Skip to content
Closed
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions kernel/arm64/dznrm2_thunderx2t99_fast.c
Original file line number Diff line number Diff line change
Expand Up @@ -155,7 +155,10 @@ static double nrm2_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x)
" cmp "J", xzr \n"
" beq .Lnrm2_kernel_F1 \n"

/* https://github.com/llvm/llvm-project/issues/149547 */
#if !(defined(__clang__) && defined(OS_WINDOWS))
" .align 5 \n"
#endif
".Lnrm2_kernel_F: \n"
" "KERNEL_F" \n"
" subs "J", "J", #1 \n"
Expand Down
3 changes: 3 additions & 0 deletions kernel/arm64/scnrm2_thunderx2t99.c
Original file line number Diff line number Diff line change
Expand Up @@ -238,7 +238,10 @@ static double nrm2_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x)
" cmp "J", xzr \n"
" beq 5f //nrm2_kernel_S_BEGIN \n"

/* https://github.com/llvm/llvm-project/issues/149547 */
#if !(defined(__clang__) && defined(OS_WINDOWS))
" .align 5 \n"
#endif
"2: //nrm2_kernel_F: \n"
" "KERNEL_F" \n"
" subs "J", "J", #1 \n"
Expand Down
3 changes: 3 additions & 0 deletions kernel/arm64/zdot_thunderx2t99.c
Original file line number Diff line number Diff line change
Expand Up @@ -236,7 +236,10 @@ static void zdot_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x, FLOAT *y, BLASLON
" cmp "J", xzr \n"
" beq 3f //dot_kernel_F1 \n"

/* https://github.com/llvm/llvm-project/issues/149547 */
#if !(defined(__clang__) && defined(OS_WINDOWS))
" .align 5 \n"
#endif
"2: //dot_kernel_F: \n"
" "KERNEL_F" \n"
" subs "J", "J", #1 \n"
Expand Down
3 changes: 3 additions & 0 deletions kernel/arm64/zsum_thunderx2t99.c
Original file line number Diff line number Diff line change
Expand Up @@ -136,7 +136,10 @@ static FLOAT zasum_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x)
" cmp "J", xzr \n"
" beq 3f //asum_kernel_F1 \n"

/* https://github.com/llvm/llvm-project/issues/149547 */
#if !(defined(__clang__) && defined(OS_WINDOWS))
".align 5 \n"
#endif
"2: //asum_kernel_F16: \n"
" "KERNEL_F16" \n"
" subs "J", "J", #1 \n"
Expand Down
Loading