| 1 | /* Copyright (C) 2010-2020 The RetroArch team |
| 2 | * |
| 3 | * --------------------------------------------------------------------------------------- |
| 4 | * The following license statement only applies to this file (s16_to_float_neon.S). |
| 5 | * --------------------------------------------------------------------------------------- |
| 6 | * |
| 7 | * Permission is hereby granted, free of charge, |
| 8 | * to any person obtaining a copy of this software and associated documentation files (the "Software"), |
| 9 | * to deal in the Software without restriction, including without limitation the rights to |
| 10 | * use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, |
| 11 | * and to permit persons to whom the Software is furnished to do so, subject to the following conditions: |
| 12 | * |
| 13 | * The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. |
| 14 | * |
| 15 | * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, |
| 16 | * INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
| 17 | * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. |
| 18 | * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, |
| 19 | * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, |
| 20 | * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. |
| 21 | */ |
| 22 | #if defined(__ARM_NEON__) && defined(HAVE_ARM_NEON_ASM_OPTIMIZATIONS) |
| 23 | |
| 24 | #ifndef __MACH__ |
| 25 | .arm |
| 26 | #endif |
| 27 | |
| 28 | .align 4 |
| 29 | .globl convert_s16_float_asm |
| 30 | #ifndef __MACH__ |
| 31 | .type convert_s16_float_asm, %function |
| 32 | #endif |
| 33 | .globl _convert_s16_float_asm |
| 34 | #ifndef __MACH__ |
| 35 | .type _convert_s16_float_asm, %function |
| 36 | #endif |
| 37 | # convert_s16_float_asm(float *out, const int16_t *in, size_t samples, const float *gain) |
| 38 | convert_s16_float_asm: |
| 39 | _convert_s16_float_asm: |
| 40 | # Hacky way to get a constant of 2^-15. |
| 41 | # Might be faster to just load a constant from memory. |
| 42 | # It's just done once however ... |
| 43 | vmov.f32 q8, #0.25 |
| 44 | vmul.f32 q8, q8, q8 |
| 45 | vmul.f32 q8, q8, q8 |
| 46 | vmul.f32 q8, q8, q8 |
| 47 | vadd.f32 q8, q8, q8 |
| 48 | |
| 49 | # Apply gain |
| 50 | vld1.f32 {d6[0]}, [r3] |
| 51 | vmul.f32 q8, q8, d6[0] |
| 52 | |
| 53 | 1: |
| 54 | # Preload here? |
| 55 | vld1.s16 {q0}, [r1]! |
| 56 | |
| 57 | # Widen to 32-bit |
| 58 | vmovl.s16 q1, d0 |
| 59 | vmovl.s16 q2, d1 |
| 60 | |
| 61 | # Convert to float |
| 62 | vcvt.f32.s32 q1, q1 |
| 63 | vcvt.f32.s32 q2, q2 |
| 64 | |
| 65 | vmul.f32 q1, q1, q8 |
| 66 | vmul.f32 q2, q2, q8 |
| 67 | |
| 68 | vst1.f32 {q1-q2}, [r0]! |
| 69 | |
| 70 | # Guaranteed to get samples in multiples of 8. |
| 71 | subs r2, r2, #8 |
| 72 | bne 1b |
| 73 | |
| 74 | bx lr |
| 75 | |
| 76 | #endif |