/external/XNNPACK/src/qs8-gavgpool/gen/ |
D | 7x-minmax-wasmsimd-c16-acc2.c | 108 const v128_t vabsaccCDEF = wasm_i32x4_abs(vaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c16_acc2() local 113 const v128_t vsgnaccCDEF = wasm_i32x4_gt(vabsaccCDEF, vaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c16_acc2() 121 const v128_t vabsaccCD = wasm_v32x4_shuffle(vabsaccCDEF, vzero, 0, 4, 1, 5); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c16_acc2() 122 const v128_t vabsaccEF = wasm_v32x4_shuffle(vabsaccCDEF, vzero, 2, 6, 3, 7); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c16_acc2()
|
D | 7x-minmax-wasmsimd-c24-acc2.c | 123 const v128_t vabsaccCDEF = wasm_i32x4_abs(vaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c24_acc2() local 130 const v128_t vsgnaccCDEF = wasm_i32x4_gt(vabsaccCDEF, vaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c24_acc2() 140 const v128_t vabsaccCD = wasm_v32x4_shuffle(vabsaccCDEF, vzero, 0, 4, 1, 5); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c24_acc2() 141 const v128_t vabsaccEF = wasm_v32x4_shuffle(vabsaccCDEF, vzero, 2, 6, 3, 7); in xnn_qs8_gavgpool_minmax_ukernel_7x__wasmsimd_c24_acc2()
|
D | 7x-minmax-sse41-c16-acc2.c | 108 const __m128i vabsaccCDEF = _mm_abs_epi32(vaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7x__sse41_c16_acc2() local 113 const __m128i vabsaccDF = _mm_shuffle_epi32(vabsaccCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gavgpool_minmax_ukernel_7x__sse41_c16_acc2() 121 const __m128i vabsprodCE = _mm_mul_epu32(vabsaccCDEF, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__sse41_c16_acc2()
|
D | 7p7x-minmax-wasmsimd-c16-acc2.c | 235 const v128_t vabsaccCDEF = wasm_i32x4_abs(vaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c16_acc2() local 240 const v128_t vsgnaccCDEF = wasm_i32x4_gt(vabsaccCDEF, vaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c16_acc2() 248 const v128_t vabsaccCD = wasm_v32x4_shuffle(vabsaccCDEF, vzero, 0, 4, 1, 5); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c16_acc2() 249 const v128_t vabsaccEF = wasm_v32x4_shuffle(vabsaccCDEF, vzero, 2, 6, 3, 7); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c16_acc2()
|
D | 7x-minmax-ssse3-c16-acc2.c | 124 const __m128i vabsaccCDEF = _mm_abs_epi32(vaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7x__ssse3_c16_acc2() local 129 const __m128i vabsaccDF = _mm_shuffle_epi32(vabsaccCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gavgpool_minmax_ukernel_7x__ssse3_c16_acc2() 137 const __m128i vabsprodCE = _mm_mul_epu32(vabsaccCDEF, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__ssse3_c16_acc2()
|
D | 7x-minmax-sse41-c24-acc2.c | 123 const __m128i vabsaccCDEF = _mm_abs_epi32(vaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7x__sse41_c24_acc2() local 130 const __m128i vabsaccDF = _mm_shuffle_epi32(vabsaccCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gavgpool_minmax_ukernel_7x__sse41_c24_acc2() 140 const __m128i vabsprodCE = _mm_mul_epu32(vabsaccCDEF, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__sse41_c24_acc2()
|
D | 7x-minmax-sse2-c16-acc2.c | 129 const __m128i vabsaccCDEF = _mm_sub_epi32(_mm_xor_si128(vaccCDEF, vsgnaccCDEF), vsgnaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7x__sse2_c16_acc2() local 134 const __m128i vabsaccDF = _mm_shuffle_epi32(vabsaccCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gavgpool_minmax_ukernel_7x__sse2_c16_acc2() 142 const __m128i vabsprodCE = _mm_mul_epu32(vabsaccCDEF, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__sse2_c16_acc2()
|
D | 7p7x-minmax-sse41-c16-acc2.c | 237 const __m128i vabsaccCDEF = _mm_abs_epi32(vaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__sse41_c16_acc2() local 242 const __m128i vabsaccDF = _mm_shuffle_epi32(vabsaccCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__sse41_c16_acc2() 250 const __m128i vabsprodCE = _mm_mul_epu32(vabsaccCDEF, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__sse41_c16_acc2()
|
D | 7x-minmax-sse2-c24-acc2.c | 154 const __m128i vabsaccCDEF = _mm_sub_epi32(_mm_xor_si128(vaccCDEF, vsgnaccCDEF), vsgnaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7x__sse2_c24_acc2() local 161 const __m128i vabsaccDF = _mm_shuffle_epi32(vabsaccCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gavgpool_minmax_ukernel_7x__sse2_c24_acc2() 171 const __m128i vabsprodCE = _mm_mul_epu32(vabsaccCDEF, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__sse2_c24_acc2()
|
D | 7x-minmax-ssse3-c24-acc2.c | 147 const __m128i vabsaccCDEF = _mm_abs_epi32(vaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7x__ssse3_c24_acc2() local 154 const __m128i vabsaccDF = _mm_shuffle_epi32(vabsaccCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gavgpool_minmax_ukernel_7x__ssse3_c24_acc2() 164 const __m128i vabsprodCE = _mm_mul_epu32(vabsaccCDEF, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7x__ssse3_c24_acc2()
|
D | 7p7x-minmax-wasmsimd-c24-acc2.c | 358 const v128_t vabsaccCDEF = wasm_i32x4_abs(vaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c24_acc2() local 365 const v128_t vsgnaccCDEF = wasm_i32x4_gt(vabsaccCDEF, vaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c24_acc2() 375 const v128_t vabsaccCD = wasm_v32x4_shuffle(vabsaccCDEF, vzero, 0, 4, 1, 5); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c24_acc2() 376 const v128_t vabsaccEF = wasm_v32x4_shuffle(vabsaccCDEF, vzero, 2, 6, 3, 7); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__wasmsimd_c24_acc2()
|
D | 7p7x-minmax-ssse3-c16-acc2.c | 285 const __m128i vabsaccCDEF = _mm_abs_epi32(vaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__ssse3_c16_acc2() local 290 const __m128i vabsaccDF = _mm_shuffle_epi32(vabsaccCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__ssse3_c16_acc2() 298 const __m128i vabsprodCE = _mm_mul_epu32(vabsaccCDEF, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__ssse3_c16_acc2()
|
D | 7p7x-minmax-sse2-c16-acc2.c | 290 const __m128i vabsaccCDEF = _mm_sub_epi32(_mm_xor_si128(vaccCDEF, vsgnaccCDEF), vsgnaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__sse2_c16_acc2() local 295 const __m128i vabsaccDF = _mm_shuffle_epi32(vabsaccCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__sse2_c16_acc2() 303 const __m128i vabsprodCE = _mm_mul_epu32(vabsaccCDEF, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__sse2_c16_acc2()
|
D | 7p7x-minmax-sse41-c24-acc2.c | 362 const __m128i vabsaccCDEF = _mm_abs_epi32(vaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__sse41_c24_acc2() local 369 const __m128i vabsaccDF = _mm_shuffle_epi32(vabsaccCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__sse41_c24_acc2() 379 const __m128i vabsprodCE = _mm_mul_epu32(vabsaccCDEF, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__sse41_c24_acc2()
|
D | 7p7x-minmax-sse2-c24-acc2.c | 457 const __m128i vabsaccCDEF = _mm_sub_epi32(_mm_xor_si128(vaccCDEF, vsgnaccCDEF), vsgnaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__sse2_c24_acc2() local 464 const __m128i vabsaccDF = _mm_shuffle_epi32(vabsaccCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__sse2_c24_acc2() 474 const __m128i vabsprodCE = _mm_mul_epu32(vabsaccCDEF, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__sse2_c24_acc2()
|
D | 7p7x-minmax-ssse3-c24-acc2.c | 450 const __m128i vabsaccCDEF = _mm_abs_epi32(vaccCDEF); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__ssse3_c24_acc2() local 457 const __m128i vabsaccDF = _mm_shuffle_epi32(vabsaccCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__ssse3_c24_acc2() 467 const __m128i vabsprodCE = _mm_mul_epu32(vabsaccCDEF, vmultiplier); in xnn_qs8_gavgpool_minmax_ukernel_7p7x__ssse3_c24_acc2()
|
/external/XNNPACK/src/qs8-dwconv/gen/ |
D | up16x9-minmax-ssse3-mul16.c | 291 const __m128i vabsaccCDEF = _mm_abs_epi32(vaccCDEF); in xnn_qs8_dwconv_minmax_ukernel_up16x9__ssse3_mul16() local 302 const __m128i vabsaccDF = _mm_shuffle_epi32(vabsaccCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up16x9__ssse3_mul16() 303 const __m128i vabsprodCE = _mm_mul_epu32(vabsaccCDEF, vmultiplier); in xnn_qs8_dwconv_minmax_ukernel_up16x9__ssse3_mul16()
|
D | up16x9-minmax-sse2-mul16.c | 291 const __m128i vabsaccCDEF = _mm_sub_epi32(_mm_xor_si128(vaccCDEF, vnmaskCDEF), vnmaskCDEF); in xnn_qs8_dwconv_minmax_ukernel_up16x9__sse2_mul16() local 302 const __m128i vabsaccDF = _mm_shuffle_epi32(vabsaccCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up16x9__sse2_mul16() 303 const __m128i vabsprodCE = _mm_mul_epu32(vabsaccCDEF, vmultiplier); in xnn_qs8_dwconv_minmax_ukernel_up16x9__sse2_mul16()
|
D | up24x9-minmax-ssse3-mul16.c | 367 const __m128i vabsaccCDEF = _mm_abs_epi32(vaccCDEF); in xnn_qs8_dwconv_minmax_ukernel_up24x9__ssse3_mul16() local 380 const __m128i vabsaccDF = _mm_shuffle_epi32(vabsaccCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up24x9__ssse3_mul16() 381 const __m128i vabsprodCE = _mm_mul_epu32(vabsaccCDEF, vmultiplier); in xnn_qs8_dwconv_minmax_ukernel_up24x9__ssse3_mul16()
|
D | up24x9-minmax-sse2-mul16.c | 367 const __m128i vabsaccCDEF = _mm_sub_epi32(_mm_xor_si128(vaccCDEF, vnmaskCDEF), vnmaskCDEF); in xnn_qs8_dwconv_minmax_ukernel_up24x9__sse2_mul16() local 380 const __m128i vabsaccDF = _mm_shuffle_epi32(vabsaccCDEF, _MM_SHUFFLE(3, 3, 1, 1)); in xnn_qs8_dwconv_minmax_ukernel_up24x9__sse2_mul16() 381 const __m128i vabsprodCE = _mm_mul_epu32(vabsaccCDEF, vmultiplier); in xnn_qs8_dwconv_minmax_ukernel_up24x9__sse2_mul16()
|