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
11
Issues
11
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
0278642d
Commit
0278642d
authored
Jul 31, 2017
by
Andreas Marek
Browse files
Options
Browse Files
Download
Email Patches
Plain Diff
Fix error in AVX512 real Block 4 and 6 kernels
parent
e5d03b59
Changes
2
Pipelines
1
Hide whitespace changes
Inline
Side-by-side
Showing
2 changed files
with
29 additions
and
26 deletions
+29
-26
src/elpa2/kernels/real_avx512_4hv_template.Xc
src/elpa2/kernels/real_avx512_4hv_template.Xc
+14
-14
src/elpa2/kernels/real_avx512_6hv_template.Xc
src/elpa2/kernels/real_avx512_6hv_template.Xc
+15
-12
No files found.
src/elpa2/kernels/real_avx512_4hv_template.Xc
View file @
0278642d
...
...
@@ -227,28 +227,28 @@ void quad_hh_trafo_real_avx512_4hv_single(float* q, float* hh, int* pnb, int* pn
{
hh_trafo_kernel_24_AVX512_4hv_double(&q[i], hh, nb, ldq, ldh, s_1_2, s_1_3, s_2_3, s_1_4, s_2_4, s_3_4);
}
#endif
#ifdef DOUBLE_PRECISION_REAL
if (nq-i == 16)
{
else
{
if (nq-i == 16)
{
hh_trafo_kernel_16_AVX512_4hv_double(&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_8_AVX512_4hv_double(&q[i], hh, nb, ldq, ldh, s_1_2, s_1_3, s_2_3, s_1_4, s_2_4, s_3_4);
}
}
#endif
#ifdef SINGLE_PRECISION_REAL
if (nq-i == 32)
{
hh_trafo_kernel_32_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);
}
#endif
else
#ifdef DOUBLE_PRECISION_REAL
{
hh_trafo_kernel_8_AVX512_4hv_double(&q[i], hh, nb, ldq, ldh, s_1_2, s_1_3, s_2_3, s_1_4, s_2_4, s_3_4);
}
#endif
#ifdef SINGLE_PRECISION_REAL
else
{
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_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);
}
#endif
}
...
...
src/elpa2/kernels/real_avx512_6hv_template.Xc
View file @
0278642d
...
...
@@ -47,6 +47,7 @@
#include "config-f90.h"
#include <x86intrin.h>
#include <stdio.h>
#define __forceinline __attribute__((always_inline)) static
...
...
@@ -289,23 +290,25 @@ void hexa_hh_trafo_real_avx512_6hv_single(float* q, float* hh, int* pnb, int* pn
{
hh_trafo_kernel_24_AVX512_6hv_double(&q[i], hh, nb, ldq, ldh, scalarprods);
}
if (nq-i == 16)
{
hh_trafo_kernel_16_AVX512_6hv_double(&q[i], hh, nb, ldq, ldh, scalarprods);
}
#endif
else
{
#ifdef DOUBLE_PRECISION_REAL
hh_trafo_kernel_8_AVX512_6hv_double(&q[i], hh, nb, ldq, ldh, scalarprods);
{
if (nq-i == 16)
{
hh_trafo_kernel_16_AVX512_6hv_double(&q[i], hh, nb, ldq, ldh, scalarprods);
}
else
{
hh_trafo_kernel_8_AVX512_6hv_double(&q[i], hh, nb, ldq, ldh, scalarprods);
}
}
#endif
#ifdef SINGLE_PRECISION_REAL
if (nq-i == 16) {
hh_trafo_kernel_16_AVX512_6hv_single(&q[i], hh, nb, ldq, ldh, scalarprods);
} else {
printf("ERROR in avx512 kernel\n");
}
#endif
}
}
/**
...
...
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