adam_ripple_fnl_kernels

RIPPLE, the FNL (OpenACC) device kernels.

The device twins of the CPU loops. Device arrays are transposed, (b, i, j, k, v): the block index is stride-1, so it is the innermost loop of every nest. Host scalars are firstprivate, device arrays DEVICEVAR (FUNDAL's deviceptr), private work arrays have constant bounds.

Source: src/app/ripple/fnl/adam_ripple_fnl_kernels.F90

Dependencies

Contents

• compute_face_fluxes_dev
• compute_flux_difference_dev
• pack_face_skin_dev
• apply_reflux_face_dev

Subroutines

compute_face_fluxes_dev

Face fluxes along the axis of unit step (di, dj, dk), device twin of the CPU face_fluxes: the split fluxes a+ u (cells m = 1-S ... S-1) and a- u (cells m+1) of the stencil of face i+1/2, reconstructed by WENO and summed.

subroutine compute_face_fluxes_dev(di, dj, dk, ni, nj, nk, ngc, blocks_number, S, a, weno_a_gpu, weno_p_gpu, weno_d_gpu, weno_zeps, q_gpu, fl_gpu)

Arguments

Name	Type	Intent	Attributes	Description
di	integer(kind=I4P)	in		Unit step.
dj	integer(kind=I4P)	in		Unit step.
dk	integer(kind=I4P)	in		Unit step.
ni	integer(kind=I4P)	in		Grid dimensions.
nj	integer(kind=I4P)	in		Grid dimensions.
nk	integer(kind=I4P)	in		Grid dimensions.
ngc	integer(kind=I4P)	in		Grid dimensions.
blocks_number	integer(kind=I4P)	in		Blocks in use.
S	integer(kind=I4P)	in		WENO half stencil.
a	real(kind=R8P)	in		Velocity component.
weno_a_gpu	real(kind=R8P)	in		WENO optimal weights.
weno_p_gpu	real(kind=R8P)	in		WENO polynomials coefficients.
weno_d_gpu	real(kind=R8P)	in		WENO smoothness indicators coefficients.
weno_zeps	real(kind=R8P)	in		WENO parameter avoiding division by zero.
q_gpu	real(kind=R8P)	in		State.
fl_gpu	real(kind=R8P)	inout		Face fluxes.

Call graph

compute_flux_difference_dev

Residuals dq = -sum_d w_d (F_{i+1/2} - F_{i-1/2}) / dx_d, w_d 1 on an active axis, 0 on a null one.

subroutine compute_flux_difference_dev(ni, nj, nk, ngc, blocks_number, w, dxyz_gpu, flx_gpu, fly_gpu, flz_gpu, dq_gpu)

Arguments

Name	Type	Intent	Attributes	Description
ni	integer(kind=I4P)	in		Grid dimensions.
nj	integer(kind=I4P)	in		Grid dimensions.
nk	integer(kind=I4P)	in		Grid dimensions.
ngc	integer(kind=I4P)	in		Grid dimensions.
blocks_number	integer(kind=I4P)	in		Blocks in use.
w	real(kind=R8P)	in		Axis weights.
dxyz_gpu	real(kind=R8P)	in		Block spacings [nb, 3].
flx_gpu	real(kind=R8P)	in		X-face fluxes.
fly_gpu	real(kind=R8P)	in		Y-face fluxes.
flz_gpu	real(kind=R8P)	in		Z-face fluxes.
dq_gpu	real(kind=R8P)	inout		Residuals.

Call graph

pack_face_skin_dev

Pack the face fluxes of face fec of block b into skin_gpu(1, c), c running over the two tangential axes, inner fastest (x faces: j, k; y faces: i, k; z faces: i, j): the flux register's order.

subroutine pack_face_skin_dev(fec, b, ni, nj, nk, flx_gpu, fly_gpu, flz_gpu, skin_gpu)

Arguments

Name	Type	Intent	Attributes	Description
fec	integer(kind=I4P)	in		Face (1..6: -x, +x, -y, +y, -z, +z), block.
b	integer(kind=I4P)	in		Face (1..6: -x, +x, -y, +y, -z, +z), block.
ni	integer(kind=I4P)	in		Grid dimensions.
nj	integer(kind=I4P)	in		Grid dimensions.
nk	integer(kind=I4P)	in		Grid dimensions.
flx_gpu	real(kind=R8P)	in		X-face fluxes.
fly_gpu	real(kind=R8P)	in		Y-face fluxes.
flz_gpu	real(kind=R8P)	in		Z-face fluxes.
skin_gpu	real(kind=R8P)	inout		Skin (1, cells).

Call graph

apply_reflux_face_dev

Add scale delta(1, c) to the coarse skin cells of block b on the face of normal axis and side sgn.

subroutine apply_reflux_face_dev(axis, sgn, b, ni, nj, nk, ngc, nface_cells, scale, delta_gpu, q_gpu)

Arguments

Name	Type	Intent	Attributes	Description
axis	integer(kind=I4P)	in		Face normal axis (1..3) and side (+-1).
sgn	integer(kind=I4P)	in		Face normal axis (1..3) and side (+-1).
b	integer(kind=I4P)	in		Coarse block.
ni	integer(kind=I4P)	in		Grid dimensions.
nj	integer(kind=I4P)	in		Grid dimensions.
nk	integer(kind=I4P)	in		Grid dimensions.
ngc	integer(kind=I4P)	in		Grid dimensions.
nface_cells	integer(kind=I4P)	in		Skin cells.
scale	real(kind=R8P)	in		sgn dt / dx_coarse.
delta_gpu	real(kind=R8P)	in		Flux mismatch F_coarse - F_fine (1, cells).
q_gpu	real(kind=R8P)	inout		State.

Call graph