optimize avx

src/emu/x64runavxf20f38.c

@@ -8,7 +8,6 @@
 #include <signal.h>
 #include <sys/types.h>
 #include <unistd.h>
-
 #include "debug.h"
 #include "box64stack.h"
 #include "x64emu.h"
@@ -26,7 +25,6 @@
 #include "custommem.h"
 #include "../dynarec/native_lock.h"
 #endif
-
 #include "modrm.h"
 
 #ifdef TEST_INTERPRETER
@@ -35,30 +33,19 @@ uintptr_t TestAVX_F20F38(x64test_t *test, vex_t vex, uintptr_t addr, int *step)
 uintptr_t RunAVX_F20F38(x64emu_t *emu, vex_t vex, uintptr_t addr, int *step)
 #endif
 {
-    uint8_t opcode;
+    uint8_t opcode = F8;
     uint8_t nextop;
-    uint8_t tmp8u, u8;
-    int8_t tmp8s;
-    int16_t tmp16s;
-    uint16_t tmp16u;
-    int32_t tmp32s, tmp32s2;
-    uint32_t tmp32u, tmp32u2;
-    uint64_t tmp64u, tmp64u2;
-    int64_t tmp64s;
+    uint8_t u8;
+    uint32_t tmp32u;
+    uint64_t tmp64u;
     reg64_t *oped, *opgd, *opvd;
-    sse_regs_t *opex, *opgx, *opvx, eax1, eax2;
-    sse_regs_t *opey, *opgy, *opvy, eay1, eay2;
-
-
+    sse_regs_t *opex, *opgx, *opvx;
 #ifdef TEST_INTERPRETER
     x64emu_t *emu = test->emu;
 #endif
-    opcode = F8;
-
     rex_t rex = vex.rex;
 
     switch(opcode) {
-
         case 0xF5:  /* PDEP Gd, Ed, Vd */
             nextop = F8;
             GETED(0);
@@ -67,16 +54,22 @@ uintptr_t RunAVX_F20F38(x64emu_t *emu, vex_t vex, uintptr_t addr, int *step)
             if(rex.w) {
                 tmp64u = 0;
                 u8 = 0;
-                for(int i=0; i<64; ++i)
-                    if(ED->q[0]&(1LL<<i))
-                        tmp64u |= ((VD->q[0]>>(u8++))&1LL)<<i;
+                for(int i=0; i<64; ++i) {
+                    if(ED->q[0] & (1ULL << i)) {
+                        tmp64u |= ((VD->q[0] >> u8) & 1ULL) << i;
+                        ++u8;
+                    }
+                }
                 GD->q[0] = tmp64u;
             } else {
                 tmp32u = 0;
                 u8 = 0;
-                for(int i=0; i<32; ++i)
-                    if(ED->dword[0]&(1<<i))
-                        tmp32u |= ((VD->dword[0]>>(u8++))&1)<<i;
+                for(int i=0; i<32; ++i) {
+                    if(ED->dword[0] & (1U << i)) {
+                        tmp32u |= ((VD->dword[0] >> u8) & 1U) << i;
+                        ++u8;
+                    }
+                }
                 GD->q[0] = tmp32u;
             }
             break;
@@ -86,13 +79,13 @@ uintptr_t RunAVX_F20F38(x64emu_t *emu, vex_t vex, uintptr_t addr, int *step)
             GETGD;
             GETVD;
             if(rex.w) {
-                unsigned __int128 res = (unsigned __int128)R_RDX * ED->q[0];
-                VD->q[0] = res&0xFFFFFFFFFFFFFFFFLL;
-                GD->q[0] = res>>64;
+                __uint128_t res = (__uint128_t)R_RDX * ED->q[0];
+                VD->q[0] = (uint64_t)res;
+                GD->q[0] = (uint64_t)(res >> 64);
             } else {
                 tmp64u = (uint64_t)R_EDX * ED->dword[0];
-                VD->q[0] = tmp64u&0xFFFFFFFF;
-                GD->q[0] = tmp64u>>32;
+                VD->q[0] = (uint32_t)tmp64u;
+                GD->q[0] = (uint32_t)(tmp64u >> 32);
             }
             break;
         case 0xF7:  /* SHRX Gd, Ed, Vd */
@@ -108,7 +101,6 @@ uintptr_t RunAVX_F20F38(x64emu_t *emu, vex_t vex, uintptr_t addr, int *step)
                 GD->q[0] = ED->dword[0] >> u8;
             }
             break;
-
         default:
             return 0;
     }