Browse Source

Merge pull request #3384 from martin-frbg/issue3383

Modify ARMV8 kernels to leave x18 unused as it is reserved on OSX
tags/v0.3.18
Martin Kroeker GitHub 4 years ago
parent
commit
5edd88c919
No known key found for this signature in database GPG Key ID: 4AEE18F83AFDEB23
6 changed files with 11 additions and 10 deletions
  1. +3
    -4
      kernel/arm64/dgemm_tcopy_8.S
  2. +2
    -1
      kernel/arm64/dtrmm_kernel_8x4.S
  3. +1
    -1
      kernel/arm64/sgemm_tcopy_16.S
  4. +2
    -1
      kernel/arm64/strmm_kernel_16x4.S
  5. +2
    -2
      kernel/arm64/zgemm_kernel_4x4.S
  6. +1
    -1
      kernel/arm64/ztrmm_kernel_4x4.S

+ 3
- 4
kernel/arm64/dgemm_tcopy_8.S View File

@@ -50,11 +50,10 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#define B03 x16
#define B04 x17

#define I x18
#define J x19
#define I x19
#define J x20

#define TEMP1 x20
#define TEMP2 x21
#define TEMP1 x21

#define A_PREFETCH 2560
#define B_PREFETCH 256


+ 2
- 1
kernel/arm64/dtrmm_kernel_8x4.S View File

@@ -49,9 +49,10 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#define pCRow3 x15
#define pA x16
#define alpha x17
#define temp x18
//#define temp x18
#define tempOffset x19
#define tempK x20
#define temp x21

#define alpha0 d10
#define alphaV0 v10.d[0]


+ 1
- 1
kernel/arm64/sgemm_tcopy_16.S View File

@@ -30,7 +30,7 @@ All rights reserved.
#define B00 x22
#define I x18
#define I x21
#define J x19
#define TEMP1 x20


+ 2
- 1
kernel/arm64/strmm_kernel_16x4.S View File

@@ -49,9 +49,10 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#define pCRow3 x15
#define pA x16
#define alpha w17
#define temp x18
//#define temp x18
#define tempOffset x19
#define tempK x20
#define temp x21

#define alpha0 s10
#define alphaV0 v10.s[0]


+ 2
- 2
kernel/arm64/zgemm_kernel_4x4.S View File

@@ -48,8 +48,8 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#define pCRow2 x14
#define pCRow3 x15
#define pA x16
#define alphaR x17
#define alphaI x18
#define alphaR x19
#define alphaI x20

#define alpha0_R d10
#define alphaV0_R v10.d[0]


+ 1
- 1
kernel/arm64/ztrmm_kernel_4x4.S View File

@@ -49,7 +49,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#define pCRow3 x15
#define pA x16
#define alphaR x17
#define alphaI x18
#define alphaI x22
#define temp x19
#define tempOffset x20
#define tempK x21


Loading…
Cancel
Save