diff options
| author | rodri <rgl@antares-labs.eu> | 2023-11-24 15:39:06 +0000 |
|---|---|---|
| committer | rodri <rgl@antares-labs.eu> | 2023-11-24 15:39:06 +0000 |
| commit | d850c3b7f47e58556c160f9d03ea20aa52452020 (patch) | |
| tree | 20471edd63ca1c23349a1c8e340155fabedf5b27 /sse.h | |
| parent | 9404d16a4263a87559af64bfb18c91ccebaa601d (diff) | |
| download | amd64-simd-d850c3b7f47e58556c160f9d03ea20aa52452020.tar.gz amd64-simd-d850c3b7f47e58556c160f9d03ea20aa52452020.tar.bz2 amd64-simd-d850c3b7f47e58556c160f9d03ea20aa52452020.zip | |
add more avx instructions and a bench9 benchmark file.
Diffstat (limited to 'sse.h')
| -rw-r--r-- | sse.h | 11 |
1 files changed, 11 insertions, 0 deletions
@@ -41,6 +41,8 @@ BYTE $(((~r)<<7)|((~v)<<3)|((l)<<2)|(p)) #define VOP(o, m, ro, rm) BYTE $(o); \ BYTE $(((m)<<6)|((ro)<<3)|(rm)) +#define VOPi(o, m, ro, rm, i) VOP((o), (m), (ro), (rm)); \ + BYTE $(i) /* MOVLPD */ //opcode = 660F12 @@ -65,6 +67,15 @@ //imm8 = 0011 0001 #define DPPD(s, d) OP4i(0x413A, 0x3, (d), (s), 0x31) +/* VMOVAPD */ +#define VMOVUPD_128mr(off, s, d) VEX3(0,0,0,VEX_m_0F,0,0,VEX_L_128,VEX_p_66); \ + VOPi(0x10, 0x1, (d), (s), (off)) +#define VMOVAPD_128rr(s, d) VEX3(0,0,0,VEX_m_0F,0,0,VEX_L_128,VEX_p_66); \ + VOP(0x28, 0x3, (d), (s)) +/* VDPPD */ +#define VDPPD(s0, s1, d) VEX3(0,0,0,VEX_m_0F3A,0,(s0),VEX_L_128,VEX_p_66); \ + VOPi(0x41, 0x3, (d), (s1), 0x31) + /* VFMADD231SD (128 bit) */ #define VFMADD231SD(s0, s1, d) VEX3(0,0,0,VEX_m_0F38,1,(s0),VEX_L_128,VEX_p_66); \ VOP(0xB9, 0x3, (d), (s1)) |