Skip to content
GitLab
Explore
Sign in
Primary navigation
Search or go to…
Project
pystencils
Manage
Activity
Members
Labels
Plan
Issues
Issue boards
Milestones
Wiki
Code
Merge requests
Repository
Branches
Commits
Tags
Repository graph
Compare revisions
Snippets
Build
Pipelines
Jobs
Pipeline schedules
Artifacts
Deploy
Releases
Model registry
Operate
Environments
Monitor
Incidents
Analyze
Value stream analytics
Contributor analytics
CI/CD analytics
Repository analytics
Model experiments
Help
Help
Support
GitLab documentation
Compare GitLab plans
Community forum
Contribute to GitLab
Provide feedback
Terms and privacy
Keyboard shortcuts
?
Snippets
Groups
Projects
Show more breadcrumbs
pycodegen
pystencils
Commits
f47fabfa
Commit
f47fabfa
authored
5 years ago
by
Michael Kuron
Browse files
Options
Downloads
Patches
Plain Diff
AES-NI: Implement _my_cvtepu64_pd in terms of SSE instructions
parent
ddbe84da
Branches
Branches containing commit
Tags
Tags containing commit
1 merge request
!46
AES-NI vectorization improvements
Changes
1
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
pystencils/include/aesni_rand.h
+13
-6
13 additions, 6 deletions
pystencils/include/aesni_rand.h
with
13 additions
and
6 deletions
pystencils/include/aesni_rand.h
+
13
−
6
View file @
f47fabfa
#if !defined(__AES__) || !defined(__SSE2__)
#if !defined(__AES__) || !defined(__SSE2__)
|| (!defined(__AVX512VL__) && !defined(__SSE4_1__))
#error AES-NI and SSE2 need to be enabled
#error AES-NI and SSE2
, as well as AVX512 or SSE4.1
need to be enabled
#endif
#endif
#include
<emmintrin.h>
// SSE2
#include
<emmintrin.h>
// SSE2
#include
<wmmintrin.h>
// AES
#include
<wmmintrin.h>
// AES
#ifdef __AVX512VL__
#ifdef __AVX512VL__
#include
<immintrin.h>
// AVX*
#include
<immintrin.h>
// AVX*
#else
#include
<smmintrin.h>
// SSE4
#endif
#endif
#include
<cstdint>
#include
<cstdint>
...
@@ -44,14 +46,19 @@ QUALIFIERS __m128 _my_cvtepu32_ps(const __m128i v)
...
@@ -44,14 +46,19 @@ QUALIFIERS __m128 _my_cvtepu32_ps(const __m128i v)
#endif
#endif
}
}
#if !defined(__AVX512VL__) && defined(__FAST_MATH__) && defined(__GNUC__) && __GNUC__ >= 5
__attribute__
((
optimize
(
"no-fast-math"
)))
#endif
QUALIFIERS
__m128d
_my_cvtepu64_pd
(
const
__m128i
x
)
QUALIFIERS
__m128d
_my_cvtepu64_pd
(
const
__m128i
x
)
{
{
#ifdef __AVX512VL__
#ifdef __AVX512VL__
return
_mm_cvtepu64_pd
(
x
);
return
_mm_cvtepu64_pd
(
x
);
#else
#else
uint64
r
[
2
];
__m128i
xH
=
_mm_srli_epi64
(
x
,
32
);
_mm_storeu_si128
((
__m128i
*
)
r
,
x
);
xH
=
_mm_or_si128
(
xH
,
_mm_castpd_si128
(
_mm_set1_pd
(
19342813113834066795298816
.)));
// 2^84
return
_mm_set_pd
((
double
)
r
[
1
],
(
double
)
r
[
0
]);
__m128i
xL
=
_mm_blend_epi16
(
x
,
_mm_castpd_si128
(
_mm_set1_pd
(
0x0010000000000000
)),
0xcc
);
// 2^52
__m128d
f
=
_mm_sub_pd
(
_mm_castsi128_pd
(
xH
),
_mm_set1_pd
(
19342813118337666422669312
.));
// 2^84 + 2^52
return
_mm_add_pd
(
f
,
_mm_castsi128_pd
(
xL
));
#endif
#endif
}
}
...
@@ -110,4 +117,4 @@ QUALIFIERS void aesni_float4(uint32 ctr0, uint32 ctr1, uint32 ctr2, uint32 ctr3,
...
@@ -110,4 +117,4 @@ QUALIFIERS void aesni_float4(uint32 ctr0, uint32 ctr1, uint32 ctr2, uint32 ctr3,
rnd2
=
r
[
1
];
rnd2
=
r
[
1
];
rnd3
=
r
[
2
];
rnd3
=
r
[
2
];
rnd4
=
r
[
3
];
rnd4
=
r
[
3
];
}
}
\ No newline at end of file
This diff is collapsed.
Click to expand it.
Preview
0%
Loading
Try again
or
attach a new file
.
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Save comment
Cancel
Please
register
or
sign in
to comment