Skip to content
GitLab
Projects
Groups
Snippets
Help
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in
Toggle navigation
elpa
Project overview
Project overview
Details
Activity
Releases
Repository
Repository
Files
Commits
Branches
Tags
Contributors
Graph
Compare
Issues
14
Issues
14
List
Boards
Labels
Service Desk
Milestones
Merge Requests
0
Merge Requests
0
Operations
Operations
Incidents
Environments
Analytics
Analytics
Repository
Value Stream
Wiki
Wiki
Members
Members
Collapse sidebar
Close sidebar
Activity
Graph
Create a new issue
Commits
Issue Boards
Open sidebar
elpa
elpa
Commits
0a06af6a
Commit
0a06af6a
authored
Sep 20, 2016
by
Andreas Marek
Browse files
Options
Browse Files
Download
Email Patches
Plain Diff
Try different step size for real single precision AVX block4 kernel
parent
7560a7b3
Changes
1
Pipelines
1
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
with
35 additions
and
18 deletions
+35
-18
src/elpa2_kernels/elpa2_kernels_real_avx-avx2_4hv_single_precision.c
...ernels/elpa2_kernels_real_avx-avx2_4hv_single_precision.c
+35
-18
No files found.
src/elpa2_kernels/elpa2_kernels_real_avx-avx2_4hv_single_precision.c
View file @
0a06af6a
...
...
@@ -135,8 +135,7 @@ void quad_hh_trafo_real_avx_avx2_4hv_single(float* q, float* hh, int* pnb, int*
s_1_4
+=
hh
[
i
-
3
]
*
hh
[
i
+
(
ldh
*
3
)];
}
// Production level kernel calls with padding
for
(
i
=
0
;
i
<
nq
-
20
;
i
+=
24
)
for
(
i
=
0
;
i
<
nq
-
16
;
i
+=
24
)
{
hh_trafo_kernel_24_AVX_4hv_single
(
&
q
[
i
],
hh
,
nb
,
ldq
,
ldh
,
s_1_2
,
s_1_3
,
s_2_3
,
s_1_4
,
s_2_4
,
s_3_4
);
}
...
...
@@ -144,29 +143,47 @@ void quad_hh_trafo_real_avx_avx2_4hv_single(float* q, float* hh, int* pnb, int*
{
return
;
}
if
(
nq
-
i
==
20
)
if
(
nq
-
i
==
16
)
{
hh_trafo_kernel_16_AVX_4hv_single
(
&
q
[
i
],
hh
,
nb
,
ldq
,
ldh
,
s_1_2
,
s_1_3
,
s_2_3
,
s_1_4
,
s_2_4
,
s_3_4
);
hh_trafo_kernel_4_AVX_4hv_single
(
&
q
[
i
+
16
],
hh
,
nb
,
ldq
,
ldh
,
s_1_2
,
s_1_3
,
s_2_3
,
s_1_4
,
s_2_4
,
s_3_4
);
}
else
if
(
nq
-
i
==
16
)
{
hh_trafo_kernel_16_AVX_4hv_single
(
&
q
[
i
],
hh
,
nb
,
ldq
,
ldh
,
s_1_2
,
s_1_3
,
s_2_3
,
s_1_4
,
s_2_4
,
s_3_4
);
}
else
if
(
nq
-
i
==
12
)
else
{
hh_trafo_kernel_8_AVX_4hv_single
(
&
q
[
i
],
hh
,
nb
,
ldq
,
ldh
,
s_1_2
,
s_1_3
,
s_2_3
,
s_1_4
,
s_2_4
,
s_3_4
);
hh_trafo_kernel_4_AVX_4hv_single
(
&
q
[
i
+
8
],
hh
,
nb
,
ldq
,
ldh
,
s_1_2
,
s_1_3
,
s_2_3
,
s_1_4
,
s_2_4
,
s_3_4
);
}
else
if
(
nq
-
i
==
8
)
{
hh_trafo_kernel_8_AVX_4hv_single
(
&
q
[
i
],
hh
,
nb
,
ldq
,
ldh
,
s_1_2
,
s_1_3
,
s_2_3
,
s_1_4
,
s_2_4
,
s_3_4
);
}
else
{
hh_trafo_kernel_4_AVX_4hv_single
(
&
q
[
i
],
hh
,
nb
,
ldq
,
ldh
,
s_1_2
,
s_1_3
,
s_2_3
,
s_1_4
,
s_2_4
,
s_3_4
);
}
// // Production level kernel calls with padding
// for (i = 0; i < nq-20; i+=24)
// {
// hh_trafo_kernel_24_AVX_4hv_single(&q[i], hh, nb, ldq, ldh, s_1_2, s_1_3, s_2_3, s_1_4, s_2_4, s_3_4);
// }
// if (nq == i)
// {
// return;
// }
// if (nq-i == 20)
// {
// hh_trafo_kernel_16_AVX_4hv_single(&q[i], hh, nb, ldq, ldh, s_1_2, s_1_3, s_2_3, s_1_4, s_2_4, s_3_4);
// hh_trafo_kernel_4_AVX_4hv_single(&q[i+16], hh, nb, ldq, ldh, s_1_2, s_1_3, s_2_3, s_1_4, s_2_4, s_3_4);
// }
// else if (nq-i == 16)
// {
// hh_trafo_kernel_16_AVX_4hv_single(&q[i], hh, nb, ldq, ldh, s_1_2, s_1_3, s_2_3, s_1_4, s_2_4, s_3_4);
// }
// else if (nq-i == 12)
// {
// hh_trafo_kernel_8_AVX_4hv_single(&q[i], hh, nb, ldq, ldh, s_1_2, s_1_3, s_2_3, s_1_4, s_2_4, s_3_4);
// hh_trafo_kernel_4_AVX_4hv_single(&q[i+8], hh, nb, ldq, ldh, s_1_2, s_1_3, s_2_3, s_1_4, s_2_4, s_3_4);
// }
// else if (nq-i == 8)
// {
// hh_trafo_kernel_8_AVX_4hv_single(&q[i], hh, nb, ldq, ldh, s_1_2, s_1_3, s_2_3, s_1_4, s_2_4, s_3_4);
// }
// else
// {
// hh_trafo_kernel_4_AVX_4hv_single(&q[i], hh, nb, ldq, ldh, s_1_2, s_1_3, s_2_3, s_1_4, s_2_4, s_3_4);
// }
}
...
...
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