intent(in)	my_pos,
		integer, dimension(0:), intent(in), contiguous	p2p,
		integer, dimension(:, :, 0:), intent(in), contiguous	yzp,
		integer, dimension(0:), intent(in), contiguous	nray,
		integer, dimension(:, :, 0:), intent(in), contiguous	bo,
		complex(kind=dp), dimension(:, :, :), intent(inout), contiguous	tb,
		type(fft_scratch_type), intent(in)	fft_scratch
	)

...

Parameters

sb	...
group	...
my_pos	...
p2p	...
yzp	...
nray	...
bo	...
tb	...
fft_scratch	...

History

Feb. 2006 : single precision all_to_all

Author: JGH (14-Jan-2001)

Definition at line 1370 of file fft_tools.F.

Here is the caller graph for this function:

◆ yz_to_x()

subroutine, public fft_tools::yz_to_x	(	complex(kind=dp), dimension(:, :, :), intent(in), contiguous	tb,
		class(mp_comm_type), intent(in)	group,
		integer, intent(in)	my_pos,
		integer, dimension(0:), intent(in), contiguous	p2p,
		integer, dimension(:, :, 0:), intent(in), contiguous	yzp,
		integer, dimension(0:), intent(in), contiguous	nray,
		integer, dimension(:, :, 0:), intent(in)	bo,
		complex(kind=dp), dimension(:, :), intent(inout), contiguous	sb,
		type(fft_scratch_type), intent(in)	fft_scratch
	)

...

Parameters

tb	...
group	...
my_pos	...
p2p	...
yzp	...
nray	...
bo	...
sb	...
fft_scratch	...

History

Feb. 2006 : single precision all_to_all

Author: JGH (14-Jan-2001)

Definition at line 1486 of file fft_tools.F.

Here is the caller graph for this function:

◆ yz_to_xz()

subroutine, public fft_tools::yz_to_xz	(	complex(kind=dp), dimension(:, :), intent(in), contiguous	sb,
		class(mp_comm_type), intent(in)	group,
		integer, dimension(2), intent(in)	dims,
		integer, intent(in)	my_pos,
		integer, dimension(0:), intent(in), contiguous	p2p,
		integer, dimension(:, :, 0:), intent(in), contiguous	yzp,
		integer, dimension(0:), intent(in), contiguous	nray,
		integer, dimension(:, :, 0:), intent(in), contiguous	bo,
		complex(kind=dp), dimension(:, :), intent(inout), contiguous	tb,
		type(fft_scratch_type), intent(inout)	fft_scratch
	)

...

Parameters

sb	...
group	...
dims	...
my_pos	...
p2p	...
yzp	...
nray	...
bo	...
tb	...
fft_scratch	...

History

Feb. 2006 : single precision all_to_all

Author: JGH (18-Jan-2001)

Definition at line 1603 of file fft_tools.F.

Here is the caller graph for this function:

◆ xz_to_yz()

subroutine, public fft_tools::xz_to_yz	(	complex(kind=dp), dimension(:, :), intent(in), contiguous	sb,
		class(mp_comm_type), intent(in)	group,
		integer, dimension(2), intent(in)	dims,
		integer, intent(in)	my_pos,
		integer, dimension(0:), intent(in), contiguous	p2p,
		integer, dimension(:, :, 0:), intent(in), contiguous	yzp,
		integer, dimension(0:), intent(in), contiguous	nray,
		integer, dimension(:, :, 0:), intent(in), contiguous	bo,
		complex(kind=dp), dimension(:, :), intent(inout), contiguous	tb,
		type(fft_scratch_type), intent(inout)	fft_scratch
	)

...

Parameters

sb	...
group	...
dims	...
my_pos	...
p2p	...
yzp	...
nray	...
bo	...
tb	...
fft_scratch	...

History

Feb. 2006 : single precision all_to_all

Author: JGH (19-Jan-2001)

Definition at line 1813 of file fft_tools.F.

Here is the caller graph for this function:

◆ cube_transpose_1()

subroutine, public fft_tools::cube_transpose_1	(	complex(kind=dp), dimension(:, :), intent(in), contiguous	cin,
		integer, dimension(:, :, 0:), intent(in), contiguous	boin,
		integer, dimension(:, :, 0:), intent(in), contiguous	boout,
		complex(kind=dp), dimension(:, :), intent(out), contiguous	sout,
		type(fft_scratch_type), intent(in)	fft_scratch
	)

...

Parameters

cin	...
boin	...
boout	...
sout	...
fft_scratch	...

History: none

Author: JGH (20-Jan-2001)

Definition at line 2007 of file fft_tools.F.

Here is the caller graph for this function:

◆ cube_transpose_2()

subroutine, public fft_tools::cube_transpose_2	(	complex(kind=dp), dimension(:, :), intent(in), contiguous	cin,
		integer, dimension(:, :, 0:), intent(in), contiguous	boin,
		integer, dimension(:, :, 0:), intent(in), contiguous	boout,
		complex(kind=dp), dimension(:, :), intent(out), contiguous	sout,
		type(fft_scratch_type), intent(in)	fft_scratch
	)

...

Parameters

cin	...
boin	...
boout	...
sout	...
fft_scratch	...

Definition at line 2097 of file fft_tools.F.

Here is the call graph for this function:

Here is the caller graph for this function:

◆ get_fft_scratch()

subroutine, public fft_tools::get_fft_scratch	(	type(fft_scratch_type), pointer	fft_scratch,
		integer, intent(in)	tf_type,
		integer, dimension(:), intent(in)	n,
		type(fft_scratch_sizes), intent(in), optional	fft_sizes
	)

...

Parameters

fft_scratch	...
tf_type	...
n	...
fft_sizes	...

Definition at line 2866 of file fft_tools.F.

Here is the call graph for this function:

Here is the caller graph for this function:

◆ release_fft_scratch()

subroutine, public fft_tools::release_fft_scratch ( type(fft_scratch_type), pointer fft_scratch )

...

Parameters

fft_scratch ...

Definition at line 3244 of file fft_tools.F.

Here is the caller graph for this function:

Variable Documentation

◆ init_fft_pool

integer, save fft_tools::init_fft_pool = 0

Definition at line 128 of file fft_tools.F.

◆ tick_fft_pool

integer, save fft_tools::tick_fft_pool = 0

Definition at line 130 of file fft_tools.F.

◆ fft_pool_scratch_limit

integer, save fft_tools::fft_pool_scratch_limit = 15

Definition at line 132 of file fft_tools.F.

◆ fft_scratch_first

type(fft_scratch_pool_type), pointer, save fft_tools::fft_scratch_first

Definition at line 133 of file fft_tools.F.

◆ fwfft

integer, parameter, public fft_tools::fwfft = +1

Definition at line 144 of file fft_tools.F.

◆ bwfft

integer, parameter, public fft_tools::bwfft = -1

Definition at line 144 of file fft_tools.F.

◆ fft_radix_closest

integer, parameter, public fft_tools::fft_radix_closest = 493

Definition at line 145 of file fft_tools.F.

◆ fft_radix_next

integer, parameter, public fft_tools::fft_radix_next = 494

Definition at line 145 of file fft_tools.F.

◆ fft_radix_next_odd

integer, parameter, public fft_tools::fft_radix_next_odd = 497

Definition at line 147 of file fft_tools.F.

◆ fft_type

integer, save, public fft_tools::fft_type = 0

Definition at line 152 of file fft_tools.F.

◆ fft_plan_style

integer, save, public fft_tools::fft_plan_style = 1

Definition at line 155 of file fft_tools.F.

Data Types

Functions/Subroutines

Variables

Detailed Description

Function/Subroutine Documentation

◆ init_fft()

◆ finalize_fft()

◆ fft_radix_operations()

◆ fft_fw1d()

◆ x_to_yz()

◆ yz_to_x()

◆ yz_to_xz()

◆ xz_to_yz()

◆ cube_transpose_1()

◆ cube_transpose_2()

◆ get_fft_scratch()

◆ release_fft_scratch()

Variable Documentation

◆ init_fft_pool

◆ tick_fft_pool

◆ fft_pool_scratch_limit

◆ fft_scratch_first

◆ fwfft

◆ bwfft

◆ fft_radix_closest

◆ fft_radix_next

◆ fft_radix_next_odd

◆ fft_type

◆ fft_plan_style