| /external/XNNPACK/src/f32-argmaxpool/ |
| D | 9p8x-wasmsimd-c4.c | 90 const v128_t vm4 = wasm_f32x4_gt(vi4, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4() local
91 vmax = wasm_v128_bitselect(vi4, vmax, vm4); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4()
92 vidx = wasm_v128_bitselect(wasm_i32x4_const_splat(4), vidx, vm4); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4()
182 const v128_t vm4 = wasm_f32x4_gt(vi4, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4() local
184 vmax = wasm_v128_bitselect(vi4, vmax, vm4); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4()
185 vidx = wasm_v128_bitselect(vidx4, vidx, vm4); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4()
297 const v128_t vm4 = wasm_f32x4_gt(vi4, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4() local
299 vmax = wasm_v128_bitselect(vi4, vmax, vm4); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4()
300 vidx = wasm_v128_bitselect(vidx4, vidx, vm4); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4()
354 const v128_t vm4 = wasm_f32x4_gt(vi4, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4() local
[all …]
|
| D | 9p8x-neon-c4.c | 81 const uint32x4_t vm4 = vcgtq_f32(vi4, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4() local
82 vmax = vbslq_f32(vm4, vi4, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4()
83 vidx = vbslq_u32(vm4, vmovq_n_u32(4), vidx); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4()
163 const uint32x4_t vm4 = vcgtq_f32(vi4, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4() local
165 vmax = vbslq_f32(vm4, vi4, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4()
166 vidx = vbslq_u32(vm4, vidx4, vidx); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4()
266 const uint32x4_t vm4 = vcgtq_f32(vi4, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4() local
268 vmax = vbslq_f32(vm4, vi4, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4()
269 vidx = vbslq_u32(vm4, vidx4, vidx); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4()
321 const uint32x4_t vm4 = vcgtq_f32(vi4, vmax); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4() local
[all …]
|
| D | 9p8x-sse2-c4.c | 90 const __m128i vm4 = _mm_castps_si128(_mm_cmpgt_ps(vi4, vmax)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4() local
92 vidx = _mm_or_si128(_mm_andnot_si128(vm4, vidx), _mm_and_si128(vm4, _mm_set1_epi32(4))); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4()
182 const __m128i vm4 = _mm_castps_si128(_mm_cmpgt_ps(vi4, vmax)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4() local
185 vidx = _mm_or_si128(_mm_andnot_si128(vm4, vidx), _mm_and_si128(vm4, vidx4)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4()
297 const __m128i vm4 = _mm_castps_si128(_mm_cmpgt_ps(vi4, vmax)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4() local
300 vidx = _mm_or_si128(_mm_andnot_si128(vm4, vidx), _mm_and_si128(vm4, vidx4)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4()
354 const __m128i vm4 = _mm_castps_si128(_mm_cmpgt_ps(vi4, vmax)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4() local
357 vidx = _mm_or_si128(_mm_andnot_si128(vm4, vidx), _mm_and_si128(vm4, vidx4)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4()
|
| D | 9x-neon-c4.c | 100 const uint32x4_t vm4 = vcgtq_f32(vi4, vmax); in xnn_f32_argmaxpool_ukernel_9x__neon_c4() local
101 vmax = vbslq_f32(vm4, vi4, vmax); in xnn_f32_argmaxpool_ukernel_9x__neon_c4()
102 vidx = vbslq_u32(vm4, vmovq_n_u32(4), vidx); in xnn_f32_argmaxpool_ukernel_9x__neon_c4()
149 const uint32x4_t vm4 = vcgtq_f32(vi4, vmax); in xnn_f32_argmaxpool_ukernel_9x__neon_c4() local
150 vmax = vbslq_f32(vm4, vi4, vmax); in xnn_f32_argmaxpool_ukernel_9x__neon_c4()
151 vidx = vbslq_u32(vm4, vmovq_n_u32(4), vidx); in xnn_f32_argmaxpool_ukernel_9x__neon_c4()
|
| D | 9x-wasmsimd-c4.c | 110 const v128_t vm4 = wasm_f32x4_gt(vi4, vmax); in xnn_f32_argmaxpool_ukernel_9x__wasmsimd_c4() local
111 vmax = wasm_v128_bitselect(vi4, vmax, vm4); in xnn_f32_argmaxpool_ukernel_9x__wasmsimd_c4()
112 vidx = wasm_v128_bitselect(wasm_i32x4_const_splat(4), vidx, vm4); in xnn_f32_argmaxpool_ukernel_9x__wasmsimd_c4()
161 const v128_t vm4 = wasm_f32x4_gt(vi4, vmax); in xnn_f32_argmaxpool_ukernel_9x__wasmsimd_c4() local
162 vmax = wasm_v128_bitselect(vi4, vmax, vm4); in xnn_f32_argmaxpool_ukernel_9x__wasmsimd_c4()
163 vidx = wasm_v128_bitselect(wasm_i32x4_const_splat(4), vidx, vm4); in xnn_f32_argmaxpool_ukernel_9x__wasmsimd_c4()
|
| D | 9x-sse2-c4.c | 109 const __m128i vm4 = _mm_castps_si128(_mm_cmpgt_ps(vi4, vmax)); in xnn_f32_argmaxpool_ukernel_9x__sse2_c4() local
111 vidx = _mm_or_si128(_mm_andnot_si128(vm4, vidx), _mm_and_si128(vm4, _mm_set1_epi32(4))); in xnn_f32_argmaxpool_ukernel_9x__sse2_c4()
160 const __m128i vm4 = _mm_castps_si128(_mm_cmpgt_ps(vi4, vmax)); in xnn_f32_argmaxpool_ukernel_9x__sse2_c4() local
162 vidx = _mm_or_si128(_mm_andnot_si128(vm4, vidx), _mm_and_si128(vm4, _mm_set1_epi32(4))); in xnn_f32_argmaxpool_ukernel_9x__sse2_c4()
|
| /external/arm-trusted-firmware/plat/arm/board/fvp/fdts/ |
| D | fvp_spmc_manifest.dts | 53 vm4 {
|
| /external/XNNPACK/src/f16-raddstoreexpminusmax/gen/ |
| D | neonfp16arith-rr2-p2-x40.c | 104 const uint16x8_t vm4 = vcltq_f16(vx4, vdenorm_cutoff); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x40() local
110 vf4 = vreinterpretq_f16_u16(vbicq_u16(vreinterpretq_u16_f16(vf4), vm4)); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x40()
|
| D | neonfp16arith-rr2-p2-x40-acc2.c | 105 const uint16x8_t vm4 = vcltq_f16(vx4, vdenorm_cutoff); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x40_acc2() local
111 vf4 = vreinterpretq_f16_u16(vbicq_u16(vreinterpretq_u16_f16(vf4), vm4)); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x40_acc2()
|
| D | neonfp16arith-rr2-p2-x40-acc5.c | 108 const uint16x8_t vm4 = vcltq_f16(vx4, vdenorm_cutoff); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x40_acc5() local
114 vf4 = vreinterpretq_f16_u16(vbicq_u16(vreinterpretq_u16_f16(vf4), vm4)); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x40_acc5()
|
| D | neonfp16arith-rr2-p2-x48.c | 113 const uint16x8_t vm4 = vcltq_f16(vx4, vdenorm_cutoff); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x48() local
121 vf4 = vreinterpretq_f16_u16(vbicq_u16(vreinterpretq_u16_f16(vf4), vm4)); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x48()
|
| D | neonfp16arith-rr2-p2-x48-acc3.c | 115 const uint16x8_t vm4 = vcltq_f16(vx4, vdenorm_cutoff); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x48_acc3() local
123 vf4 = vreinterpretq_f16_u16(vbicq_u16(vreinterpretq_u16_f16(vf4), vm4)); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x48_acc3()
|
| D | neonfp16arith-rr2-p2-x48-acc2.c | 114 const uint16x8_t vm4 = vcltq_f16(vx4, vdenorm_cutoff); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x48_acc2() local
122 vf4 = vreinterpretq_f16_u16(vbicq_u16(vreinterpretq_u16_f16(vf4), vm4)); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x48_acc2()
|
| D | neonfp16arith-rr2-p2-x64-acc4.c | 134 const uint16x8_t vm4 = vcltq_f16(vx4, vdenorm_cutoff); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x64_acc4() local
146 vf4 = vreinterpretq_f16_u16(vbicq_u16(vreinterpretq_u16_f16(vf4), vm4)); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x64_acc4()
|
| D | neonfp16arith-rr2-p2-x64-acc2.c | 132 const uint16x8_t vm4 = vcltq_f16(vx4, vdenorm_cutoff); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x64_acc2() local
144 vf4 = vreinterpretq_f16_u16(vbicq_u16(vreinterpretq_u16_f16(vf4), vm4)); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x64_acc2()
|
| D | neonfp16arith-rr2-p2-x64.c | 131 const uint16x8_t vm4 = vcltq_f16(vx4, vdenorm_cutoff); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x64() local
143 vf4 = vreinterpretq_f16_u16(vbicq_u16(vreinterpretq_u16_f16(vf4), vm4)); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x64()
|
| D | neonfp16arith-rr2-p2-x72.c | 140 const uint16x8_t vm4 = vcltq_f16(vx4, vdenorm_cutoff); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x72() local
154 vf4 = vreinterpretq_f16_u16(vbicq_u16(vreinterpretq_u16_f16(vf4), vm4)); in xnn_f16_raddstoreexpminusmax_ukernel__neonfp16arith_rr2_p2_x72()
|
| /external/arm-trusted-firmware/plat/arm/board/tc/fdts/ |
| D | tc_spmc_manifest.dts | 46 vm4 {
|
| /external/XNNPACK/src/f16-vsigmoid/gen/ |
| D | vsigmoid-neonfp16arith-rr2-p2-div-x40.c | 120 const uint16x8_t vm4 = vcltq_f16(vx4, vmovq_n_f16(0.0f)); in xnn_f16_vsigmoid_ukernel__neonfp16arith_rr2_p2_div_x40() local
126 vf4 = vbslq_f16(vm4, vf4, vsubq_f16(vone, vf4)); in xnn_f16_vsigmoid_ukernel__neonfp16arith_rr2_p2_div_x40()
|
| D | vsigmoid-neonfp16arith-rr2-p2-nr1recps-x40.c | 138 const uint16x8_t vm4 = vcltq_f16(vx4, vmovq_n_f16(0.0f)); in xnn_f16_vsigmoid_ukernel__neonfp16arith_rr2_p2_nr1recps_x40() local
144 vf4 = vbslq_f16(vm4, vf4, vsubq_f16(vone, vf4)); in xnn_f16_vsigmoid_ukernel__neonfp16arith_rr2_p2_nr1recps_x40()
|
| D | vsigmoid-neonfp16arith-rr2-p2-nr1fma-x40.c | 138 const uint16x8_t vm4 = vcltq_f16(vx4, vmovq_n_f16(0.0f)); in xnn_f16_vsigmoid_ukernel__neonfp16arith_rr2_p2_nr1fma_x40() local
144 vf4 = vbslq_f16(vm4, vf4, vsubq_f16(vone, vf4)); in xnn_f16_vsigmoid_ukernel__neonfp16arith_rr2_p2_nr1fma_x40()
|
| D | vsigmoid-neonfp16arith-rr2-p2-div-x48.c | 133 const uint16x8_t vm4 = vcltq_f16(vx4, vmovq_n_f16(0.0f)); in xnn_f16_vsigmoid_ukernel__neonfp16arith_rr2_p2_div_x48() local
140 vf4 = vbslq_f16(vm4, vf4, vsubq_f16(vone, vf4)); in xnn_f16_vsigmoid_ukernel__neonfp16arith_rr2_p2_div_x48()
|
| D | vsigmoid-neonfp16arith-rr2-p2-nr1fma-x48.c | 154 const uint16x8_t vm4 = vcltq_f16(vx4, vmovq_n_f16(0.0f)); in xnn_f16_vsigmoid_ukernel__neonfp16arith_rr2_p2_nr1fma_x48() local
161 vf4 = vbslq_f16(vm4, vf4, vsubq_f16(vone, vf4)); in xnn_f16_vsigmoid_ukernel__neonfp16arith_rr2_p2_nr1fma_x48()
|
| D | vsigmoid-neonfp16arith-rr2-p2-div-x56.c | 146 const uint16x8_t vm4 = vcltq_f16(vx4, vmovq_n_f16(0.0f)); in xnn_f16_vsigmoid_ukernel__neonfp16arith_rr2_p2_div_x56() local
154 vf4 = vbslq_f16(vm4, vf4, vsubq_f16(vone, vf4)); in xnn_f16_vsigmoid_ukernel__neonfp16arith_rr2_p2_div_x56()
|
| D | vsigmoid-neonfp16arith-rr2-p2-nr1recps-x48.c | 154 const uint16x8_t vm4 = vcltq_f16(vx4, vmovq_n_f16(0.0f)); in xnn_f16_vsigmoid_ukernel__neonfp16arith_rr2_p2_nr1recps_x48() local
161 vf4 = vbslq_f16(vm4, vf4, vsubq_f16(vone, vf4)); in xnn_f16_vsigmoid_ukernel__neonfp16arith_rr2_p2_nr1recps_x48()
|