Skip to content

*fix* corrected offsets when executing teampolicy things #3

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 2 commits into
base: master
Choose a base branch
from
Open

*fix* corrected offsets when executing teampolicy things #3

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
2180239
*fix* corrected offsets when executing teampolicy things (previous co…
Nov 6, 2017
dafb876
*fix* we have to loop of course over chunk->y-halo_depth
Nov 13, 2017
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 6 additions & 6 deletions 2d/c_kernels/kokkos_np/cg.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -133,10 +133,10 @@ struct CGCalcW
void operator()(const team_member& team, value_type& pw) const
{
double pw_team = 0.0;
const int team_offset = (team.league_rank() + halo_depth)*y;
const int team_offset = (team.league_rank() + halo_depth)*x;

Kokkos::parallel_reduce(
Kokkos::TeamThreadRange(team, halo_depth, y-halo_depth),
Kokkos::TeamThreadRange(team, halo_depth, x-halo_depth),
[&] (const int &j, double& pw_thread)
{
const int index = team_offset + j;
Expand Down Expand Up @@ -178,10 +178,10 @@ struct CGCalcUR
void operator()(const team_member& team, value_type& rrn) const
{
double rrn_team = 0.0;
const int team_offset = (team.league_rank() + halo_depth)*y;
const int team_offset = (team.league_rank() + halo_depth)*x;

Kokkos::parallel_reduce(
Kokkos::TeamThreadRange(team, halo_depth, y-halo_depth),
Kokkos::TeamThreadRange(team, halo_depth, x-halo_depth),
[&] (const int &j, double& rrn_thread)
{
const int index = team_offset + j;
Expand Down Expand Up @@ -221,10 +221,10 @@ struct CGCalcP
KOKKOS_INLINE_FUNCTION
void operator()(const team_member& team) const
{
const int team_offset = (team.league_rank() + halo_depth)*y;
const int team_offset = (team.league_rank() + halo_depth)*x;

Kokkos::parallel_for(
Kokkos::TeamThreadRange(team, halo_depth, y-halo_depth),
Kokkos::TeamThreadRange(team, halo_depth, x-halo_depth),
[&] (const int &j)
{
const int index = team_offset + j;
Expand Down
8 changes: 4 additions & 4 deletions 2d/c_kernels/kokkos_np/cheby.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -58,10 +58,10 @@ struct ChebyCalcU
KOKKOS_INLINE_FUNCTION
void operator()(const team_member& team) const
{
const int team_offset = (team.league_rank() + halo_depth)*y;
const int team_offset = (team.league_rank() + halo_depth)*x;

Kokkos::parallel_for(
Kokkos::TeamThreadRange(team, halo_depth, y-halo_depth),
Kokkos::TeamThreadRange(team, halo_depth, x-halo_depth),
[&] (const int &j)
{
const int index = team_offset + j;
Expand Down Expand Up @@ -94,10 +94,10 @@ struct ChebyIterate
KOKKOS_INLINE_FUNCTION
void operator()(const team_member& team) const
{
const int team_offset = (team.league_rank() + halo_depth)*y;
const int team_offset = (team.league_rank() + halo_depth)*x;

Kokkos::parallel_for(
Kokkos::TeamThreadRange(team, halo_depth, y-halo_depth),
Kokkos::TeamThreadRange(team, halo_depth, x-halo_depth),
[&] (const int &j)
{
const int index = team_offset + j;
Expand Down
14 changes: 7 additions & 7 deletions 2d/c_kernels/kokkos_np/kernel_interface.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -230,7 +230,7 @@ void run_cg_calc_w(Chunk* chunk, Settings* settings, double* pw)
chunk->p, chunk->kx, chunk->ky);

parallel_reduce(
TeamPolicy<DEVICE>(chunk->x - 2*settings->halo_depth, AUTO),
TeamPolicy<DEVICE>(chunk->y - 2*settings->halo_depth, AUTO),
cg_calc_w, *pw);

STOP_PROFILING(settings->kernel_profile, __func__);
Expand All @@ -246,7 +246,7 @@ void run_cg_calc_ur(
chunk->p, chunk->w, alpha);

parallel_reduce(
TeamPolicy<DEVICE>(chunk->x - 2*settings->halo_depth, AUTO),
TeamPolicy<DEVICE>(chunk->y - 2*settings->halo_depth, AUTO),
cg_calc_ur, *rrn);

STOP_PROFILING(settings->kernel_profile, __func__);
Expand All @@ -260,7 +260,7 @@ void run_cg_calc_p(Chunk* chunk, Settings* settings, double beta)
chunk->x, chunk->y, settings->halo_depth, beta, chunk->p, chunk->r);

parallel_for(
TeamPolicy<DEVICE>(chunk->x - 2*settings->halo_depth, AUTO),
TeamPolicy<DEVICE>(chunk->y - 2*settings->halo_depth, AUTO),
cg_calc_p);

STOP_PROFILING(settings->kernel_profile, __func__);
Expand Down Expand Up @@ -292,14 +292,14 @@ void run_cheby_iterate(
chunk->kx, chunk->ky);

parallel_for(
TeamPolicy<DEVICE>(chunk->x - 2*settings->halo_depth, AUTO),
TeamPolicy<DEVICE>(chunk->y - 2*settings->halo_depth, AUTO),
cheby_iterate);

ChebyCalcU<DEVICE> cheby_calc_u(
chunk->x, chunk->y, settings->halo_depth, chunk->p, chunk->u);

parallel_for(
TeamPolicy<DEVICE>(chunk->x - 2*settings->halo_depth, AUTO),
TeamPolicy<DEVICE>(chunk->y - 2*settings->halo_depth, AUTO),
cheby_calc_u);

STOP_PROFILING(settings->kernel_profile, __func__);
Expand Down Expand Up @@ -363,15 +363,15 @@ void run_ppcg_inner_iteration(
chunk->u, chunk->kx, chunk->ky);

parallel_for(
TeamPolicy<DEVICE>(chunk->x - 2*settings->halo_depth, AUTO),
TeamPolicy<DEVICE>(chunk->y - 2*settings->halo_depth, AUTO),
ppcg_calc_ur);

PPCGCalcSd<DEVICE> ppcg_calc_sd(
chunk->x, chunk->y, settings->halo_depth, chunk->theta, alpha, beta,
chunk->sd, chunk->r);

parallel_for(
TeamPolicy<DEVICE>(chunk->x - 2*settings->halo_depth, AUTO),
TeamPolicy<DEVICE>(chunk->y - 2*settings->halo_depth, AUTO),
ppcg_calc_sd);

STOP_PROFILING(settings->kernel_profile, __func__);
Expand Down
8 changes: 4 additions & 4 deletions 2d/c_kernels/kokkos_np/ppcg.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -54,10 +54,10 @@ struct PPCGCalcUR
KOKKOS_INLINE_FUNCTION
void operator()(const team_member& team) const
{
const int team_offset = (team.league_rank() + halo_depth)*y;
const int team_offset = (team.league_rank() + halo_depth)*x;

Kokkos::parallel_for(
Kokkos::TeamThreadRange(team, halo_depth, y-halo_depth),
Kokkos::TeamThreadRange(team, halo_depth, x-halo_depth),
[&] (const int &j)
{
const int index = team_offset + j;
Expand Down Expand Up @@ -93,10 +93,10 @@ struct PPCGCalcSd
KOKKOS_INLINE_FUNCTION
void operator()(const team_member& team) const
{
const int team_offset = (team.league_rank() + halo_depth)*y;
const int team_offset = (team.league_rank() + halo_depth)*x;

Kokkos::parallel_for(
Kokkos::TeamThreadRange(team, halo_depth, y-halo_depth),
Kokkos::TeamThreadRange(team, halo_depth, x-halo_depth),
[&] (const int &j)
{
const int index = team_offset + j;
Expand Down