diff options
| author | Nils Wallménius <nils@rockbox.org> | 2013-05-22 23:05:07 +0200 |
|---|---|---|
| committer | Timothy B. Terriberry <tterribe@xiph.org> | 2013-05-22 15:33:22 -0700 |
| commit | 70485d895487563b0558ff5c7e52fd2f3d4ee2ef (patch) | |
| tree | 0f1fd33825f2a7cbf4a5c4a05099793a339007a7 /celt/arm | |
| parent | 85ede2c6aa066da29fce5186394f46927358be3b (diff) | |
| download | opus-70485d895487563b0558ff5c7e52fd2f3d4ee2ef.tar.gz | |
Faster MULT32_32_Q31 for ARM.
Uses a C implementation with a 32*32 => 64 multiplication, which
ARM has.
Speeds up decoding of a 64 kbps test file by 0.5MHz on an ARM7TDMI
and 1.0MHz on an ARM9TDMI.
0.2% speedup on a 96 kbps enc+dec test on a Cortex A8.
Signed-off-by: Timothy B. Terriberry <tterribe@xiph.org>
Diffstat (limited to 'celt/arm')
| -rw-r--r-- | celt/arm/fixed_armv4.h | 5 |
1 files changed, 5 insertions, 0 deletions
diff --git a/celt/arm/fixed_armv4.h b/celt/arm/fixed_armv4.h index 73e4f434..bcacc343 100644 --- a/celt/arm/fixed_armv4.h +++ b/celt/arm/fixed_armv4.h @@ -68,4 +68,9 @@ static inline opus_val32 MULT16_32_Q15_armv4(opus_val16 a, opus_val32 b) #undef MAC16_32_Q15 #define MAC16_32_Q15(c, a, b) ADD32(c, MULT16_32_Q15(a, b)) + +/** 32x32 multiplication, followed by a 31-bit shift right. Results fits in 32 bits */ +#undef MULT32_32_Q31 +#define MULT32_32_Q31(a,b) (opus_val32)((((opus_int64)(a)) * ((opus_int64)(b)))>>31) + #endif |