sysutils/mingw-w64
sysutils/mingw-w64
1
0
Fork 0
mirror of https://git.code.sf.net/p/mingw-w64/mingw-w64 synced 2025-04-18 17:44:18 +03:00
Code

crt: Use aarch64 assembly variants for ARM64EC math functions.

Browse Source 
Signed-off-by: Jacek Caban <jacek@codeweavers.com>
This commit is contained in:
Billy Laws 2025-04-03 15:43:30 +02:00 committed by Jacek Caban
parent dbb60ad07c
commit cf2b16e2a3
12 changed files with 26 additions and 25 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
mingw-w64-crt/math/fma.c
View File

@ -17,7 +17,7 @@ double fma(double x, double y, double z){
return z;
}
#elif defined(_ARM64_) || defined(__aarch64__)
#elif defined(_ARM64_) || defined(__aarch64__) || defined(_ARM64EC_) || defined(__arm64ec__)
/* Use hardware FMA on ARM64. */
double fma(double x, double y, double z){

2
mingw-w64-crt/math/fmaf.c
View File

@ -17,7 +17,7 @@ float fmaf(float x, float y, float z){
return z;
}
#elif defined(_ARM64_) || defined(__aarch64__)
#elif defined(_ARM64_) || defined(__aarch64__) || defined(_ARM64EC_) || defined(__arm64ec__)
/* Use hardware FMA on ARM64. */
float fmaf(float x, float y, float z){

2
mingw-w64-crt/math/llrint.c
View File

@ -9,7 +9,7 @@
long long llrint (double x)
{
long long retval = 0ll;
#if defined(_AMD64_) || defined(__x86_64__) || defined(_X86_) || defined(__i386__)
#if (defined(_AMD64_) && !defined(_ARM64EC_)) || (defined(__x86_64__) && !defined(__arm64ec__)) || defined(_X86_) || defined(__i386__)
__asm__ __volatile__ ("fistpll %0" : "=m" (retval) : "t" (x) : "st");
#else
int mode = fegetround();

2
mingw-w64-crt/math/llrintf.c
View File

@ -9,7 +9,7 @@
long long llrintf (float x)
{
long long retval = 0ll;
#if defined(_AMD64_) || defined(__x86_64__) || defined(_X86_) || defined(__i386__)
#if (defined(_AMD64_) && !defined(_ARM64EC_)) || (defined(__x86_64__) && !defined(__arm64ec__)) || defined(_X86_) || defined(__i386__)
__asm__ __volatile__ ("fistpll %0" : "=m" (retval) : "t" (x) : "st");
#else
int mode = fegetround();

3
mingw-w64-crt/math/llrintl.c
View File

@ -9,7 +9,8 @@
long long llrintl (long double x)
{
long long retval = 0ll;
#if defined(_AMD64_) || defined(__x86_64__) || defined(_X86_) || defined(__i386__)
#if (defined(_AMD64_) && !defined(_ARM64EC_)) || (defined(__x86_64__) && !defined(__arm64ec__)) || \
defined(_X86_) || defined(__i386__)
__asm__ __volatile__ ("fistpll %0" : "=m" (retval) : "t" (x) : "st");
#else
int mode = fegetround();

14
mingw-w64-crt/math/lrint.c
View File

@ -5,14 +5,19 @@
*/
#include <math.h>
#if defined(_AMD64_) || defined(__x86_64__)
#if (defined(_AMD64_) && !defined(_ARM64EC_)) || (defined(__x86_64__) && !defined(__arm64ec__))
#include <xmmintrin.h>
#endif
long lrint (double x)
{
long retval = 0L;
#if defined(_AMD64_) || defined(__x86_64__)
#if defined(__aarch64__) || defined(_ARM64_) || defined(__arm64ec__) || defined(_ARM64EC_)
__asm__ __volatile__ (
"frintx %d1, %d1\n\t"
"fcvtzs %w0, %d1\n\t"
: "=r" (retval), "+w" (x));
#elif defined(_AMD64_) || defined(__x86_64__)
retval = _mm_cvtsd_si32(_mm_load_sd(&x));
#elif defined(_X86_) || defined(__i386__)
__asm__ __volatile__ ("fistpl %0" : "=m" (retval) : "t" (x) : "st");
@ -22,11 +27,6 @@ long lrint (double x)
"vcvtr.s32.f64 %[tmp], %[src]\n\t"
"fmrs %[dst], %[tmp]\n\t"
: [dst] "=r" (retval), [tmp] "=t" (temp) : [src] "w" (x));
#elif defined(__aarch64__) || defined(_ARM64_)
__asm__ __volatile__ (
"frintx %d1, %d1\n\t"
"fcvtzs %w0, %d1\n\t"
: "=r" (retval), "+w" (x));
#endif
return retval;
}

14
mingw-w64-crt/math/lrintf.c
View File

@ -5,14 +5,19 @@
*/
#include <math.h>
#if defined(_AMD64_) || defined(__x86_64__)
#if (defined(_AMD64_) && !defined(_ARM64EC_)) || (defined(__x86_64__) && !defined(__arm64ec__))
#include <xmmintrin.h>
#endif
long lrintf (float x)
{
long retval = 0l;
#if defined(_AMD64_) || defined(__x86_64__)
#if defined(__aarch64__) || defined(_ARM64_) || defined(__arm64ec__) || defined(_ARM64EC_)
__asm__ __volatile__ (
"frintx %s1, %s1\n\t"
"fcvtzs %w0, %s1\n\t"
: "=r" (retval), "+w" (x));
#elif defined(_AMD64_) || defined(__x86_64__)
retval = _mm_cvtss_si32(_mm_load_ss(&x));
#elif defined(_X86_) || defined(__i386__)
__asm__ __volatile__ ("fistpl %0" : "=m" (retval) : "t" (x) : "st");
@ -21,11 +26,6 @@ long lrintf (float x)
"vcvtr.s32.f32 %[src], %[src]\n\t"
"fmrs %[dst], %[src]\n\t"
: [dst] "=r" (retval), [src] "+w" (x));
#elif defined(__aarch64__) || defined(_ARM64_)
__asm__ __volatile__ (
"frintx %s1, %s1\n\t"
"fcvtzs %w0, %s1\n\t"
: "=r" (retval), "+w" (x));
#endif
return retval;
}

2
mingw-w64-crt/math/modf.c
View File

@ -12,7 +12,7 @@ modf (double value, double* iptr)
{
double int_part = 0.0;
/* truncate */
#if defined(_AMD64_) || defined(__x86_64__)
#if (defined(_AMD64_) && !defined(_ARM64EC_)) || (defined(__x86_64__) && !defined(__arm64ec__))
asm volatile ("subq $8, %%rsp\n"
"fnstcw 4(%%rsp)\n"
"movzwl 4(%%rsp), %%eax\n"

2
mingw-w64-crt/math/modff.c
View File

@ -13,7 +13,7 @@ modff (float value, float* iptr)
float int_part = 0.0F;
/* truncate */
/* truncate */
#if defined(_AMD64_) || defined(__x86_64__)
#if (defined(_AMD64_) && !defined(_ARM64EC_)) || (defined(__x86_64__) && !defined(__arm64ec__))
asm volatile ("subq $8, %%rsp\n"
"fnstcw 4(%%rsp)\n"
"movzwl 4(%%rsp), %%eax\n"

2
mingw-w64-crt/math/modfl.c
View File

@ -12,7 +12,7 @@ modfl (long double value, long double* iptr)
{
long double int_part = 0.0L;
/* truncate */
#if defined(_AMD64_) || defined(__x86_64__)
#if (defined(_AMD64_) && !defined(_ARM64EC_)) || (defined(__x86_64__) && !defined(__arm64ec__))
asm volatile ("subq $8, %%rsp\n"
"fnstcw 4(%%rsp)\n"
"movzwl 4(%%rsp), %%eax\n"

4
mingw-w64-crt/math/signbitl.c
View File

@ -9,11 +9,11 @@
int __signbitl (long double x) {
#if defined(__x86_64__) || defined(_AMD64_)
#if (defined(_AMD64_) && !defined(_ARM64EC_)) || (defined(__x86_64__) && !defined(__arm64ec__))
__mingw_ldbl_type_t ld;
ld.x = x;
return ((ld.lh.sign_exponent & 0x8000) != 0);
#elif defined(__arm__) || defined(_ARM_) || defined(__aarch64__) || defined(_ARM64_)
#elif defined(__arm__) || defined(_ARM_) || defined(__aarch64__) || defined(_ARM64_) || defined(__arm64ec__) || defined(_ARM64EC_)
return __signbit(x);
#elif defined(__i386__) || defined(_X86_)
unsigned short sw;

2
mingw-w64-crt/math/sqrt.def.h
View File

@ -77,7 +77,7 @@ __FLT_ABI (sqrt) (__FLT_TYPE x)
#else
asm volatile ("fsqrtd %[dst], %[src];\n" : [dst] "=w" (res) : [src] "w" (x));
#endif
#elif defined(__aarch64__) || defined(_ARM64_)
#elif defined(__aarch64__) || defined(_ARM64_) || defined(__arm64ec__) || defined(_ARM64EC_)
#if _NEW_COMPLEX_FLOAT
asm volatile ("fsqrt %s[dst], %s[src]\n" : [dst] "=w" (res) : [src] "w" (x));
#else