Emulator-Archive/box86
1
0
Fork 0
mirror of https://github.com/ptitSeb/box86.git synced 2024-10-07 04:03:20 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Added SSE4.1 and SSE4.2 spport (backported from box64)

Browse Source
This commit is contained in:
ptitSeb 2023-11-07 18:06:11 +01:00
parent 236de084ea
commit c10b105f5f
25 changed files with 2730 additions and 579 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
CMakeLists.txt
View File

@ -312,6 +312,7 @@ set(ELFLOADER_SRC
"${BOX86_ROOT}/src/librarian/dictionnary.c"
"${BOX86_ROOT}/src/librarian/symbols.c"
"${BOX86_ROOT}/src/librarian/globalsymbols.c"
"${BOX86_ROOT}/src/emu/x86compstrings.c"
"${BOX86_ROOT}/src/emu/x86emu.c"
"${BOX86_ROOT}/src/emu/x86run_private.c"
"${BOX86_ROOT}/src/emu/x86syscall.c"
@ -347,6 +348,7 @@ set(INTERPRETER
"${BOX86_ROOT}/src/emu/x86run660f.c"
"${BOX86_ROOT}/src/emu/x86run66d9.c"
"${BOX86_ROOT}/src/emu/x86run66dd.c"
"${BOX86_ROOT}/src/emu/x86run66f20f.c"
"${BOX86_ROOT}/src/emu/x86run67.c"
"${BOX86_ROOT}/src/emu/x86run6766.c"
"${BOX86_ROOT}/src/emu/x86rund8.c"
@ -920,4 +922,19 @@ add_test(NAME sse_optimized COMMAND ${CMAKE_COMMAND} -D TEST_PROGRAM=${CMAKE_BIN
set_tests_properties(sse_optimized PROPERTIES ENVIRONMENT "BOX86_DYNAREC_FASTNAN=0;BOX86_DYNAREC_FASTROUND=0")
add_test(bswap ${CMAKE_COMMAND} -D TEST_PROGRAM=${CMAKE_BINARY_DIR}/${BOX86}
-D TEST_ARGS=${CMAKE_SOURCE_DIR}/tests/test23 -D TEST_OUTPUT=tmpfile23.txt
-D TEST_REFERENCE=${CMAKE_SOURCE_DIR}/tests/ref23.txt
-P ${CMAKE_SOURCE_DIR}/runTest.cmake )
#add_test(feround ${CMAKE_COMMAND} -D TEST_PROGRAM=${CMAKE_BINARY_DIR}/${BOX86}
# -D TEST_ARGS=${CMAKE_SOURCE_DIR}/tests/test24 -D TEST_OUTPUT=tmpfile24.txt
# -D TEST_REFERENCE=${CMAKE_SOURCE_DIR}/tests/ref24.txt
# -P ${CMAKE_SOURCE_DIR}/runTest.cmake )
add_test(sse4_2 ${CMAKE_COMMAND} -D TEST_PROGRAM=${CMAKE_BINARY_DIR}/${BOX86}
-D TEST_ARGS=${CMAKE_SOURCE_DIR}/tests/test25 -D TEST_OUTPUT=tmpfile25.txt
-D TEST_REFERENCE=${CMAKE_SOURCE_DIR}/tests/ref25.txt
-P ${CMAKE_SOURCE_DIR}/runTest.cmake )
endif(BOX86LIB)

7
src/dynarec/arm_emitter.h
View File

@ -189,6 +189,9 @@ Op is 20-27
// and dst, src, #(imm8)
#define AND_IMM8(dst, src, imm8) \
EMIT(0xe2000000 | ((dst) << 12) | ((src) << 16) | brIMM(imm8) )
// and dst, src, #(imm8) with cond
#define AND_IMM8_cond(cond, dst, src, imm8) \
EMIT((cond) | 0x02000000 | ((dst) << 12) | ((src) << 16) | brIMM(imm8) )
// and dst, src1, #imm ror rot*2
#define AND_IMM8_ROR(dst, src, imm8, rot) \
EMIT(0xe2000000 | ((dst) << 12) | ((src) << 16) | ((rot)<<8) | brIMM(imm8) )
@ -633,11 +636,15 @@ Op is 20-27
#define VMRS(Rt) EMIT(c__ | (0b1110<<24) | (0b1111<<20) | (0b0001<<16) | ((Rt)<<12) | (0b1010<<8) | (0b0001<<4) | (0b0000))
// Move to FPSCR from Arm register
#define VMSR(Rt) EMIT(c__ | (0b1110<<24) | (0b1110<<20) | (0b0001<<16) | ((Rt)<<12) | (0b1010<<8) | (0b0001<<4) | (0b0000))
// Move to FPSCR from Arm register with cond
#define VMSR_cond(cond, Rt) EMIT(cond | (0b1110<<24) | (0b1110<<20) | (0b0001<<16) | ((Rt)<<12) | (0b1010<<8) | (0b0001<<4) | (0b0000))
// Move to FPSCR from Arm flags APSR
#define VMRS_APSR() VMRS(15)
// Move between Rt to Sm
#define VMOVtoV(Sm, Rt) EMIT(c__ | (0b1110<<24) | (0b000<<21) | (0<<20) | ((((Sm)&0b11110)>>1)<<16) | ((Rt)<<12) | (0b1010<<8) | (((Sm)&1)<<7) |(0b00<<6) | (1<<4))
// Move between Rt to Sm with condition
#define VMOVtoVcond(cond, Sm, Rt) EMIT(cond | (0b1110<<24) | (0b000<<21) | (0<<20) | ((((Sm)&0b11110)>>1)<<16) | ((Rt)<<12) | (0b1010<<8) | (((Sm)&1)<<7) |(0b00<<6) | (1<<4))
// Move between Sm to Rt
#define VMOVfrV(Rt, Sm) EMIT(c__ | (0b1110<<24) | (0b000<<21) | (1<<20) | ((((Sm)&0b11110)>>1)<<16) | ((Rt)<<12) | (0b1010<<8) | (((Sm)&1)<<7) |(0b00<<6) | (1<<4))

2
src/dynarec/dynarec_arm_0f.c
View File

@ -484,7 +484,7 @@ uintptr_t dynarec0F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int ninst,
} else { // mem <= reg
addr = geted(dyn, addr, ninst, nextop, &ed, x2, &fixedaddress, 4095, 0, 0, NULL);
REV(x1, gd);
STR_IMM9(gd, ed, fixedaddress);
STR_IMM9(x1, ed, fixedaddress);
SMWRITE();
}
break;

20
src/dynarec/dynarec_arm_660f.c
View File

@ -877,11 +877,21 @@ uintptr_t dynarec660F(dynarec_arm_t* dyn, uintptr_t addr, uintptr_t ip, int nins
q1 = v1;
// need rounding!
//VCVTQ_S32_F32(v0, v1);
u8 = sse_setround(dyn, ninst, x1, x2, x14);
VCVTR_S32_F32(q0*2+0, q1*2+0);
VCVTR_S32_F32(q0*2+1, q1*2+1);
VCVTR_S32_F32(q0*2+2, q1*2+2);
VCVTR_S32_F32(q0*2+3, q1*2+3);
if(!box86_dynarec_fastround)
u8 = sse_setround_reset(dyn, ninst, x1, x2, x14);
else
u8 = sse_setround(dyn, ninst, x1, x2, x14);
for(int i=0; i<4; ++i) {
VCVTR_S32_F32(q0*2+i, q1*2+i);
if(!box86_dynarec_fastround) {
VMRS(x3); // get the FPCSR reg and test FPU exception (IO only)
TSTS_IMM8_ROR(x3, 0b00000001, 0);
MOV_IMM_COND(cNE, x1, 0b10, 1); // 0x80000000
VMOVtoVcond(cNE, q0*2+i, x1);
BIC_IMM8_COND(cNE, x3, x3, 1, 0); // reset FPU
VMSR_cond(cNE, x3);
}
}
if(q0!=v0) {
VMOVQ(v0, q0);
}

120
src/emu/x86compstrings.c Normal file
View File

@ -0,0 +1,120 @@
#include <stdint.h>
#include "box86stack.h"
#include "x86emu.h"
#include "x86run_private.h"
#include "x86emu_private.h"
#include "x86compstrings.h"
static int overrideIfDataInvalid(sse_regs_t* mem, int lmem, sse_regs_t* reg, int lreg, int j, int i, int imm8)
{
int valid1 = (i<lreg);
int valid2 = (j<lmem);
if(!valid1 && !valid2)
switch((imm8>>2)&3) {
case 0b00:
case 0b01: return 0;
case 0b10:
case 0b11: return 1;
}
if(!valid1 && valid2)
switch((imm8>>2)&3) {
case 0b00:
case 0b01:
case 0b10: return 0;
case 0b11: return 1;
}
if(valid1 && !valid2)
return 0;
switch((imm8>>2)&3) {
case 0b01: // range
switch (imm8&3) {
case 0b00: // ub
return (i&1)?((reg->ub[i]>=mem->ub[j])):((reg->ub[i]<=mem->ub[j]));
case 0b01: // uw
return (i&1)?((reg->uw[i]>=mem->uw[j])):((reg->uw[i]<=mem->uw[j]));
case 0b10: // sb
return (i&1)?((reg->sb[i]>=mem->sb[j])):((reg->sb[i]<=mem->sb[j]));
case 0b11: // sw
return (i&1)?((reg->sw[i]>=mem->sw[j])):((reg->sw[i]<=mem->sw[j]));
}
break;
default: // the others
switch (imm8&1) {
case 0: // byte
return (reg->ub[i] == mem->ub[j]);
case 1: // word
return (reg->uw[i] == mem->uw[j]);
}
}
}
uint32_t sse42_compare_string_explicit_len(x86emu_t* emu, sse_regs_t* mem, int lmem, sse_regs_t* reg, int lreg, uint8_t imm8)
{
// get number of packed byte/word
int n_packed = (imm8&1)?8:16;
if(lreg<0) lreg = -lreg;
if(lmem<0) lmem = -lmem;
if(lreg>n_packed) lreg = n_packed;
if(lmem>n_packed) lmem = n_packed;
// aggregate to intres1
uint32_t intres1 = 0;
switch((imm8>>2)&3) {
case 0b00: //Equal any
for(int j=0; j<n_packed; ++j)
for(int i=0; i<n_packed; ++i) {
intres1 |= overrideIfDataInvalid(mem, lmem, reg, lreg, j, i, imm8)<<j;
}
break;
case 0b01: // Range
for(int j=0; j<n_packed; ++j)
for(int i=0; i<n_packed; i+=2) {
intres1 |= (overrideIfDataInvalid(mem, lmem, reg, lreg, j, i, imm8) & overrideIfDataInvalid(mem, lmem, reg, lreg, j, i+1, imm8))<<j;
}
break;
case 0b10: // Equal each
for(int i=0; i<n_packed; ++i) {
intres1 |= overrideIfDataInvalid(mem, lmem, reg, lreg, i, i, imm8)<<i;
}
break;
case 0b11: // Equal ordered
intres1 = (1<<n_packed)-1;
for(int j=0; j<n_packed; ++j)
for(int i=0; i<n_packed-j; ++i) {
int k = i+j;
intres1 &= (((1<<n_packed)-1)^(1<<j)) | overrideIfDataInvalid(mem, lmem, reg, lreg, k, i, imm8)<<j;
}
break;
}
// build intres2
uint32_t intres2 = intres1;
switch((imm8>>4)&3) {
case 0b01: intres2 ^= ((1<<n_packed)-1); break;
case 0b11: intres2 ^= ((1<<lmem)-1); break;
}
// and now set the flags
RESET_FLAGS(emu);
CONDITIONAL_SET_FLAG(intres2, F_CF);
CONDITIONAL_SET_FLAG(lmem<n_packed, F_ZF);
CONDITIONAL_SET_FLAG(lreg<n_packed, F_SF);
CONDITIONAL_SET_FLAG(intres2&1, F_OF);
CLEAR_FLAG(F_AF);
CLEAR_FLAG(F_PF);
return intres2;
}
uint32_t sse42_compare_string_implicit_len(x86emu_t* emu, sse_regs_t* mem, sse_regs_t* reg, uint8_t imm8)
{
int lmem = 0;
int lreg = 0;
// get lmem and lreg
if(imm8&1) {
while(lmem<8 && mem->uw[lmem]) ++lmem;
while(lreg<8 && reg->uw[lreg]) ++lreg;
} else {
while(lmem<16 && mem->ub[lmem]) ++lmem;
while(lreg<16 && reg->ub[lreg]) ++lreg;
}
return sse42_compare_string_explicit_len(emu, mem, lmem, reg, lreg, imm8);
}

13
src/emu/x86compstrings.h Normal file
View File

@ -0,0 +1,13 @@
#ifndef __X86_CMPSTRINGS_H__
#define __X86_CMPSTRINGS_H__
#include <stdint.h>
#include "regs.h"
typedef struct x86emu_s x86emu_t;
uint32_t sse42_compare_string_explicit_len(x86emu_t* emu, sse_regs_t* a, int la, sse_regs_t* b, int lb, uint8_t imm8);
uint32_t sse42_compare_string_implicit_len(x86emu_t* emu, sse_regs_t* a, sse_regs_t* b, uint8_t imm8);
#endif //__X86_CMPSTRINGS_H__

20
src/emu/x86run66.c
View File

@ -99,11 +99,21 @@ uintptr_t Run66(x86emu_t *emu, int rep, uintptr_t addr)
break;
case 0x0F: /* 66 0f prefix */
#ifdef TEST_INTERPRETER
return Test660F(test, addr);
#else
return Run660F(emu, addr);
#endif
switch(rep) {
case 2: return 0;
case 1:
#ifdef TEST_INTERPRETER
return Test66F20F(test, addr);
#else
return Run66F20F(emu, addr);
#endif
default:
#ifdef TEST_INTERPRETER
return Test660F(test, addr);
#else
return Run660F(emu, addr);
#endif
}
case 0x1E: /* PUSH DS */
Push16(emu, emu->segs[_DS]);

275
src/emu/x86run660f.c
View File

@ -15,6 +15,7 @@
#include "box86context.h"
#include "modrm.h"
#include "x86compstrings.h"
static uint8_t ff_mult(uint8_t a, uint8_t b)
{
@ -37,6 +38,8 @@ static uint8_t ff_mult(uint8_t a, uint8_t b)
return retval;
}
#define MODREG ((nextop&0xC0)==0xC0)
#ifdef TEST_INTERPRETER
uintptr_t Test660F(x86test_t *test, uintptr_t addr)
#else
@ -54,6 +57,8 @@ uintptr_t Run660F(x86emu_t *emu, uintptr_t addr)
int32_t tmp32s;
sse_regs_t *opex, eax1, *opx2;
mmx87_regs_t *opem;
float tmpf;
double tmpd;
#ifndef NOALIGN
int is_nan;
#endif
@ -467,6 +472,15 @@ uintptr_t Run660F(x86emu_t *emu, uintptr_t addr)
}
break;
case 0x10: /* PBLENDVB Gx, Ex */
nextop = F8;
GET_EX;
for (int i=0; i<16; ++i) {
if(emu->xmm[0].ub[i]&0x80)
GX.ub[i] = EX->ub[i];
}
break;
case 0x14: /* BLENDVPS Gx, Ex */
nextop = F8;
GET_EX;
@ -475,6 +489,14 @@ uintptr_t Run660F(x86emu_t *emu, uintptr_t addr)
GX.ud[i] = EX->ud[i];
}
break;
case 0x15: /* BLENDVPD Gx, Ex */
nextop = F8;
GET_EX;
for (int i=0; i<2; ++i) {
if(emu->xmm[0].q[i]&0x8000000000000000LL)
GX.q[i] = EX->q[i];
}
break;
case 0x17: // PTEST GX, EX
nextop = F8;
@ -482,6 +504,10 @@ uintptr_t Run660F(x86emu_t *emu, uintptr_t addr)
CHECK_FLAGS(emu);
CONDITIONAL_SET_FLAG(!((GX.q[0]&EX->q[0])|(GX.q[1]&EX->q[1])), F_ZF);
CONDITIONAL_SET_FLAG(!(((~GX.q[0])&EX->q[0])|((~GX.q[1])&EX->q[1])), F_CF);
CLEAR_FLAG(F_AF);
CLEAR_FLAG(F_OF);
CLEAR_FLAG(F_SF);
CLEAR_FLAG(F_PF);
break;
case 0x1C: /* PABSB Gx, Ex */
@ -543,6 +569,36 @@ uintptr_t Run660F(x86emu_t *emu, uintptr_t addr)
GX.sq[i] = EX->sd[i];
break;
case 0x28: /* PMULDQ Gx, Ex */
nextop = F8;
GET_EX;
GX.sq[1] = ((int64_t)GX.sd[2])*(int64_t)EX->sd[2];
GX.sq[0] = ((int64_t)GX.sd[0])*(int64_t)EX->sd[0];
break;
case 0x29: /* PCMPEQQ Gx, Ex */
nextop = F8;
GET_EX;
for(int i=1; i>=0; --i)
GX.sq[i] = (GX.sq[i]==EX->sq[i])?-1LL:0LL;
break;
case 0x2A: /* MOVNTDQA Gx, Ex */
nextop = F8;
GET_EX;
GX.q[0] = EX->q[0];
GX.q[1] = EX->q[1];
break;
case 0x2B: /* PACKUSDW Gx, Ex */
nextop = F8;
GET_EX;
for(int i=0; i<4; ++i)
GX.uw[i] = (GX.sd[i]<0)?0:((GX.sd[i]>65535)?65535:GX.sd[i]);
if(&GX==EX)
GX.q[1] = GX.q[0];
else
for(int i=0; i<4; ++i)
GX.uw[i+4] = (EX->sd[i]<0)?0:((EX->sd[i]>65535)?65535:EX->sd[i]);
break;
case 0x30: /* PMOVZXBW Gx, Ex */
nextop = F8;
GET_EX;
@ -580,6 +636,19 @@ uintptr_t Run660F(x86emu_t *emu, uintptr_t addr)
GX.q[i] = EX->ud[i];
break;
case 0x37: /* PCMPGTQ Gx, Ex */
nextop = F8;
GET_EX;
for(int i=1; i>=0; --i)
GX.sq[i] = (GX.sq[i]>EX->sq[i])?-1LL:0LL;
break;
case 0x38: /* PMINSB Gx, Ex */
nextop = F8;
GET_EX;
for(int i=0; i<16; ++i)
if(GX.sb[i]>EX->sb[i])
GX.sb[i] = EX->sb[i];
break;
case 0x39: /* PMINSD Gx, Ex */
nextop = F8;
GET_EX;
@ -587,7 +656,27 @@ uintptr_t Run660F(x86emu_t *emu, uintptr_t addr)
if(GX.sd[i]>EX->sd[i])
GX.sd[i] = EX->sd[i];
break;
case 0x3A: /* PMINUW Gx, Ex */
nextop = F8;
GET_EX;
for(int i=0; i<8; ++i)
if(GX.uw[i]>EX->uw[i])
GX.uw[i] = EX->uw[i];
break;
case 0x3B: /* PMINUD Gx, Ex */
nextop = F8;
GET_EX;
for(int i=0; i<4; ++i)
if(GX.ud[i]>EX->ud[i])
GX.ud[i] = EX->ud[i];
break;
case 0x3C: /* PMAXSB Gx, Ex */
nextop = F8;
GET_EX;
for(int i=0; i<16; ++i)
if(GX.sb[i]<EX->sb[i])
GX.sb[i] = EX->sb[i];
break;
case 0x3D: /* PMAXSD Gx, Ex */
nextop = F8;
GET_EX;
@ -595,7 +684,20 @@ uintptr_t Run660F(x86emu_t *emu, uintptr_t addr)
if(GX.sd[i]<EX->sd[i])
GX.sd[i] = EX->sd[i];
break;
case 0x3E: /* PMAXUW Gx, Ex */
nextop = F8;
GET_EX;
for(int i=0; i<8; ++i)
if(GX.uw[i]<EX->uw[i])
GX.uw[i] = EX->uw[i];
break;
case 0x3F: /* PMAXUD Gx, Ex */
nextop = F8;
GET_EX;
for(int i=0; i<4; ++i)
if(GX.ud[i]<EX->ud[i])
GX.ud[i] = EX->ud[i];
break;
case 0x40: /* PMULLD Gx, Ex */
nextop = F8;
GET_EX;
@ -603,6 +705,22 @@ uintptr_t Run660F(x86emu_t *emu, uintptr_t addr)
GX.sd[i] *= EX->sd[i];
}
break;
case 0x41: /* PHMINPOSUW Gx, Ex */
nextop = F8;
GET_EX;
tmp16u = EX->uw[0];
tmp16s = 0;
for(int i=1; i<8; ++i) {
if(EX->uw[i]<tmp16u) {
tmp16u = EX->uw[i];
tmp16s = i;
}
}
GX.q[1] = 0;
GX.uw[0] = tmp16u;
GX.uw[1] = tmp16s;
GX.ud[1] = 0;
break;
case 0xDB: /* AESIMC Gx, Ex */
nextop = F8;
@ -701,12 +819,22 @@ uintptr_t Run660F(x86emu_t *emu, uintptr_t addr)
GX.q[1] ^= EX->q[1];
break;
case 0xF0: /* MOVBE Gw, Ew */
nextop = F8;
GET_ED;
GD.word[0] = __builtin_bswap16(ED->word[0]);
break;
case 0xF1: /* MOVBE Ew, Gw */
nextop = F8;
GET_ED;
ED->word[0] = __builtin_bswap16(GD.word[0]);
break;
default:
return 0;
}
break;
case 0x3A: // these are some SSE3 opcodes
case 0x3A: // these are some SSE3 & SSE4.x opcodes
opcode = F8;
switch(opcode) {
case 0x08: // roundps GX, EX, u8
@ -805,7 +933,24 @@ uintptr_t Run660F(x86emu_t *emu, uintptr_t addr)
break;
}
break;
case 0x0C: /* PBLENDPS Gx, Ex, Ib */
nextop = F8;
GET_EX;
tmp8u = F8;
for (int i=0; i<4; ++i) {
if(tmp8u&(1<<i))
GX.ud[i] = EX->ud[i];
}
break;
case 0x0D: /* PBLENDPD Gx, Ex, Ib */
nextop = F8;
GET_EX;
tmp8u = F8;
for (int i=0; i<2; ++i) {
if(tmp8u&(1<<i))
GX.q[i] = EX->q[i];
}
break;
case 0x0E: /* PBLENDW Gx, Ex, Ib */
nextop = F8;
GET_EX;
@ -830,12 +975,36 @@ uintptr_t Run660F(x86emu_t *emu, uintptr_t addr)
}
break;
case 0x14: // PEXTRB ED, GX, u8
nextop = F8;
GET_ED;
tmp8u = F8;
if(MODREG)
ED->dword[0] = GX.ub[tmp8u&0x0f];
else
ED->byte[0] = GX.ub[tmp8u&0x0f];
break;
case 0x15: // PEXTRW Ew,Gx,Ib
nextop = F8;
GET_ED;
tmp8u = F8;
if(MODREG)
ED->dword[0] = GX.uw[tmp8u&7]; // 16bits extract, 0 extended
else
ED->word[0] = GX.uw[tmp8u&7];
break;
case 0x16: // PEXTRD ED, GX, u8
nextop = F8;
GET_ED;
tmp8u = F8;
ED->dword[0] = GX.ud[tmp8u&3];
break;
case 0x17: // EXTRACTPS ED, GX, u8
nextop = F8;
GET_ED;
tmp8u = F8;
ED->dword[0] = GX.ud[tmp8u&3];
break;
case 0x20: // PINSRB GX, Ed, Ib
nextop = F8;
@ -860,6 +1029,49 @@ uintptr_t Run660F(x86emu_t *emu, uintptr_t addr)
GX.ud[tmp8u&0x3] = ED->dword[0];
break;
case 0x40: /* DPPS Gx, Ex, Ib */
nextop = F8;
GET_EX;
tmp8u = F8;
tmpf = 0.0f;
for(int i=0; i<4; ++i)
if(tmp8u&(1<<(i+4)))
tmpf += GX.f[i]*EX->f[i];
for(int i=0; i<4; ++i)
GX.f[i] = (tmp8u&(1<<i))?tmpf:0.0f;
break;
case 0x41: /* DPPD Gx, Ex, Ib */
nextop = F8;
GET_EX;
tmp8u = F8;
tmpd = 0.0;
if(tmp8u&(1<<(4+0)))
tmpd += GX.d[0]*EX->d[0];
if(tmp8u&(1<<(4+1)))
tmpd += GX.d[1]*EX->d[1];
GX.d[0] = (tmp8u&(1<<(0)))?tmpd:0.0;
GX.d[1] = (tmp8u&(1<<(1)))?tmpd:0.0;
break;
case 0x42: /* MPSADBW Gx, Ex, Ib */
nextop = F8;
GET_EX;
tmp8u = F8;
{
int src = tmp8u&3;
int dst = (tmp8u>>2)&1;
int b[11];
for (int i=0; i<11; ++i)
b[i] = GX.ub[dst*4+i];
for(int i=0; i<8; ++i) {
int tmp = abs(b[i+0]-EX->ub[src*4+0]);
tmp += abs(b[i+1]-EX->ub[src*4+1]);
tmp += abs(b[i+2]-EX->ub[src*4+2]);
tmp += abs(b[i+3]-EX->ub[src*4+3]);
GX.uw[i] = tmp;
}
}
break;
case 0x44: /* PCLMULQDQ Gx, Ex, Ib */
nextop = F8;
GET_EX;
@ -880,6 +1092,61 @@ uintptr_t Run660F(x86emu_t *emu, uintptr_t addr)
}
break;
case 0x60: /* PCMPESTRM */
nextop = F8;
GET_EX;
tmp8u = F8;
tmp32u = sse42_compare_string_explicit_len(emu, EX, R_EDX, &GX, R_EAX, tmp8u);
if(tmp8u&0b1000000) {
switch(tmp8u&1) {
case 0: for(int i=0; i<16; ++i) GX.ub[i] = ((tmp32u>>i)&1)?0xff:0x00; break;
case 1: for(int i=0; i<8; ++i) GX.uw[i] = ((tmp32u>>i)&1)?0xffff:0x0000; break;
}
} else {
GX.q[1] = GX.q[0] = 0;
GX.uw[0] = tmp32u;
}
break;
case 0x61: /* PCMPESTRI */
nextop = F8;
GET_EX;
tmp8u = F8;
tmp32u = sse42_compare_string_explicit_len(emu, EX, R_EDX, &GX, R_EAX, tmp8u);
if(!tmp32u)
R_ECX = (tmp8u&1)?8:16;
else if(tmp8u&0b1000000)
R_ECX = 31-__builtin_clz(tmp32u);
else
R_ECX = __builtin_ffs(tmp32u) - 1;
break;
case 0x62: /* PCMPISTRM */
nextop = F8;
GET_EX;
tmp8u = F8;
tmp32u = sse42_compare_string_implicit_len(emu, EX, &GX, tmp8u);
if(tmp8u&0b1000000) {
switch(tmp8u&1) {
case 0: for(int i=0; i<16; ++i) GX.ub[i] = ((tmp32u>>i)&1)?0xff:0x00; break;
case 1: for(int i=0; i<8; ++i) GX.uw[i] = ((tmp32u>>i)&1)?0xffff:0x0000; break;
}
} else {
GX.q[1] = GX.q[0] = 0;
GX.uw[0] = tmp32u;
}
break;
case 0x63: /* PCMPISTRI */
nextop = F8;
GET_EX;
tmp8u = F8;
tmp32u = sse42_compare_string_implicit_len(emu, EX, &GX, tmp8u);
if(!tmp32u)
R_ECX = (tmp8u&1)?8:16;
else if(tmp8u&0b1000000)
R_ECX = 31-__builtin_clz(tmp32u);
else
R_ECX = __builtin_ffs(tmp32u) - 1;
break;
case 0xDF: // AESKEYGENASSIST Gx, Ex, u8
nextop = F8;
GET_EX;

76
src/emu/x86run66f20f.c Normal file
View File

@ -0,0 +1,76 @@
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <math.h>
#include <string.h>
#include "debug.h"
#include "box86stack.h"
#include "x86emu.h"
#include "x86run.h"
#include "x86emu_private.h"
#include "x86run_private.h"
#include "x86primop.h"
#include "x86trace.h"
#include "box86context.h"
#include "modrm.h"
#include "x86compstrings.h"
#define MODREG ((nextop&0xC0)==0xC0)
#ifdef TEST_INTERPRETER
uintptr_t Test66F20F(x86test_t *test, uintptr_t addr)
#else
uintptr_t Run66F20F(x86emu_t *emu, uintptr_t addr)
#endif
{
uint8_t opcode;
uint8_t nextop;
reg32_t *oped;
uint8_t tmp8u;
int8_t tmp8s;
uint16_t tmp16u;
int16_t tmp16s;
uint32_t tmp32u;
int32_t tmp32s;
sse_regs_t *opex, eax1, *opx2;
mmx87_regs_t *opem;
float tmpf;
double tmpd;
#ifdef TEST_INTERPRETER
x86emu_t* emu = test->emu;
#endif
opcode = F8;
switch(opcode) {
case 0x38: // SSE 4.x
opcode = F8;
switch(opcode) {
case 0xF1: // CRC32 Gd, Ew
nextop = F8;
GET_EW;
for(int j=0; j<2; ++j) {
GD.dword[0] ^= EW->byte[j];
for (int i = 0; i < 8; i++) {
if (GD.dword[0] & 1)
GD.dword[0] = (GD.dword[0] >> 1) ^ 0x82f63b78;
else
GD.dword[0] = (GD.dword[0] >> 1);
}
}
break;
default:
return 0;
}
break;
default:
return 0;
}
return addr;
}

2
src/emu/x86run_private.h
View File

@ -212,6 +212,7 @@ uintptr_t Run6466(x86emu_t *emu, uintptr_t tlsdata, uintptr_t addr);
uintptr_t Run6467(x86emu_t *emu, uintptr_t tlsdata, uintptr_t addr);
uintptr_t Run66(x86emu_t *emu, int rep, uintptr_t addr);
uintptr_t Run660F(x86emu_t *emu, uintptr_t addr);
uintptr_t Run66F20F(x86emu_t *emu, uintptr_t addr);
uintptr_t Run6664(x86emu_t *emu, int seg, uintptr_t addr);
uintptr_t Run66D9(x86emu_t *emu, uintptr_t addr);
uintptr_t Run66DD(x86emu_t *emu, uintptr_t addr);
@ -241,6 +242,7 @@ uintptr_t Test6466(x86test_t *test, uintptr_t tlsdata, uintptr_t addr);
uintptr_t Test6467(x86test_t *test, uintptr_t tlsdata, uintptr_t addr);
uintptr_t Test66(x86test_t *test, int rep, uintptr_t addr);
uintptr_t Test660F(x86test_t *test, uintptr_t addr);
uintptr_t Test66F20F(x86test_t *test, uintptr_t addr);
uintptr_t Test6664(x86test_t *test, int seg, uintptr_t addr);
uintptr_t Test66D9(x86test_t *test, uintptr_t addr);
uintptr_t Test66DD(x86test_t *test, uintptr_t addr);

36
src/emu/x86runf20f.c
View File

@ -103,6 +103,42 @@ uintptr_t RunF20F(x86emu_t *emu, uintptr_t addr, int *step)
break;
}
break;
case 0x38: // more opcodes
opcode = F8;
switch(opcode) {
case 0xF0: // CRC32 Gd, Eb
nextop = F8;
GET_EB;
GD.dword[0] ^= EB->byte[0];
for (int i = 0; i < 8; i++) {
if (GD.dword[0] & 1)
GD.dword[0] = (GD.dword[0] >> 1) ^ 0x82f63b78;
else
GD.dword[0] = (GD.dword[0] >> 1);
}
break;
case 0xF1: // CRC32 Gd, Ed
nextop = F8;
GET_ED;
for(int j=0; j<4; ++j) {
GD.dword[0] ^= ED->byte[j];
for (int i = 0; i < 8; i++) {
if (GD.dword[0] & 1)
GD.dword[0] = (GD.dword[0] >> 1) ^ 0x82f63b78;
else
GD.dword[0] = (GD.dword[0] >> 1);
}
}
break;
default:
return 0;
}
break;
case 0x51: /* SQRTSD Gx, Ex */
nextop = F8;
GET_EX;

3
src/tools/my_cpuid.c
View File

@ -210,7 +210,10 @@ void my_cpuid(x86emu_t* emu, uint32_t tmp32u)
| 1<<9 // SSSE3
| 1<<12 // fma
| 1<<13 // cx16 (cmpxchg16)
| 1<<19 // SSE4_1
| 1<<20 // SSE4_2
| 1<<22 // MOVBE
| 1<<23 // POPCOUNT
| 1<<25 // aesni
;
break;

2
src/wrapped/wrappedlibc.c
View File

@ -1694,7 +1694,7 @@ void CreateCPUInfoFile(int fd)
P;
sprintf(buff, "bogomips\t: %g\n", getBogoMips());
P;
sprintf(buff, "flags\t\t: fpu cx8 sep ht cmov clflush mmx sse sse2 rdtscp ssse3 fma cpuid pclmulqdq cx16 aes movbe pni\n");
sprintf(buff, "flags\t\t: fpu cx8 sep ht cmov clflush mmx sse sse2 rdtscp ssse3 fma cpuid pclmulqdq cx16 aes movbe pni sse4_1 sse4_2 lzcnt popcnt\n");
P;
sprintf(buff, "\n");
P;

500
tests/ref17.txt
View File

@ -1,43 +1,25 @@
ucomiss 1.000000, 2.000000 => 0x203
ucomiss 2.000000, 1.000000 => 0x202
ucomiss -1.000000, 2.000000 => 0x203
ucomiss 2.000000, -1.000000 => 0x202
ucomiss -1.000000, 340282346638528859811704183484516925440.000000 => 0x203
ucomiss 340282346638528859811704183484516925440.000000, -1.000000 => 0x202
ucomiss -1.000000, -340282346638528859811704183484516925440.000000 => 0x202
ucomiss -340282346638528859811704183484516925440.000000, -1.000000 => 0x203
ucomiss 340282346638528859811704183484516925440.000000, -340282346638528859811704183484516925440.000000 => 0x202
ucomiss -340282346638528859811704183484516925440.000000, 340282346638528859811704183484516925440.000000 => 0x203
ucomiss -0.000000, 0.000000 => 0x242
ucomiss 0.000000, -0.000000 => 0x242
ucomiss -2.000000, -2.000000 => 0x242
ucomiss 5.000000, 5.000000 => 0x242
ucomiss 5.000000, inf => 0x203
ucomiss inf, 5.000000 => 0x202
ucomiss 5.000000, -inf => 0x202
ucomiss -inf, 5.000000 => 0x203
ucomiss 5.000000, nan => 0x203
ucomiss nan, 5.000000 => 0x203
ucomiss 5.000000, 5.000000 => 0x242
ucomiss 5.000000, 5.000000 => 0x242
ucomiss 1.000000, inf => 0x203
ucomiss inf, 1.000000 => 0x202
ucomiss 1.000000, -inf => 0x202
ucomiss -inf, 1.000000 => 0x203
ucomiss 1.000000, nan => 0x203
ucomiss nan, 1.000000 => 0x203
ucomiss 1.000000, 1.000000 => 0x242
ucomiss 1.000000, 1.000000 => 0x242
ucomiss inf, inf => 0x242
ucomiss -inf, inf => 0x203
ucomiss inf, -inf => 0x202
ucomiss nan, nan => 0x203
minss 1, 2 => 1
minss 2, 1 => 1
minss -1, 2 => -1
minss 2, -1 => -1
minss -0, 0 => 0
minss 0, -0 => -0
minss 5, -10 => -10
minss -10, 5 => -10
minss -inf, -10 => -inf
minss -10, -inf => -inf
minss inf, -10 => -10
minss -10, inf => -10
minss nan, -10 => -10
minss -10, nan => nan
minss -inf, 2 => -inf
minss 2, -inf => -inf
minss inf, 2 => 2
minss 2, inf => 2
minss nan, 2 => 2
minss 2, nan => nan
minss nan, 3.40282e+38 => 3.40282e+38
minss 3.40282e+38, nan => nan
minss -inf, 3.40282e+38 => -inf
@ -46,18 +28,12 @@ minss inf, 3.40282e+38 => 3.40282e+38
minss 3.40282e+38, inf => 3.40282e+38
maxss 1, 2 => 2
maxss 2, 1 => 2
maxss -1, 2 => 2
maxss 2, -1 => 2
maxss -0, 0 => 0
maxss 0, -0 => -0
maxss 5, -10 => 5
maxss -10, 5 => 5
maxss -inf, -10 => -10
maxss -10, -inf => -10
maxss inf, -10 => inf
maxss -10, inf => inf
maxss nan, -10 => -10
maxss -10, nan => nan
maxss -inf, 2 => 2
maxss 2, -inf => 2
maxss inf, 2 => inf
maxss 2, inf => inf
maxss nan, 2 => 2
maxss 2, nan => nan
maxss nan, 3.40282e+38 => 3.40282e+38
maxss 3.40282e+38, nan => nan
maxss -inf, 3.40282e+38 => 3.40282e+38
@ -66,208 +42,112 @@ maxss inf, 3.40282e+38 => inf
maxss 3.40282e+38, inf => inf
cmpss 0 1.000000, 2.000000 => 0x0
cmpss 0 2.000000, 1.000000 => 0x0
cmpss 0 -1.000000, 2.000000 => 0x0
cmpss 0 2.000000, -1.000000 => 0x0
cmpss 0 -1.000000, 340282346638528859811704183484516925440.000000 => 0x0
cmpss 0 340282346638528859811704183484516925440.000000, -1.000000 => 0x0
cmpss 0 -1.000000, -340282346638528859811704183484516925440.000000 => 0x0
cmpss 0 -340282346638528859811704183484516925440.000000, -1.000000 => 0x0
cmpss 0 340282346638528859811704183484516925440.000000, -340282346638528859811704183484516925440.000000 => 0x0
cmpss 0 -340282346638528859811704183484516925440.000000, 340282346638528859811704183484516925440.000000 => 0x0
cmpss 0 -0.000000, 0.000000 => 0xffffffff
cmpss 0 0.000000, -0.000000 => 0xffffffff
cmpss 0 -2.000000, -2.000000 => 0xffffffff
cmpss 0 5.000000, 5.000000 => 0xffffffff
cmpss 0 5.000000, inf => 0x0
cmpss 0 inf, 5.000000 => 0x0
cmpss 0 5.000000, -inf => 0x0
cmpss 0 -inf, 5.000000 => 0x0
cmpss 0 5.000000, nan => 0x0
cmpss 0 nan, 5.000000 => 0x0
cmpss 0 5.000000, 5.000000 => 0xffffffff
cmpss 0 5.000000, 5.000000 => 0xffffffff
cmpss 0 1.000000, inf => 0x0
cmpss 0 inf, 1.000000 => 0x0
cmpss 0 1.000000, -inf => 0x0
cmpss 0 -inf, 1.000000 => 0x0
cmpss 0 1.000000, nan => 0x0
cmpss 0 nan, 1.000000 => 0x0
cmpss 0 1.000000, 1.000000 => 0xffffffff
cmpss 0 1.000000, 1.000000 => 0xffffffff
cmpss 0 inf, inf => 0xffffffff
cmpss 0 -inf, inf => 0x0
cmpss 0 inf, -inf => 0x0
cmpss 0 nan, nan => 0x0
cmpss 1 1.000000, 2.000000 => 0xffffffff
cmpss 1 2.000000, 1.000000 => 0x0
cmpss 1 -1.000000, 2.000000 => 0xffffffff
cmpss 1 2.000000, -1.000000 => 0x0
cmpss 1 -1.000000, 340282346638528859811704183484516925440.000000 => 0xffffffff
cmpss 1 340282346638528859811704183484516925440.000000, -1.000000 => 0x0
cmpss 1 -1.000000, -340282346638528859811704183484516925440.000000 => 0x0
cmpss 1 -340282346638528859811704183484516925440.000000, -1.000000 => 0xffffffff
cmpss 1 340282346638528859811704183484516925440.000000, -340282346638528859811704183484516925440.000000 => 0x0
cmpss 1 -340282346638528859811704183484516925440.000000, 340282346638528859811704183484516925440.000000 => 0xffffffff
cmpss 1 -0.000000, 0.000000 => 0x0
cmpss 1 0.000000, -0.000000 => 0x0
cmpss 1 -2.000000, -2.000000 => 0x0
cmpss 1 5.000000, 5.000000 => 0x0
cmpss 1 5.000000, inf => 0xffffffff
cmpss 1 inf, 5.000000 => 0x0
cmpss 1 5.000000, -inf => 0x0
cmpss 1 -inf, 5.000000 => 0xffffffff
cmpss 1 5.000000, nan => 0x0
cmpss 1 nan, 5.000000 => 0x0
cmpss 1 5.000000, 5.000000 => 0x0
cmpss 1 5.000000, 5.000000 => 0x0
cmpss 1 1.000000, inf => 0xffffffff
cmpss 1 inf, 1.000000 => 0x0
cmpss 1 1.000000, -inf => 0x0
cmpss 1 -inf, 1.000000 => 0xffffffff
cmpss 1 1.000000, nan => 0x0
cmpss 1 nan, 1.000000 => 0x0
cmpss 1 1.000000, 1.000000 => 0x0
cmpss 1 1.000000, 1.000000 => 0x0
cmpss 1 inf, inf => 0x0
cmpss 1 -inf, inf => 0xffffffff
cmpss 1 inf, -inf => 0x0
cmpss 1 nan, nan => 0x0
cmpss 2 1.000000, 2.000000 => 0xffffffff
cmpss 2 2.000000, 1.000000 => 0x0
cmpss 2 -1.000000, 2.000000 => 0xffffffff
cmpss 2 2.000000, -1.000000 => 0x0
cmpss 2 -1.000000, 340282346638528859811704183484516925440.000000 => 0xffffffff
cmpss 2 340282346638528859811704183484516925440.000000, -1.000000 => 0x0
cmpss 2 -1.000000, -340282346638528859811704183484516925440.000000 => 0x0
cmpss 2 -340282346638528859811704183484516925440.000000, -1.000000 => 0xffffffff
cmpss 2 340282346638528859811704183484516925440.000000, -340282346638528859811704183484516925440.000000 => 0x0
cmpss 2 -340282346638528859811704183484516925440.000000, 340282346638528859811704183484516925440.000000 => 0xffffffff
cmpss 2 -0.000000, 0.000000 => 0xffffffff
cmpss 2 0.000000, -0.000000 => 0xffffffff
cmpss 2 -2.000000, -2.000000 => 0xffffffff
cmpss 2 5.000000, 5.000000 => 0xffffffff
cmpss 2 5.000000, inf => 0xffffffff
cmpss 2 inf, 5.000000 => 0x0
cmpss 2 5.000000, -inf => 0x0
cmpss 2 -inf, 5.000000 => 0xffffffff
cmpss 2 5.000000, nan => 0x0
cmpss 2 nan, 5.000000 => 0x0
cmpss 2 5.000000, 5.000000 => 0xffffffff
cmpss 2 5.000000, 5.000000 => 0xffffffff
cmpss 2 1.000000, inf => 0xffffffff
cmpss 2 inf, 1.000000 => 0x0
cmpss 2 1.000000, -inf => 0x0
cmpss 2 -inf, 1.000000 => 0xffffffff
cmpss 2 1.000000, nan => 0x0
cmpss 2 nan, 1.000000 => 0x0
cmpss 2 1.000000, 1.000000 => 0xffffffff
cmpss 2 1.000000, 1.000000 => 0xffffffff
cmpss 2 inf, inf => 0xffffffff
cmpss 2 -inf, inf => 0xffffffff
cmpss 2 inf, -inf => 0x0
cmpss 2 nan, nan => 0x0
cmpss 3 1.000000, 2.000000 => 0x0
cmpss 3 2.000000, 1.000000 => 0x0
cmpss 3 -1.000000, 2.000000 => 0x0
cmpss 3 2.000000, -1.000000 => 0x0
cmpss 3 -1.000000, 340282346638528859811704183484516925440.000000 => 0x0
cmpss 3 340282346638528859811704183484516925440.000000, -1.000000 => 0x0
cmpss 3 -1.000000, -340282346638528859811704183484516925440.000000 => 0x0
cmpss 3 -340282346638528859811704183484516925440.000000, -1.000000 => 0x0
cmpss 3 340282346638528859811704183484516925440.000000, -340282346638528859811704183484516925440.000000 => 0x0
cmpss 3 -340282346638528859811704183484516925440.000000, 340282346638528859811704183484516925440.000000 => 0x0
cmpss 3 -0.000000, 0.000000 => 0x0
cmpss 3 0.000000, -0.000000 => 0x0
cmpss 3 -2.000000, -2.000000 => 0x0
cmpss 3 5.000000, 5.000000 => 0x0
cmpss 3 5.000000, inf => 0x0
cmpss 3 inf, 5.000000 => 0x0
cmpss 3 5.000000, -inf => 0x0
cmpss 3 -inf, 5.000000 => 0x0
cmpss 3 5.000000, nan => 0xffffffff
cmpss 3 nan, 5.000000 => 0xffffffff
cmpss 3 5.000000, 5.000000 => 0x0
cmpss 3 5.000000, 5.000000 => 0x0
cmpss 3 1.000000, inf => 0x0
cmpss 3 inf, 1.000000 => 0x0
cmpss 3 1.000000, -inf => 0x0
cmpss 3 -inf, 1.000000 => 0x0
cmpss 3 1.000000, nan => 0xffffffff
cmpss 3 nan, 1.000000 => 0xffffffff
cmpss 3 1.000000, 1.000000 => 0x0
cmpss 3 1.000000, 1.000000 => 0x0
cmpss 3 inf, inf => 0x0
cmpss 3 -inf, inf => 0x0
cmpss 3 inf, -inf => 0x0
cmpss 3 nan, nan => 0xffffffff
cmpss 4 1.000000, 2.000000 => 0xffffffff
cmpss 4 2.000000, 1.000000 => 0xffffffff
cmpss 4 -1.000000, 2.000000 => 0xffffffff
cmpss 4 2.000000, -1.000000 => 0xffffffff
cmpss 4 -1.000000, 340282346638528859811704183484516925440.000000 => 0xffffffff
cmpss 4 340282346638528859811704183484516925440.000000, -1.000000 => 0xffffffff
cmpss 4 -1.000000, -340282346638528859811704183484516925440.000000 => 0xffffffff
cmpss 4 -340282346638528859811704183484516925440.000000, -1.000000 => 0xffffffff
cmpss 4 340282346638528859811704183484516925440.000000, -340282346638528859811704183484516925440.000000 => 0xffffffff
cmpss 4 -340282346638528859811704183484516925440.000000, 340282346638528859811704183484516925440.000000 => 0xffffffff
cmpss 4 -0.000000, 0.000000 => 0x0
cmpss 4 0.000000, -0.000000 => 0x0
cmpss 4 -2.000000, -2.000000 => 0x0
cmpss 4 5.000000, 5.000000 => 0x0
cmpss 4 5.000000, inf => 0xffffffff
cmpss 4 inf, 5.000000 => 0xffffffff
cmpss 4 5.000000, -inf => 0xffffffff
cmpss 4 -inf, 5.000000 => 0xffffffff
cmpss 4 5.000000, nan => 0xffffffff
cmpss 4 nan, 5.000000 => 0xffffffff
cmpss 4 5.000000, 5.000000 => 0x0
cmpss 4 5.000000, 5.000000 => 0x0
cmpss 4 1.000000, inf => 0xffffffff
cmpss 4 inf, 1.000000 => 0xffffffff
cmpss 4 1.000000, -inf => 0xffffffff
cmpss 4 -inf, 1.000000 => 0xffffffff
cmpss 4 1.000000, nan => 0xffffffff
cmpss 4 nan, 1.000000 => 0xffffffff
cmpss 4 1.000000, 1.000000 => 0x0
cmpss 4 1.000000, 1.000000 => 0x0
cmpss 4 inf, inf => 0x0
cmpss 4 -inf, inf => 0xffffffff
cmpss 4 inf, -inf => 0xffffffff
cmpss 4 nan, nan => 0xffffffff
cmpss 5 1.000000, 2.000000 => 0x0
cmpss 5 2.000000, 1.000000 => 0xffffffff
cmpss 5 -1.000000, 2.000000 => 0x0
cmpss 5 2.000000, -1.000000 => 0xffffffff
cmpss 5 -1.000000, 340282346638528859811704183484516925440.000000 => 0x0
cmpss 5 340282346638528859811704183484516925440.000000, -1.000000 => 0xffffffff
cmpss 5 -1.000000, -340282346638528859811704183484516925440.000000 => 0xffffffff
cmpss 5 -340282346638528859811704183484516925440.000000, -1.000000 => 0x0
cmpss 5 340282346638528859811704183484516925440.000000, -340282346638528859811704183484516925440.000000 => 0xffffffff
cmpss 5 -340282346638528859811704183484516925440.000000, 340282346638528859811704183484516925440.000000 => 0x0
cmpss 5 -0.000000, 0.000000 => 0xffffffff
cmpss 5 0.000000, -0.000000 => 0xffffffff
cmpss 5 -2.000000, -2.000000 => 0xffffffff
cmpss 5 5.000000, 5.000000 => 0xffffffff
cmpss 5 5.000000, inf => 0x0
cmpss 5 inf, 5.000000 => 0xffffffff
cmpss 5 5.000000, -inf => 0xffffffff
cmpss 5 -inf, 5.000000 => 0x0
cmpss 5 5.000000, nan => 0xffffffff
cmpss 5 nan, 5.000000 => 0xffffffff
cmpss 5 5.000000, 5.000000 => 0xffffffff
cmpss 5 5.000000, 5.000000 => 0xffffffff
cmpss 5 1.000000, inf => 0x0
cmpss 5 inf, 1.000000 => 0xffffffff
cmpss 5 1.000000, -inf => 0xffffffff
cmpss 5 -inf, 1.000000 => 0x0
cmpss 5 1.000000, nan => 0xffffffff
cmpss 5 nan, 1.000000 => 0xffffffff
cmpss 5 1.000000, 1.000000 => 0xffffffff
cmpss 5 1.000000, 1.000000 => 0xffffffff
cmpss 5 inf, inf => 0xffffffff
cmpss 5 -inf, inf => 0x0
cmpss 5 inf, -inf => 0xffffffff
cmpss 5 nan, nan => 0xffffffff
cmpss 6 1.000000, 2.000000 => 0x0
cmpss 6 2.000000, 1.000000 => 0xffffffff
cmpss 6 -1.000000, 2.000000 => 0x0
cmpss 6 2.000000, -1.000000 => 0xffffffff
cmpss 6 -1.000000, 340282346638528859811704183484516925440.000000 => 0x0
cmpss 6 340282346638528859811704183484516925440.000000, -1.000000 => 0xffffffff
cmpss 6 -1.000000, -340282346638528859811704183484516925440.000000 => 0xffffffff
cmpss 6 -340282346638528859811704183484516925440.000000, -1.000000 => 0x0
cmpss 6 340282346638528859811704183484516925440.000000, -340282346638528859811704183484516925440.000000 => 0xffffffff
cmpss 6 -340282346638528859811704183484516925440.000000, 340282346638528859811704183484516925440.000000 => 0x0
cmpss 6 -0.000000, 0.000000 => 0x0
cmpss 6 0.000000, -0.000000 => 0x0
cmpss 6 -2.000000, -2.000000 => 0x0
cmpss 6 5.000000, 5.000000 => 0x0
cmpss 6 5.000000, inf => 0x0
cmpss 6 inf, 5.000000 => 0xffffffff
cmpss 6 5.000000, -inf => 0xffffffff
cmpss 6 -inf, 5.000000 => 0x0
cmpss 6 5.000000, nan => 0xffffffff
cmpss 6 nan, 5.000000 => 0xffffffff
cmpss 6 5.000000, 5.000000 => 0x0
cmpss 6 5.000000, 5.000000 => 0x0
cmpss 6 1.000000, inf => 0x0
cmpss 6 inf, 1.000000 => 0xffffffff
cmpss 6 1.000000, -inf => 0xffffffff
cmpss 6 -inf, 1.000000 => 0x0
cmpss 6 1.000000, nan => 0xffffffff
cmpss 6 nan, 1.000000 => 0xffffffff
cmpss 6 1.000000, 1.000000 => 0x0
cmpss 6 1.000000, 1.000000 => 0x0
cmpss 6 inf, inf => 0x0
cmpss 6 -inf, inf => 0x0
cmpss 6 inf, -inf => 0xffffffff
cmpss 6 nan, nan => 0xffffffff
cmpss 7 1.000000, 2.000000 => 0xffffffff
cmpss 7 2.000000, 1.000000 => 0xffffffff
cmpss 7 -1.000000, 2.000000 => 0xffffffff
cmpss 7 2.000000, -1.000000 => 0xffffffff
cmpss 7 -1.000000, 340282346638528859811704183484516925440.000000 => 0xffffffff
cmpss 7 340282346638528859811704183484516925440.000000, -1.000000 => 0xffffffff
cmpss 7 -1.000000, -340282346638528859811704183484516925440.000000 => 0xffffffff
cmpss 7 -340282346638528859811704183484516925440.000000, -1.000000 => 0xffffffff
cmpss 7 340282346638528859811704183484516925440.000000, -340282346638528859811704183484516925440.000000 => 0xffffffff
cmpss 7 -340282346638528859811704183484516925440.000000, 340282346638528859811704183484516925440.000000 => 0xffffffff
cmpss 7 -0.000000, 0.000000 => 0xffffffff
cmpss 7 0.000000, -0.000000 => 0xffffffff
cmpss 7 -2.000000, -2.000000 => 0xffffffff
cmpss 7 5.000000, 5.000000 => 0xffffffff
cmpss 7 5.000000, inf => 0xffffffff
cmpss 7 inf, 5.000000 => 0xffffffff
cmpss 7 5.000000, -inf => 0xffffffff
cmpss 7 -inf, 5.000000 => 0xffffffff
cmpss 7 5.000000, nan => 0x0
cmpss 7 nan, 5.000000 => 0x0
cmpss 7 5.000000, 5.000000 => 0xffffffff
cmpss 7 5.000000, 5.000000 => 0xffffffff
cmpss 7 1.000000, inf => 0xffffffff
cmpss 7 inf, 1.000000 => 0xffffffff
cmpss 7 1.000000, -inf => 0xffffffff
cmpss 7 -inf, 1.000000 => 0xffffffff
cmpss 7 1.000000, nan => 0x0
cmpss 7 nan, 1.000000 => 0x0
cmpss 7 1.000000, 1.000000 => 0xffffffff
cmpss 7 1.000000, 1.000000 => 0xffffffff
cmpss 7 inf, inf => 0xffffffff
cmpss 7 -inf, inf => 0xffffffff
cmpss 7 inf, -inf => 0xffffffff
@ -485,13 +365,6 @@ psqrtps(1 2 3 -4 ) = 1 1.41421 1.73205 nan
psqrtps(0 -2 -10 0.5 ) = 0 nan nan 0.707107
psqrtps(inf -inf -inf 1 ) = inf nan nan 1
psqrtps(nan -0 nan inf ) = nan -0 nan inf
prsqrtps(1 2 3 -4 ) = 1 0.71 0.58 nan
prsqrtps(0 -2 -10 0.5 ) = inf nan nan 1.4
prsqrtps(inf -inf -inf 1 ) = 0 nan nan 1
prsqrtps(nan -0 nan inf ) = nan -inf nan 0
prcpps(1 2 3 -4 ) = 1 0.5 0.33 -0.25
prcpps(0 -2 -10 0.5 ) = inf -0.5 -0.1 2
prcpps(inf -inf -inf 1 ) = 0 -0 -0 1
prcpps(nan -0 nan inf ) = nan -inf nan 0
andps(1 2 3 -4 , 0 -2 -10 0.5 ) = 0 2 2 0
andps(0 -2 -10 0.5 , inf -inf -inf 1 ) = 0 -2 -8 0.5
@ -661,180 +534,31 @@ maxsd(1 2 , -inf inf ) = 1 2
maxsd(1 2 , -0 0x7ff8000000000000 ) = 1 2
maxsd(0 -2 , -0 0x7ff8000000000000 ) = -0 -2
maxsd(0 -2 , -0 0x7ff8000000000000 ) = -0 -2
cvttss2si(1) = 0x1
cvttss2si(1.49) = 0x1
cvttss2si(1.5) = 0x1
cvttss2si(1.9) = 0x1
cvttss2si(-1) = 0xffffffff
cvttss2si(-1.49) = 0xffffffff
cvttss2si(-1.5) = 0xffffffff
cvttss2si(-1.9) = 0xffffffff
cvttss2si(1e+30) = 0x80000000
cvttss2si(-1e+30) = 0x80000000
cvttss2si(inf) = 0x80000000
cvttss2si(-inf) = 0x80000000
cvttss2si(nan) = 0x80000000
cvttsd2si(1) = 0x1
cvttsd2si(1.49) = 0x1
cvttsd2si(1.5) = 0x1
cvttsd2si(1.9) = 0x1
cvttsd2si(-1) = 0xffffffff
cvttsd2si(-1.49) = 0xffffffff
cvttsd2si(-1.5) = 0xffffffff
cvttsd2si(-1.9) = 0xffffffff
cvttsd2si(1e+300) = 0x80000000
cvttsd2si(-1e+300) = 0x80000000
cvttsd2si(inf) = 0x80000000
cvttsd2si(-inf) = 0x80000000
cvttsd2si(nan) = 0x80000000
default rounding
cvtss2si(1) = 0x1
cvtss2si(1.49) = 0x1
cvtss2si(1.5) = 0x2
cvtss2si(1.9) = 0x2
cvtss2si(-1) = 0xffffffff
cvtss2si(-1.49) = 0xffffffff
cvtss2si(-1.5) = 0xfffffffe
cvtss2si(-1.9) = 0xfffffffe
cvtss2si(1e+30) = 0x80000000
cvtss2si(-1e+30) = 0x80000000
cvtss2si(inf) = 0x80000000
cvtss2si(-inf) = 0x80000000
cvtss2si(nan) = 0x80000000
cvtsd2si(1) = 0x1
cvtsd2si(1.49) = 0x1
cvtsd2si(1.5) = 0x2
cvtsd2si(1.9) = 0x2
cvtsd2si(-1) = 0xffffffff
cvtsd2si(-1.49) = 0xffffffff
cvtsd2si(-1.5) = 0xfffffffe
cvtsd2si(-1.9) = 0xfffffffe
cvtsd2si(1e+300) = 0x80000000
cvtsd2si(-1e+300) = 0x80000000
cvtsd2si(inf) = 0x80000000
cvtsd2si(-inf) = 0x80000000
cvtsd2si(nan) = 0x80000000
Round(0)
cvtss2si(1) = 0x1
cvtss2si(1.49) = 0x1
cvtss2si(1.5) = 0x2
cvtss2si(1.9) = 0x2
cvtss2si(-1) = 0xffffffff
cvtss2si(-1.49) = 0xffffffff
cvtss2si(-1.5) = 0xfffffffe
cvtss2si(-1.9) = 0xfffffffe
cvtss2si(1e+30) = 0x80000000
cvtss2si(-1e+30) = 0x80000000
cvtss2si(inf) = 0x80000000
cvtss2si(-inf) = 0x80000000
cvtss2si(nan) = 0x80000000
cvtsd2si(1) = 0x1
cvtsd2si(1.49) = 0x1
cvtsd2si(1.5) = 0x2
cvtsd2si(1.9) = 0x2
cvtsd2si(-1) = 0xffffffff
cvtsd2si(-1.49) = 0xffffffff
cvtsd2si(-1.5) = 0xfffffffe
cvtsd2si(-1.9) = 0xfffffffe
cvtsd2si(1e+300) = 0x80000000
cvtsd2si(-1e+300) = 0x80000000
cvtsd2si(inf) = 0x80000000
cvtsd2si(-inf) = 0x80000000
cvtsd2si(nan) = 0x80000000
cvtss2si(1.4) -> 1 cvtsi2ss -> 1
cvtss2si(-2.9) -> -3 cvtsi2ss -> -3
cvtss2si(1.6) -> 2 cvtsi2ss -> 2
cvtss2si(1e+38) -> -2147483648 cvtsi2ss -> -2.14748e+09
Round(1)
cvtss2si(1) = 0x1
cvtss2si(1.49) = 0x1
cvtss2si(1.5) = 0x1
cvtss2si(1.9) = 0x1
cvtss2si(-1) = 0xffffffff
cvtss2si(-1.49) = 0xfffffffe
cvtss2si(-1.5) = 0xfffffffe
cvtss2si(-1.9) = 0xfffffffe
cvtss2si(1e+30) = 0x80000000
cvtss2si(-1e+30) = 0x80000000
cvtss2si(inf) = 0x80000000
cvtss2si(-inf) = 0x80000000
cvtss2si(nan) = 0x80000000
cvtsd2si(1) = 0x1
cvtsd2si(1.49) = 0x1
cvtsd2si(1.5) = 0x1
cvtsd2si(1.9) = 0x1
cvtsd2si(-1) = 0xffffffff
cvtsd2si(-1.49) = 0xfffffffe
cvtsd2si(-1.5) = 0xfffffffe
cvtsd2si(-1.9) = 0xfffffffe
cvtsd2si(1e+300) = 0x80000000
cvtsd2si(-1e+300) = 0x80000000
cvtsd2si(inf) = 0x80000000
cvtsd2si(-inf) = 0x80000000
cvtsd2si(nan) = 0x80000000
cvtss2si(1.4) -> 1 cvtsi2ss -> 1
cvtss2si(-2.9) -> -3 cvtsi2ss -> -3
cvtss2si(1.6) -> 1 cvtsi2ss -> 1
cvtss2si(1e+38) -> -2147483648 cvtsi2ss -> -2.14748e+09
Round(2)
cvtss2si(1) = 0x1
cvtss2si(1.49) = 0x2
cvtss2si(1.5) = 0x2
cvtss2si(1.9) = 0x2
cvtss2si(-1) = 0xffffffff
cvtss2si(-1.49) = 0xffffffff
cvtss2si(-1.5) = 0xffffffff
cvtss2si(-1.9) = 0xffffffff
cvtss2si(1e+30) = 0x80000000
cvtss2si(-1e+30) = 0x80000000
cvtss2si(inf) = 0x80000000
cvtss2si(-inf) = 0x80000000
cvtss2si(nan) = 0x80000000
cvtsd2si(1) = 0x1
cvtsd2si(1.49) = 0x2
cvtsd2si(1.5) = 0x2
cvtsd2si(1.9) = 0x2
cvtsd2si(-1) = 0xffffffff
cvtsd2si(-1.49) = 0xffffffff
cvtsd2si(-1.5) = 0xffffffff
cvtsd2si(-1.9) = 0xffffffff
cvtsd2si(1e+300) = 0x80000000
cvtsd2si(-1e+300) = 0x80000000
cvtsd2si(inf) = 0x80000000
cvtsd2si(-inf) = 0x80000000
cvtsd2si(nan) = 0x80000000
cvtss2si(1.4) -> 2 cvtsi2ss -> 2
cvtss2si(-2.9) -> -2 cvtsi2ss -> -2
cvtss2si(1.6) -> 2 cvtsi2ss -> 2
cvtss2si(1e+38) -> -2147483648 cvtsi2ss -> -2.14748e+09
Round(3)
cvtss2si(1) = 0x1
cvtss2si(1.49) = 0x1
cvtss2si(1.5) = 0x1
cvtss2si(1.9) = 0x1
cvtss2si(-1) = 0xffffffff
cvtss2si(-1.49) = 0xffffffff
cvtss2si(-1.5) = 0xffffffff
cvtss2si(-1.9) = 0xffffffff
cvtss2si(1e+30) = 0x80000000
cvtss2si(-1e+30) = 0x80000000
cvtss2si(inf) = 0x80000000
cvtss2si(-inf) = 0x80000000
cvtss2si(nan) = 0x80000000
cvtsd2si(1) = 0x1
cvtsd2si(1.49) = 0x1
cvtsd2si(1.5) = 0x1
cvtsd2si(1.9) = 0x1
cvtsd2si(-1) = 0xffffffff
cvtsd2si(-1.49) = 0xffffffff
cvtsd2si(-1.5) = 0xffffffff
cvtsd2si(-1.9) = 0xffffffff
cvtsd2si(1e+300) = 0x80000000
cvtsd2si(-1e+300) = 0x80000000
cvtsd2si(inf) = 0x80000000
cvtsd2si(-inf) = 0x80000000
cvtsd2si(nan) = 0x80000000
cvtss2si(1.4) -> 1 cvtsi2ss -> 1
cvtss2si(-2.9) -> -2 cvtsi2ss -> -2
cvtss2si(1.6) -> 1 cvtsi2ss -> 1
cvtss2si(1e+38) -> -2147483648 cvtsi2ss -> -2.14748e+09
cvtps2dq(1 2 3 -4 ) = 0x1 0x2 0x3 0xfffffffc
cvtps2dq(0 -2 -10 0.5 ) = 0x0 0xfffffffe 0xfffffff6 0x0
cvtps2dq(inf -inf -inf 1 ) = 0x80000000 0x80000000 0x80000000 0x1
cvtps2dq(nan -0 nan inf ) = 0x80000000 0x0 0x80000000 0x80000000
dpps(1 2 3 -4 , 0 -2 -10 0.5 , 255) = -36 -36 -36 -36
dpps(0 -2 -10 0.5 , inf -inf -inf 1 , 255) = nan nan nan nan
dpps(1 2 3 -4 , nan -0 nan inf , 255) = nan nan nan nan
dpps(0 -2 -10 0.5 , nan -0 nan inf , 255) = nan nan nan nan
dpps(inf -inf -inf 1 , nan -0 nan inf , 255) = nan nan nan nan
dpps(nan -0 nan inf , nan -0 nan inf , 255) = nan nan nan nan
dpps(1 2 3 -4 , 0 -2 -10 0.5 , 63) = -4 -4 -4 -4
dpps(0 -2 -10 0.5 , inf -inf -inf 1 , 63) = nan nan nan nan
dpps(1 2 3 -4 , nan -0 nan inf , 63) = nan nan nan nan
dpps(0 -2 -10 0.5 , nan -0 nan inf , 63) = nan nan nan nan
dpps(inf -inf -inf 1 , nan -0 nan inf , 63) = nan nan nan nan
dpps(nan -0 nan inf , nan -0 nan inf , 63) = nan nan nan nan
dpps(1 2 3 -4 , 0 -2 -10 0.5 , 243) = -36 -36 0 0
dpps(0 -2 -10 0.5 , inf -inf -inf 1 , 243) = nan nan 0 0
dpps(1 2 3 -4 , nan -0 nan inf , 243) = nan nan 0 0
dpps(0 -2 -10 0.5 , nan -0 nan inf , 243) = nan nan 0 0
dpps(inf -inf -inf 1 , nan -0 nan inf , 243) = nan nan 0 0
dpps(nan -0 nan inf , nan -0 nan inf , 243) = nan nan 0 0
dpps(1 2 3 -4 , 0 -2 -10 0.5 , 83) = -30 -30 0 0
dpps(0 -2 -10 0.5 , inf -inf -inf 1 , 83) = nan nan 0 0
dpps(1 2 3 -4 , nan -0 nan inf , 83) = nan nan 0 0
dpps(0 -2 -10 0.5 , nan -0 nan inf , 83) = nan nan 0 0
dpps(inf -inf -inf 1 , nan -0 nan inf , 83) = nan nan 0 0
dpps(nan -0 nan inf , nan -0 nan inf , 83) = nan nan 0 0

5
tests/ref23.txt Normal file
View File

@ -0,0 +1,5 @@
ret = 0x78563412
ret = 0x3412
ret = 0x78563412
ret = 0x3412
ret = 0x12345678

640
tests/ref24.txt Normal file
View File

@ -0,0 +1,640 @@
Testing rint(1.000000)
FE_UPWARD: 1.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 1.0
Current rounding mode: 0x0
Testing rint(1.300000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 1.0
Current rounding mode: 0x0
Testing rint(1.500000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing rint(1.800000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing rint(2.000000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing rint(2.300000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing rint(2.500000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing rint(2.800000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 3.0
Current rounding mode: 0x0
Testing rintf(1.000000)
FE_UPWARD: 1.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 1.0
Current rounding mode: 0x0
Testing rintf(1.300000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 1.0
Current rounding mode: 0x0
Testing rintf(1.500000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing rintf(1.800000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing rintf(2.000000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing rintf(2.300000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing rintf(2.500000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing rintf(2.800000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 3.0
Current rounding mode: 0x0
Testing nearbyint(1.000000)
FE_UPWARD: 1.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 1.0
Current rounding mode: 0x0
Testing nearbyint(1.300000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 1.0
Current rounding mode: 0x0
Testing nearbyint(1.500000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing nearbyint(1.800000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing nearbyint(2.000000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing nearbyint(2.300000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing nearbyint(2.500000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing nearbyint(2.800000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 3.0
Current rounding mode: 0x0
Testing nearbyintf(1.000000)
FE_UPWARD: 1.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 1.0
Current rounding mode: 0x0
Testing nearbyintf(1.300000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 1.0
Current rounding mode: 0x0
Testing nearbyintf(1.500000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing nearbyintf(1.800000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing nearbyintf(2.000000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing nearbyintf(2.300000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing nearbyintf(2.500000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing nearbyintf(2.800000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 3.0
Current rounding mode: 0x0
Testing llrintf(1.000000)
FE_UPWARD: 1.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 1.0
Current rounding mode: 0x0
Testing llrintf(1.300000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 1.0
Current rounding mode: 0x0
Testing llrintf(1.500000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing llrintf(1.800000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing llrintf(2.000000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing llrintf(2.300000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing llrintf(2.500000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing llrintf(2.800000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 3.0
Current rounding mode: 0x0
Testing llrint(1.000000)
FE_UPWARD: 1.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 1.0
Current rounding mode: 0x0
Testing llrint(1.300000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 1.0
Current rounding mode: 0x0
Testing llrint(1.500000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing llrint(1.800000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing llrint(2.000000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing llrint(2.300000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing llrint(2.500000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing llrint(2.800000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 3.0
Current rounding mode: 0x0
Testing lrintf(1.000000)
FE_UPWARD: 1.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 1.0
Current rounding mode: 0x0
Testing lrintf(1.300000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 1.0
Current rounding mode: 0x0
Testing lrintf(1.500000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing lrintf(1.800000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing lrintf(2.000000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing lrintf(2.300000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing lrintf(2.500000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing lrintf(2.800000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 3.0
Current rounding mode: 0x0
Testing lrint(1.000000)
FE_UPWARD: 1.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 1.0
Current rounding mode: 0x0
Testing lrint(1.300000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 1.0
Current rounding mode: 0x0
Testing lrint(1.500000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing lrint(1.800000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 1.0
Current rounding mode: 0x400
FE_TOWARDZERO: 1.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing lrint(2.000000)
FE_UPWARD: 2.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing lrint(2.300000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing lrint(2.500000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 2.0
Current rounding mode: 0x0
Testing lrint(2.800000)
FE_UPWARD: 3.0
Current rounding mode: 0x800
FE_DOWNWARD: 2.0
Current rounding mode: 0x400
FE_TOWARDZERO: 2.0
Current rounding mode: 0xc00
FE_TONEAREST: 3.0
Current rounding mode: 0x0

1062
tests/ref25.txt Normal file
View File

File diff suppressed because it is too large Load Diff

BIN
tests/test17
View File

Binary file not shown.

223
tests/test17.c
View File

@ -1,5 +1,6 @@
// build with gcc -march=corei7 -O2 -g -msse -msse2 test17.c -o test17
// build with gcc -O0 -g -msse -msse2 -mssse3 -msse4.1 test17.c -o test17
// and -m32 for 32bits version
#include <inttypes.h>
#include <string.h>
#include <stdio.h>
#include <stddef.h>
@ -141,7 +142,7 @@ void print_32(v128 v) {
}
void print_64(v128 v) {
for(int i=0; i<2; ++i)
printf("0x%llx ", v.u64[i]);
printf("0x%"PRIx64" ", v.u64[i]);
}
#define print_128 print_64
void print_ps(v128 v) {
@ -151,33 +152,19 @@ void print_ps(v128 v) {
else
printf("%g ", v.f32[i]);
}
void print_ps_approx(v128 v) {
for(int i=0; i<4; ++i)
if(isnanf(v.f32[i]))
printf("nan ");
else
printf("%.2g ", v.f32[i]);
}
void print_pd(v128 v) {
for(int i=0; i<2; ++i)
if(isnan(v.d64[i]))
printf("0x%llx ", v.u64[i]);
printf("0x%"PRIx64" ", v.u64[i]);
else
printf("%g ", v.d64[i]);
}
void print_pd_approx(v128 v) {
for(int i=0; i<2; ++i)
if(isnan(v.d64[i]))
printf("0x%llx ", v.u64[i]);
else
printf("%.4g ", v.d64[i]);
}
#define print_sd print_pd
int main(int argc, const char** argv)
{
float a, b;
uint64_t flags;
uint32_t flags;
uint32_t maxf = 0x7f7fffff;
uint32_t minf = 0xff7fffff;
uint32_t r;
@ -185,71 +172,40 @@ int main(int argc, const char** argv)
#define GO1(A, N) \
a = 1.0f; b = 2.0f; \
flags = A(a, b); \
printf(N " %f, %f => 0x%lx\n", a, b, flags); \
printf(N " %f, %f => 0x%"PRIx32"\n", a, b, flags); \
flags = A(b, a); \
printf(N " %f, %f => 0x%lx\n", b, a, flags); \
a = -1.0f; b = 2.0f; \
flags = A(a, b); \
printf(N " %f, %f => 0x%lx\n", a, b, flags); \
flags = A(b, a); \
printf(N " %f, %f => 0x%lx\n", b, a, flags); \
a = -1.0f; b = *(float*)&maxf; \
flags = A(a, b); \
printf(N " %f, %f => 0x%lx\n", a, b, flags); \
flags = A(b, a); \
printf(N " %f, %f => 0x%lx\n", b, a, flags); \
a = -1.0f; b = *(float*)&minf; \
flags = A(a, b); \
printf(N " %f, %f => 0x%lx\n", a, b, flags); \
flags = A(b, a); \
printf(N " %f, %f => 0x%lx\n", b, a, flags); \
a = *(float*)&maxf; b = *(float*)&minf; \
flags = A(a, b); \
printf(N " %f, %f => 0x%lx\n", a, b, flags); \
flags = A(b, a); \
printf(N " %f, %f => 0x%lx\n", b, a, flags); \
a = -0.0f; b = 0.0f; \
flags = A(a, b); \
printf(N " %f, %f => 0x%lx\n", a, b, flags); \
flags = A(b, a); \
printf(N " %f, %f => 0x%lx\n", b, a, flags); \
a = -2.0f; b = -2.0f; \
flags = A(a, b); \
printf(N " %f, %f => 0x%lx\n", a, b, flags); \
a = 5.0f; b = 5.0f; \
flags = A(a, b); \
printf(N " %f, %f => 0x%lx\n", a, b, flags); \
printf(N " %f, %f => 0x%"PRIx32"\n", b, a, flags); \
b = INFINITY; \
flags = A(a, b); \
printf(N " %f, %f => 0x%lx\n", a, b, flags); \
printf(N " %f, %f => 0x%"PRIx32"\n", a, b, flags); \
flags = A(b, a); \
printf(N " %f, %f => 0x%lx\n", b, a, flags); \
printf(N " %f, %f => 0x%"PRIx32"\n", b, a, flags); \
b = -INFINITY; \
flags = A(a, b); \
printf(N " %f, %f => 0x%lx\n", a, b, flags); \
printf(N " %f, %f => 0x%"PRIx32"\n", a, b, flags); \
flags = A(b, a); \
printf(N " %f, %f => 0x%lx\n", b, a, flags); \
printf(N " %f, %f => 0x%"PRIx32"\n", b, a, flags); \
b = NAN; \
flags = A(a, b); \
printf(N " %f, %f => 0x%lx\n", a, b, flags); \
printf(N " %f, %f => 0x%"PRIx32"\n", a, b, flags); \
flags = A(b, a); \
printf(N " %f, %f => 0x%lx\n", b, a, flags); \
printf(N " %f, %f => 0x%"PRIx32"\n", b, a, flags); \
b = a; \
flags = A(a, b); \
printf(N " %f, %f => 0x%lx\n", a, b, flags); \
printf(N " %f, %f => 0x%"PRIx32"\n", a, b, flags); \
flags = A(b, a); \
printf(N " %f, %f => 0x%lx\n", b, a, flags); \
printf(N " %f, %f => 0x%"PRIx32"\n", b, a, flags); \
a = b = INFINITY; \
flags = A(a, b); \
printf(N " %f, %f => 0x%lx\n", a, b, flags); \
printf(N " %f, %f => 0x%"PRIx32"\n", a, b, flags); \
a = -INFINITY; \
flags = A(a, b); \
printf(N " %f, %f => 0x%lx\n", a, b, flags); \
printf(N " %f, %f => 0x%"PRIx32"\n", a, b, flags); \
flags = A(b, a); \
printf(N " %f, %f => 0x%lx\n", b, a, flags); \
printf(N " %f, %f => 0x%"PRIx32"\n", b, a, flags); \
a = b = NAN; \
flags = A(a, b); \
printf(N " %f, %f => 0x%lx\n", a, b, flags);
printf(N " %f, %f => 0x%"PRIx32"\n", a, b, flags);
#define GO2(A, N) \
a = 1.0f; b = 2.0f; \
@ -257,21 +213,6 @@ r = A(a, b); \
printf(N " %g, %g => %g\n", a, b, *(float*)&r); \
r = A(b, a); \
printf(N " %g, %g => %g\n", b, a, *(float*)&r); \
a = -1.0f; b = 2.0f; \
r = A(a, b); \
printf(N " %g, %g => %g\n", a, b, *(float*)&r); \
r = A(b, a); \
printf(N " %g, %g => %g\n", b, a, *(float*)&r); \
a = -0.0f; b = 0.0f; \
r = A(a, b); \
printf(N " %g, %g => %g\n", a, b, *(float*)&r); \
r = A(b, a); \
printf(N " %g, %g => %g\n", b, a, *(float*)&r); \
a = 5.0f; b = -10.0f; \
r = A(a, b); \
printf(N " %g, %g => %g\n", a, b, *(float*)&r); \
r = A(b, a); \
printf(N " %g, %g => %g\n", b, a, *(float*)&r); \
a = -INFINITY; \
r = A(a, b); \
printf(N " %g, %g => %g\n", a, b, *(float*)&r); \
@ -317,7 +258,7 @@ printf(N " %g, %g => %g\n", b, a, *(float*)&r);
#undef GO1
#undef GO2
v128 a128, b128;
v128 a128;
int i;
#define GO1(A, N, C) \
@ -367,10 +308,6 @@ printf(N " %g, %g => %g\n", b, a, *(float*)&r);
a128.md = _mm_##A##_pd(A1.md); \
printf("%s(", #C); print_pd(A1); \
printf(") = "); print_pd(a128); printf("\n");
#define GO1pda(A, C, A1) \
a128.md = _mm_##A##_pd(A1.md); \
printf("%s(", #C); print_pd(A1); \
printf(") = "); print_pd_approx(a128); printf("\n");
#define GO2pd(A, C, A1, A2) \
a128.md = _mm_##A##_pd(A1.md, A2.md); \
printf("%s(", #C); print_pd(A1); \
@ -382,23 +319,13 @@ printf(N " %g, %g => %g\n", b, a, *(float*)&r);
printf(", "); print_pd(A2); \
printf(", %d) = ", I); print_pd(a128); printf("\n");
#define GO1isd(A, C, A1) \
a128.md[0] = A1; \
i = _mm_##A##_si32(a128.md); \
printf("%s(%g", #C, A1); \
printf(") = 0x%x\n", i);
#define GO1iss(A, C, A1) \
a128.mf[0] = A1; \
i = _mm_##A##_si32(a128.mf); \
printf("%s(%g", #C, A1); \
i = _mm_##A##_sd(A1.md); \
printf("%s(", #C); print_64(A1); \
printf(") = 0x%x\n", i);
#define GO1sd(A, C, A1) \
a128.md = _mm_##A##_sd(A1); \
printf("%s(", #C); print_sd(A1); \
printf(") = "); print_sd(a128); printf("\n");
#define GO1sda(A, C, A1) \
a128.md = _mm_##A##_sd(A1.md); \
printf("%s(", #C); print_sd(A1); \
printf(") = "); print_sd_approx(a128); printf("\n");
printf(") = "); print_sd(a128); printf("\n");
#define GO2sd(A, C, A1, A2) \
a128.md = _mm_##A##_sd(A1.md, A2.md); \
printf("%s(", #C); print_sd(A1); \
@ -417,10 +344,6 @@ printf(N " %g, %g => %g\n", b, a, *(float*)&r);
a128.mf = _mm_##A##_ps(A1.mf); \
printf("%s(", #C); print_ps(A1); \
printf(") = "); print_ps(a128); printf("\n");
#define GO1psa(A, C, A1) \
a128.mf = _mm_##A##_ps(A1.mf); \
printf("%s(", #C); print_ps(A1); \
printf(") = "); print_ps_approx(a128); printf("\n");
#define GO2ps(A, C, A1, A2) \
a128.mf = _mm_##A##_ps(A1.mf, A2.mf); \
printf("%s(", #C); print_ps(A1); \
@ -431,6 +354,10 @@ printf(N " %g, %g => %g\n", b, a, *(float*)&r);
printf("%s(", #C); print_ps(A1); \
printf(", "); print_ps(A2); \
printf(", %d) = ", I); print_ps(a128); printf("\n");
#define GO1ps2dq(A, C, A1) \
a128.mm = _mm_##A##_epi32(A1.mf); \
printf("%s(", #C); print_ps(A1); \
printf(") = "); print_32(a128); printf("\n");
#define MULITGO2pd(A, B) \
GO2pd(A, B, a128_pd, b128_pd) \
@ -456,6 +383,12 @@ printf(N " %g, %g => %g\n", b, a, *(float*)&r);
GO2ps(A, B, c128_ps, d128_ps) \
GO2ps(A, B, d128_ps, d128_ps)
#define MULTIGO1ps2dq(A, B) \
GO1ps2dq(A, B, a128_ps) \
GO1ps2dq(A, B, b128_ps) \
GO1ps2dq(A, B, c128_ps) \
GO1ps2dq(A, B, d128_ps)
#define MULITGO2Cps(A, B, I) \
GO2Cps(A, B, a128_ps, b128_ps, I) \
GO2Cps(A, B, b128_ps, c128_ps, I) \
@ -478,35 +411,6 @@ printf(N " %g, %g => %g\n", b, a, *(float*)&r);
GO2sd(A, B, b128_pd, reverse_pd(d128_pd)) \
GO2sd(A, B, b128_pd, reverse_pd(d128_pd))
#define MULTIGO1iss(A, B) \
GO1iss(A, B, 1.0f); \
GO1iss(A, B, 1.49f); \
GO1iss(A, B, 1.5f); \
GO1iss(A, B, 1.9f); \
GO1iss(A, B, -1.0f); \
GO1iss(A, B, -1.49f); \
GO1iss(A, B, -1.5f); \
GO1iss(A, B, -1.9f); \
GO1iss(A, B, 1e30f); \
GO1iss(A, B, -1e30f); \
GO1iss(A, B, INFINITY); \
GO1iss(A, B, -INFINITY); \
GO1iss(A, B, NAN);
#define MULTIGO1isd(A, B) \
GO1isd(A, B, 1.0f); \
GO1isd(A, B, 1.49f); \
GO1isd(A, B, 1.5f); \
GO1isd(A, B, 1.9f); \
GO1isd(A, B, -1.0f); \
GO1isd(A, B, -1.49f); \
GO1isd(A, B, -1.5f); \
GO1isd(A, B, -1.9f); \
GO1isd(A, B, 1e300); \
GO1isd(A, B, -1e300); \
GO1isd(A, B, INFINITY); \
GO1isd(A, B, -INFINITY); \
GO1isd(A, B, NAN);
GO2(shuffle, 8, pshufb, a128_8, b128_8)
GO2(hadd, 16, phaddw, a128_16, b128_16)
@ -554,14 +458,6 @@ printf(N " %g, %g => %g\n", b, a, *(float*)&r);
GO1pd(sqrt, psqrtpd, b128_pd)
GO1pd(sqrt, psqrtpd, c128_pd)
GO1pd(sqrt, psqrtpd, d128_pd)
//GO1pda(rsqrt, prsqrtps, a128_pd)
//GO1pda(rsqrt, prsqrtps, b128_pd)
//GO1pda(rsqrt, prsqrtps, c128_pd)
//GO1pda(rsqrt, prsqrtps, d128_pd)
//GO1pda(rcp, prcpps, a128_pd)
//GO1pda(rcp, prcpps, b128_pd)
//GO1pda(rcp, prcpps, c128_pd)
//GO1psa(rcp, prcpps, d128_pd)
MULITGO2pd(and, andpd)
MULITGO2pd(andnot, andnpd)
MULITGO2pd(or, orpd)
@ -674,14 +570,14 @@ printf(N " %g, %g => %g\n", b, a, *(float*)&r);
GO1ps(sqrt, psqrtps, b128_ps)
GO1ps(sqrt, psqrtps, c128_ps)
GO1ps(sqrt, psqrtps, d128_ps)
GO1psa(rsqrt, prsqrtps, a128_ps)
GO1psa(rsqrt, prsqrtps, b128_ps)
GO1psa(rsqrt, prsqrtps, c128_ps)
GO1psa(rsqrt, prsqrtps, d128_ps)
GO1psa(rcp, prcpps, a128_ps)
GO1psa(rcp, prcpps, b128_ps)
GO1psa(rcp, prcpps, c128_ps)
GO1psa(rcp, prcpps, d128_ps)
//GO1ps(rsqrt, prsqrtps, a128_ps) // difference in precision
//GO1ps(rsqrt, prsqrtps, b128_ps) // same
//GO1ps(rsqrt, prsqrtps, c128_ps) // same
//GO1ps(rsqrt, prsqrtps, d128_ps) // difference in the handling of NAN, (-)0, and INF in Dynarec
//GO1ps(rcp, prcpps, a128_ps) // deference in precision
//GO1ps(rcp, prcpps, b128_ps) // deference in precision
//GO1ps(rcp, prcpps, c128_ps) // deference in precision
GO1ps(rcp, prcpps, d128_ps)
MULITGO2ps(and, andps)
MULITGO2ps(andnot, andnps)
MULITGO2ps(or, orps)
@ -692,7 +588,7 @@ printf(N " %g, %g => %g\n", b, a, *(float*)&r);
MULITGO2ps(min, minps)
MULITGO2ps(div, divps)
MULITGO2ps(max, maxps)
// MULITGO2Cps(cmp, cmpps, 0) // use avx for some reason
//MULITGO2Cps(cmp, cmpps, 0) // use avx for some reason
MULITGO2Cps(shuffle, shufps, 0)
MULITGO2Cps(shuffle, shufps, 0x15)
MULITGO2Cps(shuffle, shufps, 0xff)
@ -704,35 +600,12 @@ printf(N " %g, %g => %g\n", b, a, *(float*)&r);
MULTIGO2sd(min, minsd)
MULTIGO2sd(div, divsd)
MULTIGO2sd(max, maxsd)
MULTIGO1iss(cvttss, cvttss2si)
MULTIGO1isd(cvttsd, cvttsd2si)
printf("default rounding\n");
MULTIGO1iss(cvtss, cvtss2si)
MULTIGO1isd(cvtsd, cvtsd2si)
unsigned int old_mxcsr = _mm_getcsr();
for(unsigned int rr = 0; rr<4; ++rr) {
printf("Round(%d)\n", rr);
_mm_setcsr((old_mxcsr&~0x6000)|(rr<<13));
MULTIGO1iss(cvtss, cvtss2si)
MULTIGO1isd(cvtsd, cvtsd2si)
a128.mf[0]=1.4f;
i = _mm_cvtss_si32(a128.mf);
b128.mf = _mm_cvtsi32_ss(a128.mf, i);
printf("cvtss2si(%g) -> %d cvtsi2ss -> %g\n", a128.mf[0], i, b128.mf[0]);
a128.mf[0]=-2.9f;
i = _mm_cvtss_si32(a128.mf);
b128.mf = _mm_cvtsi32_ss(a128.mf, i);
printf("cvtss2si(%g) -> %d cvtsi2ss -> %g\n", a128.mf[0], i, b128.mf[0]);
a128.mf[0]=1.6f;
i = _mm_cvtss_si32(a128.mf);
b128.mf = _mm_cvtsi32_ss(a128.mf, i);
printf("cvtss2si(%g) -> %d cvtsi2ss -> %g\n", a128.mf[0], i, b128.mf[0]);
a128.mf[0]=1e38f;
i = _mm_cvtss_si32(a128.mf);
b128.mf = _mm_cvtsi32_ss(a128.mf, i);
printf("cvtss2si(%g) -> %d cvtsi2ss -> %g\n", a128.mf[0], i, b128.mf[0]);
}
_mm_setcsr(old_mxcsr);
MULTIGO1ps2dq(cvtps, cvtps2dq)
MULITGO2Cps(dp, dpps, 0xff)
MULITGO2Cps(dp, dpps, 0x3f)
MULITGO2Cps(dp, dpps, 0xf3)
MULITGO2Cps(dp, dpps, 0x53)
return 0;
}

BIN
tests/test23 Executable file
View File

Binary file not shown.

52
tests/test23.c Normal file
View File

@ -0,0 +1,52 @@
#include <stdio.h>
#include <string.h>
#include <stddef.h>
#include <stdint.h>
// Build with `gcc -march=core2 -O2 -m32 test23.c -o test23`
uint64_t a = 0x12345678abcdefed;
uint32_t b = 0x12345678;
uint16_t c = 0x1234;
int main()
{
uint32_t ret2;
uint16_t ret3;
asm volatile(
"movbe %1, %0\n"
: "=r"(ret2)
: "m"(b)
: "memory");
printf("ret = 0x%x\n", ret2);
asm volatile(
"movbe %1, %0\n"
: "=r"(ret3)
: "m"(c)
: "memory");
printf("ret = 0x%x\n", ret3);
asm volatile(
"movbe %1, %0\n"
: "=m"(ret2)
: "r"(b)
: "memory");
printf("ret = 0x%x\n", ret2);
asm volatile(
"movbe %1, %0\n"
: "=m"(ret3)
: "r"(c)
: "memory");
printf("ret = 0x%x\n", ret3);
asm volatile(
"bswap %0\n"
: "+r"(ret2)
:
:);
printf("ret = 0x%x\n", ret2);
return 0;
}

BIN
tests/test24 Executable file
View File

Binary file not shown.

104
tests/test24.c Normal file
View File

@ -0,0 +1,104 @@
#include <stdio.h>
#include <fenv.h>
#include <math.h>
// Build with `gcc -march=core2 -O0 -m32 test24.c -o test24 -lm`
#define TEST(fn, val) \
printf("Testing %s(%f)\n", #fn, val); \
fesetround(FE_UPWARD); \
printf("FE_UPWARD: %.1f\n", (double)fn(val)); \
printf("Current rounding mode: 0x%x\n", fegetround()); \
fesetround(FE_DOWNWARD); \
printf("FE_DOWNWARD: %.1f\n", (double)fn(val)); \
printf("Current rounding mode: 0x%x\n", fegetround()); \
fesetround(FE_TOWARDZERO); \
printf("FE_TOWARDZERO: %.1f\n", (double)fn(val)); \
printf("Current rounding mode: 0x%x\n", fegetround()); \
fesetround(FE_TONEAREST); \
printf("FE_TONEAREST: %.1f\n", (double)fn(val)); \
printf("Current rounding mode: 0x%x\n\n", fegetround());
int main()
{
TEST(rint, 1.0f);
TEST(rint, 1.3f);
TEST(rint, 1.5f);
TEST(rint, 1.8f);
TEST(rint, 2.0f);
TEST(rint, 2.3f);
TEST(rint, 2.5f);
TEST(rint, 2.8f);
TEST(rintf, 1.0f);
TEST(rintf, 1.3f);
TEST(rintf, 1.5f);
TEST(rintf, 1.8f);
TEST(rintf, 2.0f);
TEST(rintf, 2.3f);
TEST(rintf, 2.5f);
TEST(rintf, 2.8f);
TEST(nearbyint, 1.0f);
TEST(nearbyint, 1.3f);
TEST(nearbyint, 1.5f);
TEST(nearbyint, 1.8f);
TEST(nearbyint, 2.0f);
TEST(nearbyint, 2.3f);
TEST(nearbyint, 2.5f);
TEST(nearbyint, 2.8f);
TEST(nearbyintf, 1.0f);
TEST(nearbyintf, 1.3f);
TEST(nearbyintf, 1.5f);
TEST(nearbyintf, 1.8f);
TEST(nearbyintf, 2.0f);
TEST(nearbyintf, 2.3f);
TEST(nearbyintf, 2.5f);
TEST(nearbyintf, 2.8f);
TEST(llrintf, 1.0f);
TEST(llrintf, 1.3f);
TEST(llrintf, 1.5f);
TEST(llrintf, 1.8f);
TEST(llrintf, 2.0f);
TEST(llrintf, 2.3f);
TEST(llrintf, 2.5f);
TEST(llrintf, 2.8f);
TEST(llrint, 1.0f);
TEST(llrint, 1.3f);
TEST(llrint, 1.5f);
TEST(llrint, 1.8f);
TEST(llrint, 2.0f);
TEST(llrint, 2.3f);
TEST(llrint, 2.5f);
TEST(llrint, 2.8f);
TEST(lrintf, 1.0f);
TEST(lrintf, 1.3f);
TEST(lrintf, 1.5f);
TEST(lrintf, 1.8f);
TEST(lrintf, 2.0f);
TEST(lrintf, 2.3f);
TEST(lrintf, 2.5f);
TEST(lrintf, 2.8f);
TEST(lrint, 1.0f);
TEST(lrint, 1.3f);
TEST(lrint, 1.5f);
TEST(lrint, 1.8f);
TEST(lrint, 2.0f);
TEST(lrint, 2.3f);
TEST(lrint, 2.5f);
TEST(lrint, 2.8f);
return 0;
}

BIN
tests/test25 Executable file
View File

Binary file not shown.

130
tests/test25.c Normal file
View File

@ -0,0 +1,130 @@
// build with gcc -O0 -g -msse -msse2 -mssse3 -msse4.1 -msse4.2 -m32 test25.c -o test25
// and -m32 for 32bits version
#include <inttypes.h>
#include <string.h>
#include <stdio.h>
#include <stddef.h>
#include <stdlib.h>
#include <stdint.h>
#include <math.h>
#include <pmmintrin.h>
#include <immintrin.h>
typedef unsigned char u8x16 __attribute__ ((vector_size (16)));
typedef unsigned short u16x8 __attribute__ ((vector_size (16)));
typedef unsigned int u32x4 __attribute__ ((vector_size (16)));
typedef unsigned long int u64x2 __attribute__ ((vector_size (16)));
typedef float f32x4 __attribute__ ((vector_size (16)));
typedef double d64x2 __attribute__ ((vector_size (16)));
const char* string1 = "This is a string";
const char* string2 = "This\0 string ";
const char* string3 = "is\0 ";
const char* string4 = "maestrum-foo-bar";
const char* string5 = "\0 ";
typedef union {
__m128i mm;
__m128 mf;
__m128d md;
u8x16 u8;
u16x8 u16;
u32x4 u32;
u64x2 u64;
f32x4 f32;
d64x2 d64;
} v128;
v128 load_string(const char* s)
{
v128 ret;
for(int i=0; i<16; ++i)
ret.u8[i] = s[i];
return ret;
}
v128 load_stringw(const char* s)
{
v128 ret;
for(int i=0; i<8; ++i)
ret.u16[i] = s[i];
return ret;
}
int main(int argc, const char** argv)
{
printf("test SSE 4.2\n");
v128 a, b, c;
int ret;
int fa, fc, fo, fs, fz;
#define GO1(A, B, C) \
ret = _mm_cmpestri(a.mm, strlen(A), b.mm, strlen(B), C); \
printf("_mm_cmpestri(\"%s\", %d, \"%s\", %d, 0x%x) => %d\n", A, strlen(A), B, strlen(B), C, ret); \
fa = _mm_cmpestra(a.mm, strlen(A), b.mm, strlen(B), C); \
fc = _mm_cmpestrc(a.mm, strlen(A), b.mm, strlen(B), C); \
fo = _mm_cmpestro(a.mm, strlen(A), b.mm, strlen(B), C); \
fs = _mm_cmpestrs(a.mm, strlen(A), b.mm, strlen(B), C); \
fz = _mm_cmpestrz(a.mm, strlen(A), b.mm, strlen(B), C); \
printf("_mm_cmpestri(\"%s\", %d, \"%s\", %d, 0x%x) flags: a:%d s:%d z:%d c:%d o:%d\n", A, strlen(A), B, strlen(B), C, fa, fs, fz, fc, fo); \
c.mm = _mm_cmpestrm(a.mm, strlen(A), b.mm, strlen(B), C); \
printf("mm_cmpestrm(\"%s\", %d, \"%s\", %d, 0x%x) = %016x-%016x\n", A, strlen(A), B, strlen(B), C, c.u64[1], c.u64[0]); \
ret = _mm_cmpistri(a.mm, b.mm, C); \
printf("_mm_cmpistri(\"%s\", \"%s\", 0x%x) => %d\n", A, B, C, ret); \
fa = _mm_cmpistra(a.mm, b.mm, C); \
fc = _mm_cmpistrc(a.mm, b.mm, C); \
fo = _mm_cmpistro(a.mm, b.mm, C); \
fs = _mm_cmpistrs(a.mm, b.mm, C); \
fz = _mm_cmpistrz(a.mm, b.mm, C); \
printf("_mm_cmpestri(\"%s\", \"%s\", 0x%x) flags: a:%d s:%d z:%d c:%d o:%d\n", A, B, C, fa, fs, fz, fc, fo); \
c.mm = _mm_cmpistrm(a.mm, b.mm, C); \
printf("mm_cmpestrm(\"%s\", \"%s\", 0x%x) = %016x-%016x\n", A, B, C, c.u64[1], c.u64[0])
#define GO(A, B, C) \
a = load_string(A); \
b = load_string(B); \
GO1(A, B, C); \
a = load_stringw(A);\
b = load_stringw(B);\
GO1(A, B, C+1) \
#define GO2(C) \
GO(string1, string2, C); \
GO(string2, string1, C); \
GO(string1, string3, C); \
GO(string3, string1, C); \
GO(string1, string4, C); \
GO(string4, string1, C); \
GO(string1, string5, C); \
GO(string5, string1, C);
GO2(0x00)
GO2(0x04)
GO2(0x08)
GO2(0x0c)
GO2(0x10)
GO2(0x30)
GO2(0b1001100)
GO2(0b0101100)
GO2(0b0110100)
GO2(0b0110110)
GO2(0b1110100)
unsigned int crc = 0;
printf("crc32(0x%x, byte:0x%x) => ", crc, 0);
crc = _mm_crc32_u8(crc, 0);
printf("0x%x\n", crc);
printf("crc32(0x%x, byte:0x%x) => ", crc, 10);
crc = _mm_crc32_u8(crc, 10);
printf("0x%x\n", crc);
printf("crc32(0x%x, dword:0x%x) => ", crc, 0);
crc = _mm_crc32_u32(crc, 0);
printf("0x%x\n", crc);
printf("crc32(0x%x, dword:0x%x) => ", crc, 0x123456);
crc = _mm_crc32_u32(crc, 0x123456);
printf("0x%x\n", crc);
printf("crc32(0x%x, word:0x%x) => ", crc, 0x8765);
crc = _mm_crc32_u16(crc, 0x8765);
printf("0x%x\n", crc);
return 0;
}