More patches

This commit is contained in:
Susi Lehtola 2019-02-26 01:11:21 +01:00
parent ee6a88c55d
commit 4e591d8725
4 changed files with 4352 additions and 1 deletions

3283
1965.patch Normal file

File diff suppressed because it is too large Load Diff

960
1966.patch Normal file
View File

@ -0,0 +1,960 @@
From 63cdd8f4a04f3a5ac1733e202b6b3678c34fb8dd Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Wed, 16 Jan 2019 23:27:38 +0100
Subject: [PATCH 01/18] Tag arguments 0 and 1 as both input and output
---
kernel/x86_64/cscal_microk_bulldozer-2.c | 32 ++++++++++++------------
1 file changed, 16 insertions(+), 16 deletions(-)
diff --git a/kernel/x86_64/cscal_microk_bulldozer-2.c b/kernel/x86_64/cscal_microk_bulldozer-2.c
index 3abffc4cf..f526fd611 100644
--- a/kernel/x86_64/cscal_microk_bulldozer-2.c
+++ b/kernel/x86_64/cscal_microk_bulldozer-2.c
@@ -116,11 +116,11 @@ static void cscal_kernel_16( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"%0", "%1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
@@ -208,11 +208,11 @@ static void cscal_kernel_16_zero_r( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"%0", "%1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
@@ -285,11 +285,11 @@ static void cscal_kernel_16_zero_i( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"%0", "%1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
@@ -330,11 +330,11 @@ static void cscal_kernel_16_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"%0", "%1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
From b6136be686e415fbdb035267c5020cb08e4e49ac Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Wed, 16 Jan 2019 23:30:03 +0100
Subject: [PATCH 02/18] Tag arguments 0 and 1 as both input and output
---
kernel/x86_64/cscal_microk_haswell-2.c | 30 +++++++++++++-------------
1 file changed, 15 insertions(+), 15 deletions(-)
diff --git a/kernel/x86_64/cscal_microk_haswell-2.c b/kernel/x86_64/cscal_microk_haswell-2.c
index 0a4eb683c..8623dcd10 100644
--- a/kernel/x86_64/cscal_microk_haswell-2.c
+++ b/kernel/x86_64/cscal_microk_haswell-2.c
@@ -116,11 +116,11 @@ static void cscal_kernel_16( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"0", "1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
@@ -208,9 +208,9 @@ static void cscal_kernel_16_zero_r( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
: "cc", // "0", "1",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
@@ -285,9 +285,9 @@ static void cscal_kernel_16_zero_i( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
: "cc", //"%0", "%1",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
@@ -329,12 +329,12 @@ static void cscal_kernel_16_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
- :
- :
- "r" (n), // 0
- "r" (x), // 1
+ :
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"0", "1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
From f447fb4c54870710cd6304553df59f50ff51b8f5 Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Wed, 16 Jan 2019 23:32:48 +0100
Subject: [PATCH 03/18] Tag arguments 0 and 1 as both input and output
---
kernel/x86_64/cscal_microk_steamroller-2.c | 32 +++++++++++-----------
1 file changed, 16 insertions(+), 16 deletions(-)
diff --git a/kernel/x86_64/cscal_microk_steamroller-2.c b/kernel/x86_64/cscal_microk_steamroller-2.c
index 8346e1748..fbeb857e2 100644
--- a/kernel/x86_64/cscal_microk_steamroller-2.c
+++ b/kernel/x86_64/cscal_microk_steamroller-2.c
@@ -117,11 +117,11 @@ static void cscal_kernel_16( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"0", "1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
@@ -208,12 +208,12 @@ static void cscal_kernel_16_zero_r( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
+ :
+ "+r" (n), // 0
+ "+r" (x), // 1
:
- :
- "r" (n), // 0
- "r" (x), // 1
"r" (alpha) // 2
- : "cc", //"0", "1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
@@ -286,11 +286,11 @@ static void cscal_kernel_16_zero_i( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"%0", "%1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
@@ -331,11 +331,11 @@ static void cscal_kernel_16_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"0", "1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
From fcd7fde5702cf7270332a5dd747f83efe7be93dd Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Wed, 16 Jan 2019 23:35:18 +0100
Subject: [PATCH 04/18] Tag arguments 0 and 1 as both input and output
---
kernel/x86_64/dscal_microk_bulldozer-2.c | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/kernel/x86_64/dscal_microk_bulldozer-2.c b/kernel/x86_64/dscal_microk_bulldozer-2.c
index de53b0bc4..71d3a9846 100644
--- a/kernel/x86_64/dscal_microk_bulldozer-2.c
+++ b/kernel/x86_64/dscal_microk_bulldozer-2.c
@@ -122,9 +122,9 @@ static void dscal_kernel_8( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n1), // 0
- "r" (x), // 1
+ "+r" (n1), // 0
+ "+r" (x), // 1
+ :
"r" (alpha), // 2
"r" (n2) // 3
: "cc",
@@ -188,9 +188,9 @@ static void dscal_kernel_8_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n1), // 0
- "r" (x), // 1
+ "+r" (n1), // 0
+ "+r" (x), // 1
+ :
"r" (alpha), // 2
"r" (n2) // 3
: "cc",
From 05e961994401bfc6dc8639fa9bc159148569ca9d Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Wed, 16 Jan 2019 23:36:37 +0100
Subject: [PATCH 05/18] Tag arguments 0 and 1 as both input and output
---
kernel/x86_64/dscal_microk_haswell-2.c | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/kernel/x86_64/dscal_microk_haswell-2.c b/kernel/x86_64/dscal_microk_haswell-2.c
index e732a2718..90790cfdc 100644
--- a/kernel/x86_64/dscal_microk_haswell-2.c
+++ b/kernel/x86_64/dscal_microk_haswell-2.c
@@ -122,9 +122,9 @@ static void dscal_kernel_8( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n1), // 0
- "r" (x), // 1
+ "+r" (n1), // 0
+ "+r" (x), // 1
+ :
"r" (alpha), // 2
"r" (n2) // 3
: "cc",
@@ -187,10 +187,10 @@ static void dscal_kernel_8_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
+ :
+ "+r" (n1), // 0
+ "+r" (x), // 1
:
- :
- "r" (n1), // 0
- "r" (x), // 1
"r" (alpha), // 2
"r" (n2) // 3
: "cc",
From 7a11cc5b9f7c9669ee1f9818a1ea3f44c2f6d98d Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Wed, 16 Jan 2019 23:37:49 +0100
Subject: [PATCH 06/18] Tag arguments 0 and 1 as both input and output
---
kernel/x86_64/dscal_microk_sandy-2.c | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/kernel/x86_64/dscal_microk_sandy-2.c b/kernel/x86_64/dscal_microk_sandy-2.c
index 8d855072b..0f187ba88 100644
--- a/kernel/x86_64/dscal_microk_sandy-2.c
+++ b/kernel/x86_64/dscal_microk_sandy-2.c
@@ -122,9 +122,9 @@ static void dscal_kernel_8( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n1), // 0
- "r" (x), // 1
+ "+r" (n1), // 0
+ "+r" (x), // 1
+ :
"r" (alpha), // 2
"r" (n2) // 3
: "cc",
@@ -187,10 +187,10 @@ static void dscal_kernel_8_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
+ :
+ "+r" (n1), // 0
+ "+r" (x), // 1
:
- :
- "r" (n1), // 0
- "r" (x), // 1
"r" (alpha), // 2
"r" (n2) // 3
: "cc",
From a6c06bffe1ec60ec359b300b8cc9e18b30c72d0d Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Wed, 16 Jan 2019 23:40:28 +0100
Subject: [PATCH 07/18] Tag arguments 0 and 1 as both input and output
---
kernel/x86_64/zscal_microk_bulldozer-2.c | 16 ++++++++--------
1 file changed, 8 insertions(+), 8 deletions(-)
diff --git a/kernel/x86_64/zscal_microk_bulldozer-2.c b/kernel/x86_64/zscal_microk_bulldozer-2.c
index 03882d6b6..1ce59d2c7 100644
--- a/kernel/x86_64/zscal_microk_bulldozer-2.c
+++ b/kernel/x86_64/zscal_microk_bulldozer-2.c
@@ -116,11 +116,11 @@ static void zscal_kernel_8( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"%0", "%1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
@@ -208,11 +208,11 @@ static void zscal_kernel_8_zero_r( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"%0", "%1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
From 5efc7ce079fd87de9ab7ca20aaaf8c5c627170fa Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Wed, 16 Jan 2019 23:42:34 +0100
Subject: [PATCH 08/18] Tag arguments 0 and 1 as both input and output
---
kernel/x86_64/zscal_microk_haswell-2.c | 32 +++++++++++++-------------
1 file changed, 16 insertions(+), 16 deletions(-)
diff --git a/kernel/x86_64/zscal_microk_haswell-2.c b/kernel/x86_64/zscal_microk_haswell-2.c
index d9253c1ed..534370959 100644
--- a/kernel/x86_64/zscal_microk_haswell-2.c
+++ b/kernel/x86_64/zscal_microk_haswell-2.c
@@ -116,11 +116,11 @@ static void zscal_kernel_8( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"%0", "%1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
@@ -208,11 +208,11 @@ static void zscal_kernel_8_zero_r( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"%0", "%1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
@@ -285,11 +285,11 @@ static void zscal_kernel_8_zero_i( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"%0", "%1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
@@ -330,11 +330,11 @@ static void zscal_kernel_8_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"%0", "%1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
From 1a1471c6be597a176a4dbfe2757c134eb3780af0 Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Wed, 16 Jan 2019 23:44:42 +0100
Subject: [PATCH 09/18] Tag arguments 0 and 1 as both input and output
---
kernel/x86_64/zscal_microk_steamroller-2.c | 32 +++++++++++-----------
1 file changed, 16 insertions(+), 16 deletions(-)
diff --git a/kernel/x86_64/zscal_microk_steamroller-2.c b/kernel/x86_64/zscal_microk_steamroller-2.c
index 97b07add6..4b489d9f3 100644
--- a/kernel/x86_64/zscal_microk_steamroller-2.c
+++ b/kernel/x86_64/zscal_microk_steamroller-2.c
@@ -116,12 +116,12 @@ static void zscal_kernel_8( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
+ :
+ "+r" (n), // 0
+ "+r" (x), // 1
:
- :
- "r" (n), // 0
- "r" (x), // 1
"r" (alpha) // 2
- : "cc", //"%0", "%1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
@@ -209,11 +209,11 @@ static void zscal_kernel_8_zero_r( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"%0", "%1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
@@ -286,11 +286,11 @@ static void zscal_kernel_8_zero_i( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"%0", "%1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
@@ -331,11 +331,11 @@ static void zscal_kernel_8_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x), // 1
+ :
"r" (alpha) // 2
- : "cc", //"%0", "%1",
+ : "cc",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
"%xmm4", "%xmm5", "%xmm6", "%xmm7",
"%xmm8", "%xmm9", "%xmm10", "%xmm11",
From 90e28665183cd8da3a6129016977f57dd415c6a9 Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Thu, 17 Jan 2019 09:38:20 +0100
Subject: [PATCH 10/18] Remove stray comma
---
kernel/x86_64/cscal_microk_bulldozer-2.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/kernel/x86_64/cscal_microk_bulldozer-2.c b/kernel/x86_64/cscal_microk_bulldozer-2.c
index f526fd611..31451aa6c 100644
--- a/kernel/x86_64/cscal_microk_bulldozer-2.c
+++ b/kernel/x86_64/cscal_microk_bulldozer-2.c
@@ -117,7 +117,7 @@ static void cscal_kernel_16( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
@@ -209,7 +209,7 @@ static void cscal_kernel_16_zero_r( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
@@ -286,7 +286,7 @@ static void cscal_kernel_16_zero_i( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
@@ -331,7 +331,7 @@ static void cscal_kernel_16_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
From b8dd71bddcb41d3d88af1a1eb77f845760452f5f Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Thu, 17 Jan 2019 09:39:23 +0100
Subject: [PATCH 11/18] Remove stray comma
---
kernel/x86_64/cscal_microk_haswell-2.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/kernel/x86_64/cscal_microk_haswell-2.c b/kernel/x86_64/cscal_microk_haswell-2.c
index 8623dcd10..a04a4c4ab 100644
--- a/kernel/x86_64/cscal_microk_haswell-2.c
+++ b/kernel/x86_64/cscal_microk_haswell-2.c
@@ -117,7 +117,7 @@ static void cscal_kernel_16( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
@@ -209,7 +209,7 @@ static void cscal_kernel_16_zero_r( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc", // "0", "1",
@@ -286,7 +286,7 @@ static void cscal_kernel_16_zero_i( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc", //"%0", "%1",
@@ -331,7 +331,7 @@ static void cscal_kernel_16_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
From 8c9a6356eaba102124147856422b9a0570daeb55 Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Thu, 17 Jan 2019 09:40:25 +0100
Subject: [PATCH 12/18] Remove stray comma
---
kernel/x86_64/cscal_microk_steamroller-2.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/kernel/x86_64/cscal_microk_steamroller-2.c b/kernel/x86_64/cscal_microk_steamroller-2.c
index fbeb857e2..e8073d485 100644
--- a/kernel/x86_64/cscal_microk_steamroller-2.c
+++ b/kernel/x86_64/cscal_microk_steamroller-2.c
@@ -118,7 +118,7 @@ static void cscal_kernel_16( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
@@ -210,7 +210,7 @@ static void cscal_kernel_16_zero_r( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
@@ -287,7 +287,7 @@ static void cscal_kernel_16_zero_i( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
@@ -332,7 +332,7 @@ static void cscal_kernel_16_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
From ebe8882eb23e88d410f824d8d6a113f0fca94a3b Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Thu, 17 Jan 2019 09:41:27 +0100
Subject: [PATCH 13/18] Remove stray comma
---
kernel/x86_64/dscal_microk_bulldozer-2.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/kernel/x86_64/dscal_microk_bulldozer-2.c b/kernel/x86_64/dscal_microk_bulldozer-2.c
index 71d3a9846..096662781 100644
--- a/kernel/x86_64/dscal_microk_bulldozer-2.c
+++ b/kernel/x86_64/dscal_microk_bulldozer-2.c
@@ -123,7 +123,7 @@ static void dscal_kernel_8( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n1), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha), // 2
"r" (n2) // 3
@@ -189,7 +189,7 @@ static void dscal_kernel_8_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n1), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha), // 2
"r" (n2) // 3
From fd3e2c862286019589530ece0a61be6d86a01e92 Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Thu, 17 Jan 2019 09:42:12 +0100
Subject: [PATCH 14/18] Remove stray comma
---
kernel/x86_64/dscal_microk_sandy-2.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/kernel/x86_64/dscal_microk_sandy-2.c b/kernel/x86_64/dscal_microk_sandy-2.c
index 0f187ba88..9982b8e58 100644
--- a/kernel/x86_64/dscal_microk_sandy-2.c
+++ b/kernel/x86_64/dscal_microk_sandy-2.c
@@ -123,7 +123,7 @@ static void dscal_kernel_8( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n1), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha), // 2
"r" (n2) // 3
@@ -189,7 +189,7 @@ static void dscal_kernel_8_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n1), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha), // 2
"r" (n2) // 3
From 45339034256043b4405fd6330f918cbed3660ac4 Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Thu, 17 Jan 2019 09:43:14 +0100
Subject: [PATCH 15/18] Remove stray comma
---
kernel/x86_64/dscal_microk_haswell-2.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/kernel/x86_64/dscal_microk_haswell-2.c b/kernel/x86_64/dscal_microk_haswell-2.c
index 90790cfdc..77ed59a4e 100644
--- a/kernel/x86_64/dscal_microk_haswell-2.c
+++ b/kernel/x86_64/dscal_microk_haswell-2.c
@@ -123,7 +123,7 @@ static void dscal_kernel_8( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n1), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha), // 2
"r" (n2) // 3
@@ -189,7 +189,7 @@ static void dscal_kernel_8_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n1), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha), // 2
"r" (n2) // 3
From 3b0b5ce0f69a45753b126d8bd96a48de2f882a4c Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Thu, 17 Jan 2019 09:46:05 +0100
Subject: [PATCH 16/18] Remove stray comma
---
kernel/x86_64/zscal_microk_bulldozer-2.c | 16 ++++++++--------
1 file changed, 8 insertions(+), 8 deletions(-)
diff --git a/kernel/x86_64/zscal_microk_bulldozer-2.c b/kernel/x86_64/zscal_microk_bulldozer-2.c
index 1ce59d2c7..5e733ffda 100644
--- a/kernel/x86_64/zscal_microk_bulldozer-2.c
+++ b/kernel/x86_64/zscal_microk_bulldozer-2.c
@@ -117,7 +117,7 @@ static void zscal_kernel_8( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
@@ -209,7 +209,7 @@ static void zscal_kernel_8_zero_r( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
@@ -285,9 +285,9 @@ static void zscal_kernel_8_zero_i( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
:
- :
- "r" (n), // 0
- "r" (x), // 1
+ "+r" (n), // 0
+ "+r" (x) // 1
+ :
"r" (alpha) // 2
: "cc", //"%0", "%1",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
@@ -329,10 +329,10 @@ static void zscal_kernel_8_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
"vzeroupper \n\t"
+ :
+ "+r" (n), // 0
+ "+r" (x) // 1
:
- :
- "r" (n), // 0
- "r" (x), // 1
"r" (alpha) // 2
: "cc", //"%0", "%1",
"%xmm0", "%xmm1", "%xmm2", "%xmm3",
From c17d2f61c2387b5a6cfab22d964d70afcce69b23 Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Thu, 17 Jan 2019 09:47:12 +0100
Subject: [PATCH 17/18] Remove stray comma
---
kernel/x86_64/zscal_microk_haswell-2.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/kernel/x86_64/zscal_microk_haswell-2.c b/kernel/x86_64/zscal_microk_haswell-2.c
index 534370959..8c8f5b75c 100644
--- a/kernel/x86_64/zscal_microk_haswell-2.c
+++ b/kernel/x86_64/zscal_microk_haswell-2.c
@@ -117,7 +117,7 @@ static void zscal_kernel_8( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
@@ -209,7 +209,7 @@ static void zscal_kernel_8_zero_r( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
@@ -286,7 +286,7 @@ static void zscal_kernel_8_zero_i( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
@@ -331,7 +331,7 @@ static void zscal_kernel_8_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
From ccb2b2175751037b5625b4ec3c60ddca26a04394 Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Thu, 17 Jan 2019 09:48:40 +0100
Subject: [PATCH 18/18] Remove stray comma
---
kernel/x86_64/zscal_microk_steamroller-2.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/kernel/x86_64/zscal_microk_steamroller-2.c b/kernel/x86_64/zscal_microk_steamroller-2.c
index 4b489d9f3..c9267ee0c 100644
--- a/kernel/x86_64/zscal_microk_steamroller-2.c
+++ b/kernel/x86_64/zscal_microk_steamroller-2.c
@@ -118,7 +118,7 @@ static void zscal_kernel_8( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
@@ -210,7 +210,7 @@ static void zscal_kernel_8_zero_r( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
@@ -287,7 +287,7 @@ static void zscal_kernel_8_zero_i( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",
@@ -332,7 +332,7 @@ static void zscal_kernel_8_zero( BLASLONG n, FLOAT *alpha, FLOAT *x)
:
"+r" (n), // 0
- "+r" (x), // 1
+ "+r" (x) // 1
:
"r" (alpha) // 2
: "cc",

99
1967.patch Normal file
View File

@ -0,0 +1,99 @@
From 7ff08e4b06e2c643829b566a4f2c1daba25b1029 Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Thu, 17 Jan 2019 00:04:44 +0100
Subject: [PATCH 1/4] Tag arguments 0 and 1 as both input and output
---
kernel/x86_64/dger_microk_sandy-2.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/kernel/x86_64/dger_microk_sandy-2.c b/kernel/x86_64/dger_microk_sandy-2.c
index 2bf966a5f..944d4c6f1 100644
--- a/kernel/x86_64/dger_microk_sandy-2.c
+++ b/kernel/x86_64/dger_microk_sandy-2.c
@@ -105,9 +105,9 @@ static void dger_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha)
"vzeroupper \n\t"
:
- :
- "r" (i), // 0
- "r" (n), // 1
+ "+r" (i), // 0
+ "+r" (n), // 1
+ :
"r" (x), // 2
"r" (y), // 3
"r" (alpha) // 4
From 003583675d31ce5ddabfede7fc0f93cfbac51e5f Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Thu, 17 Jan 2019 00:05:47 +0100
Subject: [PATCH 2/4] Tag arguments 0 and 1 as both input and output
---
kernel/x86_64/sger_microk_sandy-2.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/kernel/x86_64/sger_microk_sandy-2.c b/kernel/x86_64/sger_microk_sandy-2.c
index 79180b991..d38fdd551 100644
--- a/kernel/x86_64/sger_microk_sandy-2.c
+++ b/kernel/x86_64/sger_microk_sandy-2.c
@@ -105,9 +105,9 @@ static void sger_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha)
"vzeroupper \n\t"
:
- :
- "r" (i), // 0
- "r" (n), // 1
+ "+r" (i), // 0
+ "+r" (n), // 1
+ :
"r" (x), // 2
"r" (y), // 3
"r" (alpha) // 4
From 78aeb19e4613104c1ae8ea1c67022451dcfed7e6 Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Thu, 17 Jan 2019 09:34:12 +0100
Subject: [PATCH 3/4] Remove stray comma
---
kernel/x86_64/sger_microk_sandy-2.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/kernel/x86_64/sger_microk_sandy-2.c b/kernel/x86_64/sger_microk_sandy-2.c
index d38fdd551..14f13475b 100644
--- a/kernel/x86_64/sger_microk_sandy-2.c
+++ b/kernel/x86_64/sger_microk_sandy-2.c
@@ -106,7 +106,7 @@ static void sger_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha)
:
"+r" (i), // 0
- "+r" (n), // 1
+ "+r" (n) // 1
:
"r" (x), // 2
"r" (y), // 3
From d3e7e25bfb73e16bdbf89ee07d0ab584339be2a0 Mon Sep 17 00:00:00 2001
From: Martin Kroeker <martin@ruby.chemie.uni-freiburg.de>
Date: Thu, 17 Jan 2019 09:35:56 +0100
Subject: [PATCH 4/4] Remove stray comma
---
kernel/x86_64/dger_microk_sandy-2.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/kernel/x86_64/dger_microk_sandy-2.c b/kernel/x86_64/dger_microk_sandy-2.c
index 944d4c6f1..e8494500f 100644
--- a/kernel/x86_64/dger_microk_sandy-2.c
+++ b/kernel/x86_64/dger_microk_sandy-2.c
@@ -106,7 +106,7 @@ static void dger_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha)
:
"+r" (i), // 0
- "+r" (n), // 1
+ "+r" (n) // 1
:
"r" (x), // 2
"r" (y), // 3

View File

@ -15,7 +15,7 @@
Name: openblas
Version: 0.3.5
Release: 4%{?dist}
Release: 5%{?dist}
Summary: An optimized BLAS library based on GotoBLAS2
License: BSD
URL: https://github.com/xianyi/OpenBLAS/
@ -37,6 +37,9 @@ Patch13: https://patch-diff.githubusercontent.com/raw/xianyi/OpenBLAS/pul
Patch14: https://patch-diff.githubusercontent.com/raw/xianyi/OpenBLAS/pull/2023.patch
Patch15: https://patch-diff.githubusercontent.com/raw/xianyi/OpenBLAS/pull/2024.patch
Patch16: https://patch-diff.githubusercontent.com/raw/xianyi/OpenBLAS/pull/2028.patch
Patch17: https://patch-diff.githubusercontent.com/raw/xianyi/OpenBLAS/pull/1965.patch
Patch18: https://patch-diff.githubusercontent.com/raw/xianyi/OpenBLAS/pull/1966.patch
Patch19: https://patch-diff.githubusercontent.com/raw/xianyi/OpenBLAS/pull/1967.patch
BuildRequires: gcc
BuildRequires: gcc-gfortran
@ -255,6 +258,9 @@ cd OpenBLAS-%{version}
%patch14 -p1
%patch15 -p1
%patch16 -p1
%patch17 -p1
%patch18 -p1
%patch19 -p1
# Fix source permissions
find -name \*.f -exec chmod 644 {} \;
@ -691,6 +697,9 @@ rm -rf %{buildroot}%{_libdir}/pkgconfig
%endif
%changelog
* Tue Feb 26 2019 Susi Lehtola <jussilehtola@fedoraproject.org> - 0.3.5-5
- Even more assembly kernel patches.
* Mon Feb 25 2019 Susi Lehtola <jussilehtola@fedoraproject.org> - 0.3.5-4
- Another assembly kernel patch.