dany/libwebp
1
0
Fork 0
mirror of https://github.com/webmproject/libwebp.git synced 2025-07-13 14:34:33 +02:00
Code Issues Actions Packages Projects Releases Wiki Activity

SSE2 version of simple in-loop filtering

Browse Source 
~10% faster decoding

Patch by Somnath Banerjee (somnath at google dot com)

Change-Id: I200db408272b4f61cda9d9261d2d4370a698d6c4
This commit is contained in:
Pascal Massimino
2011-06-10 15:10:18 -07:00
parent 42548da9e3
commit a11009d7fc
10 changed files with 335 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

55
src/dec/dsp.c
View File

@ -682,9 +682,62 @@ void (*VP8SimpleVFilter16i)(uint8_t*, int, int) = SimpleVFilter16i;
void (*VP8SimpleHFilter16i)(uint8_t*, int, int) = SimpleHFilter16i;
//-----------------------------------------------------------------------------
// SSE2 detection.
//
#if defined(__pic__) && defined(__i386__)
static inline void GetCPUInfo(int cpu_info[4], int info_type) {
__asm__ volatile (
"mov %%ebx, %%edi\n"
"cpuid\n"
"xchg %%edi, %%ebx\n"
: "=a"(cpu_info[0]), "=D"(cpu_info[1]), "=c"(cpu_info[2]), "=d"(cpu_info[3])
: "a"(info_type));
}
#elif defined(__i386__) || defined(__x86_64__)
static inline void GetCPUInfo(int cpu_info[4], int info_type) {
__asm__ volatile (
"cpuid\n"
: "=a"(cpu_info[0]), "=b"(cpu_info[1]), "=c"(cpu_info[2]), "=d"(cpu_info[3])
: "a"(info_type));
}
#elif defined(_MSC_VER) // Visual C++
#define GetCPUInfo __cpuid
#endif
#if defined(__i386__) || defined(__x86_64__) || defined(_MSC_VER)
static int x86CPUInfo(CPUFeature feature) {
int cpu_info[4];
GetCPUInfo(cpu_info, 1);
if (feature == kSSE2) {
return 0 != (cpu_info[3] & 0x04000000);
}
if (feature == kSSE3) {
return 0 != (cpu_info[2] & 0x00000001);
}
return 0;
}
VP8CPUInfo VP8DecGetCPUInfo = x86CPUInfo;
#else
VP8CPUInfo VP8DecGetCPUInfo = NULL;
#endif
//-----------------------------------------------------------------------------
extern void VP8DspInitSSE2(void);
void VP8DspInit(void) {
// later we'll plug some SSE2 variant here
// If defined, use CPUInfo() to overwrite some pointers with faster versions.
if (VP8DecGetCPUInfo) {
if (VP8DecGetCPUInfo(kSSE2)) {
#if defined(__SSE2__) || defined(_MSC_VER)
VP8DspInitSSE2();
#endif
}
if (VP8DecGetCPUInfo(kSSE3)) {
// later we'll plug some SSE3 variant here
}
}
}
#if defined(__cplusplus) || defined(c_plusplus)