diff --git a/kernel/x86_64/caxpy_microk_bulldozer-2.c b/kernel/x86_64/caxpy_microk_bulldozer-2.c index 33bda0943..ca2209340 100644 --- a/kernel/x86_64/caxpy_microk_bulldozer-2.c +++ b/kernel/x86_64/caxpy_microk_bulldozer-2.c @@ -114,9 +114,9 @@ static void caxpy_kernel_8( 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 @@ -180,10 +180,10 @@ static void caxpy_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha), // 4 diff --git a/kernel/x86_64/caxpy_microk_haswell-2.c b/kernel/x86_64/caxpy_microk_haswell-2.c index 00e2e6a42..b605ea34c 100644 --- a/kernel/x86_64/caxpy_microk_haswell-2.c +++ b/kernel/x86_64/caxpy_microk_haswell-2.c @@ -112,9 +112,9 @@ static void caxpy_kernel_8( 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 diff --git a/kernel/x86_64/caxpy_microk_sandy-2.c b/kernel/x86_64/caxpy_microk_sandy-2.c index a798fd977..72d37afed 100644 --- a/kernel/x86_64/caxpy_microk_sandy-2.c +++ b/kernel/x86_64/caxpy_microk_sandy-2.c @@ -95,10 +95,10 @@ static void caxpy_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha), // 4 diff --git a/kernel/x86_64/caxpy_microk_steamroller-2.c b/kernel/x86_64/caxpy_microk_steamroller-2.c index 87370b032..7ca7af070 100644 --- a/kernel/x86_64/caxpy_microk_steamroller-2.c +++ b/kernel/x86_64/caxpy_microk_steamroller-2.c @@ -113,10 +113,10 @@ static void caxpy_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha), // 4 @@ -181,9 +181,9 @@ static void caxpy_kernel_8( 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 diff --git a/kernel/x86_64/cdot_microk_bulldozer-2.c b/kernel/x86_64/cdot_microk_bulldozer-2.c index f587aa036..118655913 100644 --- a/kernel/x86_64/cdot_microk_bulldozer-2.c +++ b/kernel/x86_64/cdot_microk_bulldozer-2.c @@ -97,9 +97,9 @@ static void cdot_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vzeroupper \n\t" : - : - "r" (i), // 0 - "r" (n), // 1 + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 @@ -175,10 +175,10 @@ static void cdot_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovups %%xmm4, 16(%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/cdot_microk_haswell-2.c b/kernel/x86_64/cdot_microk_haswell-2.c index fe195a63b..8b9d6d104 100644 --- a/kernel/x86_64/cdot_microk_haswell-2.c +++ b/kernel/x86_64/cdot_microk_haswell-2.c @@ -98,9 +98,9 @@ static void cdot_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vzeroupper \n\t" : - : - "r" (i), // 0 - "r" (n), // 1 + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/cdot_microk_sandy-2.c b/kernel/x86_64/cdot_microk_sandy-2.c index 01816917d..fe142c38f 100644 --- a/kernel/x86_64/cdot_microk_sandy-2.c +++ b/kernel/x86_64/cdot_microk_sandy-2.c @@ -105,10 +105,10 @@ static void cdot_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovups %%xmm4, 16(%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/cdot_microk_steamroller-2.c b/kernel/x86_64/cdot_microk_steamroller-2.c index 76a3aa0eb..7350b21c9 100644 --- a/kernel/x86_64/cdot_microk_steamroller-2.c +++ b/kernel/x86_64/cdot_microk_steamroller-2.c @@ -97,9 +97,9 @@ static void cdot_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vzeroupper \n\t" : - : - "r" (i), // 0 - "r" (n), // 1 + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 @@ -175,10 +175,10 @@ static void cdot_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovups %%xmm4, 16(%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/daxpy_microk_bulldozer-2.c b/kernel/x86_64/daxpy_microk_bulldozer-2.c index 8c520dcf1..9c1305b97 100644 --- a/kernel/x86_64/daxpy_microk_bulldozer-2.c +++ b/kernel/x86_64/daxpy_microk_bulldozer-2.c @@ -64,9 +64,9 @@ static void daxpy_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" : - : - "r" (i), // 0 - "r" (n), // 1 + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha) // 4 diff --git a/kernel/x86_64/daxpy_microk_haswell-2.c b/kernel/x86_64/daxpy_microk_haswell-2.c index bbe8b9550..f3682e6d7 100644 --- a/kernel/x86_64/daxpy_microk_haswell-2.c +++ b/kernel/x86_64/daxpy_microk_haswell-2.c @@ -59,10 +59,10 @@ static void daxpy_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha) // 4 diff --git a/kernel/x86_64/daxpy_microk_nehalem-2.c b/kernel/x86_64/daxpy_microk_nehalem-2.c index 943d893af..8feb9f26c 100644 --- a/kernel/x86_64/daxpy_microk_nehalem-2.c +++ b/kernel/x86_64/daxpy_microk_nehalem-2.c @@ -73,9 +73,9 @@ static void daxpy_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" : - : - "r" (i), // 0 - "r" (n), // 1 + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha) // 4 diff --git a/kernel/x86_64/daxpy_microk_piledriver-2.c b/kernel/x86_64/daxpy_microk_piledriver-2.c index 95eb953b4..4b83124c7 100644 --- a/kernel/x86_64/daxpy_microk_piledriver-2.c +++ b/kernel/x86_64/daxpy_microk_piledriver-2.c @@ -78,10 +78,10 @@ static void daxpy_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "subq $16, %1 \n\t" "jnz 1b \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha) // 4 @@ -140,10 +140,10 @@ static void daxpy_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "subq $16, %1 \n\t" "jnz 1b \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha) // 4 diff --git a/kernel/x86_64/daxpy_microk_sandy-2.c b/kernel/x86_64/daxpy_microk_sandy-2.c index 85e038cef..db9a45de8 100644 --- a/kernel/x86_64/daxpy_microk_sandy-2.c +++ b/kernel/x86_64/daxpy_microk_sandy-2.c @@ -99,10 +99,10 @@ static void daxpy_kernel_8( 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 diff --git a/kernel/x86_64/daxpy_microk_steamroller-2.c b/kernel/x86_64/daxpy_microk_steamroller-2.c index e40009037..8e63fcc1d 100644 --- a/kernel/x86_64/daxpy_microk_steamroller-2.c +++ b/kernel/x86_64/daxpy_microk_steamroller-2.c @@ -78,10 +78,10 @@ static void daxpy_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "subq $16, %1 \n\t" "jnz 1b \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha) // 4 @@ -140,10 +140,10 @@ static void daxpy_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "subq $16, %1 \n\t" "jnz 1b \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha) // 4 diff --git a/kernel/x86_64/ddot_microk_bulldozer-2.c b/kernel/x86_64/ddot_microk_bulldozer-2.c index 9756ee46a..5590c5b17 100644 --- a/kernel/x86_64/ddot_microk_bulldozer-2.c +++ b/kernel/x86_64/ddot_microk_bulldozer-2.c @@ -65,10 +65,10 @@ static void ddot_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovsd %%xmm4, (%4) \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/ddot_microk_haswell-2.c b/kernel/x86_64/ddot_microk_haswell-2.c index 365737363..dbb5487f7 100644 --- a/kernel/x86_64/ddot_microk_haswell-2.c +++ b/kernel/x86_64/ddot_microk_haswell-2.c @@ -77,9 +77,9 @@ static void ddot_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vzeroupper \n\t" : - : - "r" (i), // 0 - "r" (n), // 1 + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/ddot_microk_nehalem-2.c b/kernel/x86_64/ddot_microk_nehalem-2.c index fb5ec9bca..e5e234e22 100644 --- a/kernel/x86_64/ddot_microk_nehalem-2.c +++ b/kernel/x86_64/ddot_microk_nehalem-2.c @@ -75,10 +75,10 @@ static void ddot_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "movsd %%xmm4, (%4) \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/ddot_microk_piledriver-2.c b/kernel/x86_64/ddot_microk_piledriver-2.c index ac950885c..cc4bcd90a 100644 --- a/kernel/x86_64/ddot_microk_piledriver-2.c +++ b/kernel/x86_64/ddot_microk_piledriver-2.c @@ -81,10 +81,10 @@ static void ddot_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovsd %%xmm4, (%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 @@ -145,10 +145,10 @@ static void ddot_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovsd %%xmm4, (%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/ddot_microk_sandy-2.c b/kernel/x86_64/ddot_microk_sandy-2.c index 160f95604..84493ec27 100644 --- a/kernel/x86_64/ddot_microk_sandy-2.c +++ b/kernel/x86_64/ddot_microk_sandy-2.c @@ -81,10 +81,10 @@ static void ddot_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovsd %%xmm4, (%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/ddot_microk_steamroller-2.c b/kernel/x86_64/ddot_microk_steamroller-2.c index 5ce20b5de..27d5244ce 100644 --- a/kernel/x86_64/ddot_microk_steamroller-2.c +++ b/kernel/x86_64/ddot_microk_steamroller-2.c @@ -78,10 +78,10 @@ static void ddot_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovsd %%xmm4, (%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/saxpy_microk_haswell-2.c b/kernel/x86_64/saxpy_microk_haswell-2.c index 3a743d64c..7099ba4c6 100644 --- a/kernel/x86_64/saxpy_microk_haswell-2.c +++ b/kernel/x86_64/saxpy_microk_haswell-2.c @@ -59,10 +59,10 @@ static void saxpy_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha) // 4 diff --git a/kernel/x86_64/saxpy_microk_nehalem-2.c b/kernel/x86_64/saxpy_microk_nehalem-2.c index 68f68ea3a..88bbb695d 100644 --- a/kernel/x86_64/saxpy_microk_nehalem-2.c +++ b/kernel/x86_64/saxpy_microk_nehalem-2.c @@ -73,9 +73,9 @@ static void saxpy_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" : - : - "r" (i), // 0 - "r" (n), // 1 + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha) // 4 diff --git a/kernel/x86_64/saxpy_microk_piledriver-2.c b/kernel/x86_64/saxpy_microk_piledriver-2.c index 204cf8bac..5feea7f24 100644 --- a/kernel/x86_64/saxpy_microk_piledriver-2.c +++ b/kernel/x86_64/saxpy_microk_piledriver-2.c @@ -78,10 +78,10 @@ static void saxpy_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha) // 4 @@ -139,10 +139,10 @@ static void saxpy_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha) // 4 diff --git a/kernel/x86_64/saxpy_microk_sandy-2.c b/kernel/x86_64/saxpy_microk_sandy-2.c index 0a6bef046..0d448d5f8 100644 --- a/kernel/x86_64/saxpy_microk_sandy-2.c +++ b/kernel/x86_64/saxpy_microk_sandy-2.c @@ -99,10 +99,10 @@ static void saxpy_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 diff --git a/kernel/x86_64/sdot_microk_bulldozer-2.c b/kernel/x86_64/sdot_microk_bulldozer-2.c index 36e61b077..8958a33dc 100644 --- a/kernel/x86_64/sdot_microk_bulldozer-2.c +++ b/kernel/x86_64/sdot_microk_bulldozer-2.c @@ -66,10 +66,10 @@ static void sdot_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovss %%xmm4, (%4) \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/sdot_microk_haswell-2.c b/kernel/x86_64/sdot_microk_haswell-2.c index df367b61f..91dc928d3 100644 --- a/kernel/x86_64/sdot_microk_haswell-2.c +++ b/kernel/x86_64/sdot_microk_haswell-2.c @@ -79,10 +79,10 @@ static void sdot_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovss %%xmm4, (%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/sdot_microk_nehalem-2.c b/kernel/x86_64/sdot_microk_nehalem-2.c index 1a27177f5..5a715d008 100644 --- a/kernel/x86_64/sdot_microk_nehalem-2.c +++ b/kernel/x86_64/sdot_microk_nehalem-2.c @@ -75,10 +75,10 @@ static void sdot_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "movss %%xmm4, (%4) \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/sdot_microk_sandy-2.c b/kernel/x86_64/sdot_microk_sandy-2.c index ca13536f2..ae25d5a50 100644 --- a/kernel/x86_64/sdot_microk_sandy-2.c +++ b/kernel/x86_64/sdot_microk_sandy-2.c @@ -82,10 +82,10 @@ static void sdot_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovss %%xmm4, (%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/sdot_microk_steamroller-2.c b/kernel/x86_64/sdot_microk_steamroller-2.c index 6b8b2566b..bf6a5f287 100644 --- a/kernel/x86_64/sdot_microk_steamroller-2.c +++ b/kernel/x86_64/sdot_microk_steamroller-2.c @@ -80,10 +80,10 @@ static void sdot_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovss %%xmm4, (%4) \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 @@ -143,10 +143,10 @@ static void sdot_kernel_16( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovss %%xmm4, (%4) \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/zaxpy_microk_bulldozer-2.c b/kernel/x86_64/zaxpy_microk_bulldozer-2.c index 0e15761f7..15d367971 100644 --- a/kernel/x86_64/zaxpy_microk_bulldozer-2.c +++ b/kernel/x86_64/zaxpy_microk_bulldozer-2.c @@ -113,10 +113,10 @@ static void zaxpy_kernel_4( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha), // 4 @@ -180,10 +180,10 @@ static void zaxpy_kernel_4( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha), // 4 diff --git a/kernel/x86_64/zaxpy_microk_haswell-2.c b/kernel/x86_64/zaxpy_microk_haswell-2.c index 30e8b1955..89d23daf3 100644 --- a/kernel/x86_64/zaxpy_microk_haswell-2.c +++ b/kernel/x86_64/zaxpy_microk_haswell-2.c @@ -111,10 +111,10 @@ static void zaxpy_kernel_4( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha), // 4 diff --git a/kernel/x86_64/zaxpy_microk_sandy-2.c b/kernel/x86_64/zaxpy_microk_sandy-2.c index 233af143a..17b8b24f7 100644 --- a/kernel/x86_64/zaxpy_microk_sandy-2.c +++ b/kernel/x86_64/zaxpy_microk_sandy-2.c @@ -99,10 +99,10 @@ static void zaxpy_kernel_4( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha), // 4 @@ -176,10 +176,10 @@ static void zaxpy_kernel_4( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha), // 4 diff --git a/kernel/x86_64/zaxpy_microk_steamroller-2.c b/kernel/x86_64/zaxpy_microk_steamroller-2.c index 728d09213..907b1ae00 100644 --- a/kernel/x86_64/zaxpy_microk_steamroller-2.c +++ b/kernel/x86_64/zaxpy_microk_steamroller-2.c @@ -113,10 +113,10 @@ static void zaxpy_kernel_4( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha), // 4 @@ -180,10 +180,10 @@ static void zaxpy_kernel_4( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *alpha) "jnz 1b \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (alpha), // 4 diff --git a/kernel/x86_64/zdot_microk_bulldozer-2.c b/kernel/x86_64/zdot_microk_bulldozer-2.c index 30a9552d6..db9a48cce 100644 --- a/kernel/x86_64/zdot_microk_bulldozer-2.c +++ b/kernel/x86_64/zdot_microk_bulldozer-2.c @@ -96,10 +96,10 @@ static void zdot_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovups %%xmm4, 16(%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 @@ -175,10 +175,10 @@ static void zdot_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovups %%xmm4, 16(%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/zdot_microk_haswell-2.c b/kernel/x86_64/zdot_microk_haswell-2.c index 11056a3c1..9f2fc2c1d 100644 --- a/kernel/x86_64/zdot_microk_haswell-2.c +++ b/kernel/x86_64/zdot_microk_haswell-2.c @@ -101,10 +101,10 @@ static void zdot_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovups %%xmm4, 16(%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 @@ -186,10 +186,10 @@ static void zdot_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovups %%xmm4, 16(%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/zdot_microk_sandy-2.c b/kernel/x86_64/zdot_microk_sandy-2.c index 87c5b0340..33415e26e 100644 --- a/kernel/x86_64/zdot_microk_sandy-2.c +++ b/kernel/x86_64/zdot_microk_sandy-2.c @@ -107,10 +107,10 @@ if ( n < 1280 ) "vmovups %%xmm4, 16(%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 @@ -199,10 +199,10 @@ if ( n < 1280 ) "vmovups %%xmm4, 16(%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 diff --git a/kernel/x86_64/zdot_microk_steamroller-2.c b/kernel/x86_64/zdot_microk_steamroller-2.c index 325f74ae3..87138fe9a 100644 --- a/kernel/x86_64/zdot_microk_steamroller-2.c +++ b/kernel/x86_64/zdot_microk_steamroller-2.c @@ -95,10 +95,10 @@ static void zdot_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovups %%xmm4, 16(%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4 @@ -172,10 +172,10 @@ static void zdot_kernel_8( BLASLONG n, FLOAT *x, FLOAT *y, FLOAT *dot) "vmovups %%xmm4, 16(%4) \n\t" "vzeroupper \n\t" - : - : - "r" (i), // 0 - "r" (n), // 1 + : + "+r" (i), // 0 + "+r" (n) // 1 + : "r" (x), // 2 "r" (y), // 3 "r" (dot) // 4