/dports/graphics/vapoursynth-waifu2x-ncnn-vulkan/vapoursynth-waifu2x-ncnn-vulkan-r4/deps/ncnn/src/layer/arm/ |
H A D | convolution_3x3_pack8to1_int8.h | 1088 const int* output0_tm_1 = output0_tm_0 + tiles * 1; in conv3x3s1_winograd42_pack8to1_int8_neon() local 1104 int tmp02a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd42_pack8to1_int8_neon() 1105 int tmp13a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd42_pack8to1_int8_neon() 1116 output0_tm_1 += tiles * 6; in conv3x3s1_winograd42_pack8to1_int8_neon() 1124 int tmp02a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd42_pack8to1_int8_neon() 1125 int tmp13a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd42_pack8to1_int8_neon() 1136 output0_tm_1 += tiles * 6; in conv3x3s1_winograd42_pack8to1_int8_neon()
|
H A D | convolution_3x3_pack8to1_fp16s.h | 1052 const __fp16* output0_tm_1 = output0_tm_0 + tiles * 1; in conv3x3s1_winograd64_pack8to1_fp16sa_neon() local 1063 __fp16 tmp024a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd64_pack8to1_fp16sa_neon() 1064 __fp16 tmp135a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd64_pack8to1_fp16sa_neon() 1081 output0_tm_1 += tiles * 8; in conv3x3s1_winograd64_pack8to1_fp16sa_neon()
|
H A D | convolution_3x3_pack4to1_bf16s.h | 1967 const float* output0_tm_1 = output0_tm_0 + tiles * 1; in conv3x3s1_winograd64_pack4to1_bf16s_neon() local 1978 float tmp024a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd64_pack4to1_bf16s_neon() 1979 float tmp135a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd64_pack4to1_bf16s_neon() 1996 output0_tm_1 += tiles * 8; in conv3x3s1_winograd64_pack4to1_bf16s_neon()
|
H A D | convolution_3x3_pack8_fp16s.h | 1075 const __fp16* output0_tm_1 = output0_tm_0 + tiles * 8; in conv3x3s1_winograd64_pack8_fp16sa_neon() local 1089 float16x8_t _out0tm1 = vld1q_f16(output0_tm_1); in conv3x3s1_winograd64_pack8_fp16sa_neon() 1138 output0_tm_1 += tiles * 64; in conv3x3s1_winograd64_pack8_fp16sa_neon() 2168 const __fp16* output0_tm_1 = output0_tm_0 + tiles * 8; in conv3x3s1_winograd42_pack8_fp16sa_neon() local 2180 float16x8_t _out0tm1 = vld1q_f16(output0_tm_1); in conv3x3s1_winograd42_pack8_fp16sa_neon() 2203 output0_tm_1 += tiles * 48; in conv3x3s1_winograd42_pack8_fp16sa_neon()
|
H A D | convolution_3x3_pack4_bf16s.h | 1685 const float* output0_tm_1 = output0_tm_0 + tiles * 4; in conv3x3s1_winograd64_pack4_bf16s_neon() local 1699 float32x4_t _out0tm1 = vld1q_f32(output0_tm_1); in conv3x3s1_winograd64_pack4_bf16s_neon() 1748 output0_tm_1 += tiles * 32; in conv3x3s1_winograd64_pack4_bf16s_neon() 3391 const float* output0_tm_1 = output0_tm_0 + tiles * 4; in conv3x3s1_winograd42_pack4_bf16s_neon() local 3403 float32x4_t _out0tm1 = vld1q_f32(output0_tm_1); in conv3x3s1_winograd42_pack4_bf16s_neon() 3426 output0_tm_1 += tiles * 24; in conv3x3s1_winograd42_pack4_bf16s_neon()
|
/dports/graphics/waifu2x-ncnn-vulkan/waifu2x-ncnn-vulkan-20210521/src/ncnn/src/layer/arm/ |
H A D | convolution_3x3_pack8to1_int8.h | 1088 const int* output0_tm_1 = output0_tm_0 + tiles * 1; in conv3x3s1_winograd42_pack8to1_int8_neon() local 1104 int tmp02a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd42_pack8to1_int8_neon() 1105 int tmp13a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd42_pack8to1_int8_neon() 1116 output0_tm_1 += tiles * 6; in conv3x3s1_winograd42_pack8to1_int8_neon() 1124 int tmp02a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd42_pack8to1_int8_neon() 1125 int tmp13a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd42_pack8to1_int8_neon() 1136 output0_tm_1 += tiles * 6; in conv3x3s1_winograd42_pack8to1_int8_neon()
|
H A D | convolution_3x3_pack8to1_fp16s.h | 1052 const __fp16* output0_tm_1 = output0_tm_0 + tiles * 1; in conv3x3s1_winograd64_pack8to1_fp16sa_neon() local 1063 __fp16 tmp024a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd64_pack8to1_fp16sa_neon() 1064 __fp16 tmp135a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd64_pack8to1_fp16sa_neon() 1081 output0_tm_1 += tiles * 8; in conv3x3s1_winograd64_pack8to1_fp16sa_neon()
|
H A D | convolution_3x3_pack4to1_bf16s.h | 1967 const float* output0_tm_1 = output0_tm_0 + tiles * 1; in conv3x3s1_winograd64_pack4to1_bf16s_neon() local 1978 float tmp024a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd64_pack4to1_bf16s_neon() 1979 float tmp135a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd64_pack4to1_bf16s_neon() 1996 output0_tm_1 += tiles * 8; in conv3x3s1_winograd64_pack4to1_bf16s_neon()
|
H A D | convolution_3x3_pack8_fp16s.h | 1075 const __fp16* output0_tm_1 = output0_tm_0 + tiles * 8; in conv3x3s1_winograd64_pack8_fp16sa_neon() local 1089 float16x8_t _out0tm1 = vld1q_f16(output0_tm_1); in conv3x3s1_winograd64_pack8_fp16sa_neon() 1138 output0_tm_1 += tiles * 64; in conv3x3s1_winograd64_pack8_fp16sa_neon() 2168 const __fp16* output0_tm_1 = output0_tm_0 + tiles * 8; in conv3x3s1_winograd42_pack8_fp16sa_neon() local 2180 float16x8_t _out0tm1 = vld1q_f16(output0_tm_1); in conv3x3s1_winograd42_pack8_fp16sa_neon() 2203 output0_tm_1 += tiles * 48; in conv3x3s1_winograd42_pack8_fp16sa_neon()
|
H A D | convolution_3x3_pack4_bf16s.h | 1685 const float* output0_tm_1 = output0_tm_0 + tiles * 4; in conv3x3s1_winograd64_pack4_bf16s_neon() local 1699 float32x4_t _out0tm1 = vld1q_f32(output0_tm_1); in conv3x3s1_winograd64_pack4_bf16s_neon() 1748 output0_tm_1 += tiles * 32; in conv3x3s1_winograd64_pack4_bf16s_neon() 3391 const float* output0_tm_1 = output0_tm_0 + tiles * 4; in conv3x3s1_winograd42_pack4_bf16s_neon() local 3403 float32x4_t _out0tm1 = vld1q_f32(output0_tm_1); in conv3x3s1_winograd42_pack4_bf16s_neon() 3426 output0_tm_1 += tiles * 24; in conv3x3s1_winograd42_pack4_bf16s_neon()
|
/dports/benchmarks/vkpeak/vkpeak-20210430/ncnn/src/layer/arm/ |
H A D | convolution_3x3_pack8to1_int8.h | 1088 const int* output0_tm_1 = output0_tm_0 + tiles * 1; in conv3x3s1_winograd42_pack8to1_int8_neon() local 1104 int tmp02a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd42_pack8to1_int8_neon() 1105 int tmp13a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd42_pack8to1_int8_neon() 1116 output0_tm_1 += tiles * 6; in conv3x3s1_winograd42_pack8to1_int8_neon() 1124 int tmp02a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd42_pack8to1_int8_neon() 1125 int tmp13a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd42_pack8to1_int8_neon() 1136 output0_tm_1 += tiles * 6; in conv3x3s1_winograd42_pack8to1_int8_neon()
|
H A D | convolution_3x3_pack8to1_fp16s.h | 1052 const __fp16* output0_tm_1 = output0_tm_0 + tiles * 1; in conv3x3s1_winograd64_pack8to1_fp16sa_neon() local 1063 __fp16 tmp024a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd64_pack8to1_fp16sa_neon() 1064 __fp16 tmp135a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd64_pack8to1_fp16sa_neon() 1081 output0_tm_1 += tiles * 8; in conv3x3s1_winograd64_pack8to1_fp16sa_neon()
|
H A D | convolution_3x3_pack4to1_bf16s.h | 1967 const float* output0_tm_1 = output0_tm_0 + tiles * 1; in conv3x3s1_winograd64_pack4to1_bf16s_neon() local 1978 float tmp024a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd64_pack4to1_bf16s_neon() 1979 float tmp135a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd64_pack4to1_bf16s_neon() 1996 output0_tm_1 += tiles * 8; in conv3x3s1_winograd64_pack4to1_bf16s_neon()
|
H A D | convolution_3x3_pack8_fp16s.h | 1075 const __fp16* output0_tm_1 = output0_tm_0 + tiles * 8; in conv3x3s1_winograd64_pack8_fp16sa_neon() local 1089 float16x8_t _out0tm1 = vld1q_f16(output0_tm_1); in conv3x3s1_winograd64_pack8_fp16sa_neon() 1138 output0_tm_1 += tiles * 64; in conv3x3s1_winograd64_pack8_fp16sa_neon() 2168 const __fp16* output0_tm_1 = output0_tm_0 + tiles * 8; in conv3x3s1_winograd42_pack8_fp16sa_neon() local 2180 float16x8_t _out0tm1 = vld1q_f16(output0_tm_1); in conv3x3s1_winograd42_pack8_fp16sa_neon() 2203 output0_tm_1 += tiles * 48; in conv3x3s1_winograd42_pack8_fp16sa_neon()
|
H A D | convolution_3x3_pack4_bf16s.h | 1685 const float* output0_tm_1 = output0_tm_0 + tiles * 4; in conv3x3s1_winograd64_pack4_bf16s_neon() local 1699 float32x4_t _out0tm1 = vld1q_f32(output0_tm_1); in conv3x3s1_winograd64_pack4_bf16s_neon() 1748 output0_tm_1 += tiles * 32; in conv3x3s1_winograd64_pack4_bf16s_neon() 3391 const float* output0_tm_1 = output0_tm_0 + tiles * 4; in conv3x3s1_winograd42_pack4_bf16s_neon() local 3403 float32x4_t _out0tm1 = vld1q_f32(output0_tm_1); in conv3x3s1_winograd42_pack4_bf16s_neon() 3426 output0_tm_1 += tiles * 24; in conv3x3s1_winograd42_pack4_bf16s_neon()
|
/dports/misc/ncnn/ncnn-20211208/src/layer/arm/ |
H A D | convolution_3x3_pack8to1_int8.h | 1088 const int* output0_tm_1 = output0_tm_0 + tiles * 1; in conv3x3s1_winograd42_pack8to1_int8_neon() local 1104 int tmp02a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd42_pack8to1_int8_neon() 1105 int tmp13a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd42_pack8to1_int8_neon() 1116 output0_tm_1 += tiles * 6; in conv3x3s1_winograd42_pack8to1_int8_neon() 1124 int tmp02a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd42_pack8to1_int8_neon() 1125 int tmp13a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd42_pack8to1_int8_neon() 1136 output0_tm_1 += tiles * 6; in conv3x3s1_winograd42_pack8to1_int8_neon()
|
H A D | convolution_3x3_pack8to1_fp16s.h | 1052 const __fp16* output0_tm_1 = output0_tm_0 + tiles * 1; in conv3x3s1_winograd64_pack8to1_fp16sa_neon() local 1063 __fp16 tmp024a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd64_pack8to1_fp16sa_neon() 1064 __fp16 tmp135a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd64_pack8to1_fp16sa_neon() 1081 output0_tm_1 += tiles * 8; in conv3x3s1_winograd64_pack8to1_fp16sa_neon()
|
H A D | convolution_3x3_pack4to1_bf16s.h | 1967 const float* output0_tm_1 = output0_tm_0 + tiles * 1; in conv3x3s1_winograd64_pack4to1_bf16s_neon() local 1978 float tmp024a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd64_pack4to1_bf16s_neon() 1979 float tmp135a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd64_pack4to1_bf16s_neon() 1996 output0_tm_1 += tiles * 8; in conv3x3s1_winograd64_pack4to1_bf16s_neon()
|
H A D | convolution_3x3_pack8_fp16s.h | 1075 const __fp16* output0_tm_1 = output0_tm_0 + tiles * 8; in conv3x3s1_winograd64_pack8_fp16sa_neon() local 1089 float16x8_t _out0tm1 = vld1q_f16(output0_tm_1); in conv3x3s1_winograd64_pack8_fp16sa_neon() 1138 output0_tm_1 += tiles * 64; in conv3x3s1_winograd64_pack8_fp16sa_neon() 2168 const __fp16* output0_tm_1 = output0_tm_0 + tiles * 8; in conv3x3s1_winograd42_pack8_fp16sa_neon() local 2180 float16x8_t _out0tm1 = vld1q_f16(output0_tm_1); in conv3x3s1_winograd42_pack8_fp16sa_neon() 2203 output0_tm_1 += tiles * 48; in conv3x3s1_winograd42_pack8_fp16sa_neon()
|
H A D | convolution_3x3_pack4_bf16s.h | 1685 const float* output0_tm_1 = output0_tm_0 + tiles * 4; in conv3x3s1_winograd64_pack4_bf16s_neon() local 1699 float32x4_t _out0tm1 = vld1q_f32(output0_tm_1); in conv3x3s1_winograd64_pack4_bf16s_neon() 1748 output0_tm_1 += tiles * 32; in conv3x3s1_winograd64_pack4_bf16s_neon() 3391 const float* output0_tm_1 = output0_tm_0 + tiles * 4; in conv3x3s1_winograd42_pack4_bf16s_neon() local 3403 float32x4_t _out0tm1 = vld1q_f32(output0_tm_1); in conv3x3s1_winograd42_pack4_bf16s_neon() 3426 output0_tm_1 += tiles * 24; in conv3x3s1_winograd42_pack4_bf16s_neon()
|
/dports/graphics/realsr-ncnn-vulkan/realsr-ncnn-vulkan-20210210/src/ncnn/src/layer/arm/ |
H A D | convolution_3x3_pack8to1_fp16s.h | 1052 const __fp16* output0_tm_1 = output0_tm_0 + tiles * 1; in conv3x3s1_winograd64_pack8to1_fp16sa_neon() local 1063 __fp16 tmp024a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd64_pack8to1_fp16sa_neon() 1064 __fp16 tmp135a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd64_pack8to1_fp16sa_neon() 1081 output0_tm_1 += tiles * 8; in conv3x3s1_winograd64_pack8to1_fp16sa_neon()
|
H A D | convolution_3x3_pack4to1_bf16s.h | 1967 const float* output0_tm_1 = output0_tm_0 + tiles * 1; in conv3x3s1_winograd64_pack4to1_bf16s_neon() local 1978 float tmp024a = output0_tm_1[0] + output0_tm_2[0]; in conv3x3s1_winograd64_pack4to1_bf16s_neon() 1979 float tmp135a = output0_tm_1[0] - output0_tm_2[0]; in conv3x3s1_winograd64_pack4to1_bf16s_neon() 1996 output0_tm_1 += tiles * 8; in conv3x3s1_winograd64_pack4to1_bf16s_neon()
|
H A D | convolution_3x3_pack8_fp16s.h | 1075 const __fp16* output0_tm_1 = output0_tm_0 + tiles * 8; in conv3x3s1_winograd64_pack8_fp16sa_neon() local 1089 float16x8_t _out0tm1 = vld1q_f16(output0_tm_1); in conv3x3s1_winograd64_pack8_fp16sa_neon() 1138 output0_tm_1 += tiles * 64; in conv3x3s1_winograd64_pack8_fp16sa_neon() 2168 const __fp16* output0_tm_1 = output0_tm_0 + tiles * 8; in conv3x3s1_winograd42_pack8_fp16sa_neon() local 2180 float16x8_t _out0tm1 = vld1q_f16(output0_tm_1); in conv3x3s1_winograd42_pack8_fp16sa_neon() 2203 output0_tm_1 += tiles * 48; in conv3x3s1_winograd42_pack8_fp16sa_neon()
|
H A D | convolution_3x3_pack4_bf16s.h | 1685 const float* output0_tm_1 = output0_tm_0 + tiles * 4; in conv3x3s1_winograd64_pack4_bf16s_neon() local 1699 float32x4_t _out0tm1 = vld1q_f32(output0_tm_1); in conv3x3s1_winograd64_pack4_bf16s_neon() 1748 output0_tm_1 += tiles * 32; in conv3x3s1_winograd64_pack4_bf16s_neon() 3391 const float* output0_tm_1 = output0_tm_0 + tiles * 4; in conv3x3s1_winograd42_pack4_bf16s_neon() local 3403 float32x4_t _out0tm1 = vld1q_f32(output0_tm_1); in conv3x3s1_winograd42_pack4_bf16s_neon() 3426 output0_tm_1 += tiles * 24; in conv3x3s1_winograd42_pack4_bf16s_neon()
|
/dports/misc/ncnn/ncnn-20211208/src/layer/riscv/ |
H A D | convolution_3x3_packn.h | 645 const float* output0_tm_1 = output0_tm_0 + tiles * packn; in conv3x3s1_winograd64_packn_rvv() local 659 vfloat32m1_t _out0tm1 = vle32_v_f32m1(output0_tm_1, vl); in conv3x3s1_winograd64_packn_rvv() 691 output0_tm_1 += tiles * packn * 8; in conv3x3s1_winograd64_packn_rvv() 1316 const float* output0_tm_1 = output0_tm_0 + tiles * packn; in conv3x3s1_winograd42_packn_rvv() local 1328 vfloat32m1_t _out0tm1 = vle32_v_f32m1(output0_tm_1, vl); in conv3x3s1_winograd42_packn_rvv() 1351 output0_tm_1 += tiles * packn * 6; in conv3x3s1_winograd42_packn_rvv()
|