From 02eb72ac426226566b6b9d3cffd4beaacde88672 Mon Sep 17 00:00:00 2001 From: wernsaar Date: Sun, 20 Jul 2014 10:48:41 +0200 Subject: [PATCH] bugfix in sgemv_t_microk_sandy.c --- kernel/x86_64/KERNEL.SANDYBRIDGE | 8 ++++++++ kernel/x86_64/sgemv_t_microk_sandy.c | 3 ++- 2 files changed, 10 insertions(+), 1 deletion(-) diff --git a/kernel/x86_64/KERNEL.SANDYBRIDGE b/kernel/x86_64/KERNEL.SANDYBRIDGE index 7228357ce..7d6b81d54 100644 --- a/kernel/x86_64/KERNEL.SANDYBRIDGE +++ b/kernel/x86_64/KERNEL.SANDYBRIDGE @@ -1,3 +1,11 @@ +ifdef OS_WINDOWS +#SGEMVNKERNEL = ../arm/gemv_n.c +SGEMVTKERNEL = ../arm/gemv_t.c +else +#SGEMVNKERNEL = sgemv_n_avx.c +SGEMVTKERNEL = sgemv_t_avx.c +endif + SGEMMKERNEL = sgemm_kernel_16x4_sandy.S SGEMMINCOPY = ../generic/gemm_ncopy_16.c diff --git a/kernel/x86_64/sgemv_t_microk_sandy.c b/kernel/x86_64/sgemv_t_microk_sandy.c index 1745db3a7..4ecd6d3d0 100644 --- a/kernel/x86_64/sgemv_t_microk_sandy.c +++ b/kernel/x86_64/sgemv_t_microk_sandy.c @@ -80,7 +80,8 @@ static void sgemv_kernel_16( long n, float alpha, float *a, long lda, float *x, "vhaddps %%xmm12, %%xmm12, %%xmm12\n\t" "vhaddps %%xmm12, %%xmm12, %%xmm12\n\t" - "vfmaddss (%%rdx), %%xmm12, %%xmm1, %%xmm12\n\t" + "vmulss %%xmm12, %%xmm1, %%xmm12 \n\t" + "vaddss (%%rdx), %%xmm12, %%xmm12\n\t" "vmovss %%xmm12, (%%rdx) \n\t" // store temp -> y :