/external/XNNPACK/src/f32-conv-hwc/gen/ |
D | 3x3s2p0p1c3x4-neon-2x1.c | 39 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x1() local 47 input_height_stride * (output_y_start * 2 /* vertical stride */ - input_padding_top)); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x1() 48 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x1() 49 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x1() 50 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x1() 51 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x1() 437 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x1() 438 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x1() 439 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x1() 440 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x1()
|
D | 3x3s2p1c3x4-neonfma-2x1.c | 39 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x1() local 47 input_height_stride * (output_y_start * 2 /* vertical stride */ - input_padding_top)); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x1() 48 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x1() 49 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x1() 50 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x1() 51 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x1() 435 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x1() 436 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x1() 437 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x1() 438 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x1()
|
D | 3x3s2p1c3x4-neon-2x1.c | 39 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x1() local 47 input_height_stride * (output_y_start * 2 /* vertical stride */ - input_padding_top)); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x1() 48 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x1() 49 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x1() 50 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x1() 51 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x1() 437 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x1() 438 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x1() 439 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x1() 440 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x1()
|
D | 3x3s2p0p1c3x4-neonfma-2x1.c | 39 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x1() local 47 input_height_stride * (output_y_start * 2 /* vertical stride */ - input_padding_top)); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x1() 48 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x1() 49 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x1() 50 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x1() 51 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x1() 435 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x1() 436 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x1() 437 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x1() 438 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x1()
|
D | 3x3s2p1c3x8-neon-2x1.c | 39 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neon_2x1() local 47 input_height_stride * (output_y_start * 2 /* vertical stride */ - input_padding_top)); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neon_2x1() 48 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neon_2x1() 49 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neon_2x1() 50 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neon_2x1() 51 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neon_2x1() 600 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neon_2x1() 601 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neon_2x1() 602 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neon_2x1() 603 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neon_2x1()
|
D | 3x3s2p0p1c3x8-neonfma-2x1.c | 39 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neonfma_2x1() local 47 input_height_stride * (output_y_start * 2 /* vertical stride */ - input_padding_top)); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neonfma_2x1() 48 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neonfma_2x1() 49 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neonfma_2x1() 50 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neonfma_2x1() 51 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neonfma_2x1() 598 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neonfma_2x1() 599 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neonfma_2x1() 600 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neonfma_2x1() 601 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neonfma_2x1()
|
D | 3x3s2p1c3x8-neonfma-2x1.c | 39 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neonfma_2x1() local 47 input_height_stride * (output_y_start * 2 /* vertical stride */ - input_padding_top)); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neonfma_2x1() 48 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neonfma_2x1() 49 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neonfma_2x1() 50 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neonfma_2x1() 51 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neonfma_2x1() 598 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neonfma_2x1() 599 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neonfma_2x1() 600 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neonfma_2x1() 601 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x8__neonfma_2x1()
|
D | 3x3s2p0p1c3x8-neon-2x1.c | 39 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x1() local 47 input_height_stride * (output_y_start * 2 /* vertical stride */ - input_padding_top)); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x1() 48 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x1() 49 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x1() 50 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x1() 51 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x1() 600 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x1() 601 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x1() 602 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x1() 603 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x1()
|
D | 3x3s2p0p1c3x4-neonfma-2x2.c | 39 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x2() local 47 input_height_stride * (output_y_start * 2 /* vertical stride */ - input_padding_top)); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x2() 48 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x2() 49 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x2() 50 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x2() 51 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x2() 737 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x2() 738 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x2() 739 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x2() 740 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neonfma_2x2()
|
D | 3x3s2p1c3x4-neon-2x2.c | 39 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x2() local 47 input_height_stride * (output_y_start * 2 /* vertical stride */ - input_padding_top)); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x2() 48 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x2() 49 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x2() 50 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x2() 51 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x2() 741 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x2() 742 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x2() 743 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x2() 744 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neon_2x2()
|
D | 3x3s2p0p1c3x4-neon-2x2.c | 39 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x2() local 47 input_height_stride * (output_y_start * 2 /* vertical stride */ - input_padding_top)); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x2() 48 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x2() 49 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x2() 50 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x2() 51 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x2() 741 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x2() 742 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x2() 743 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x2() 744 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__neon_2x2()
|
D | 3x3s2p1c3x4-neonfma-2x2.c | 39 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x2() local 47 input_height_stride * (output_y_start * 2 /* vertical stride */ - input_padding_top)); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x2() 48 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x2() 49 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x2() 50 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x2() 51 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x2() 737 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x2() 738 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x2() 739 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x2() 740 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__neonfma_2x2()
|
D | 3x3s2p0p1c3x8-neon-2x2.c | 39 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x2() local 47 input_height_stride * (output_y_start * 2 /* vertical stride */ - input_padding_top)); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x2() 48 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x2() 49 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x2() 50 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x2() 51 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x2() 1066 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x2() 1067 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x2() 1068 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x2() 1069 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x8__neon_2x2()
|
/external/XNNPACK/src/f32-conv-hwc2chw/ |
D | 3x3s2p1c3x4-sse-1x1.c | 34 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_1x1() local 40 …const float* i0 = (const float*) ((uintptr_t) input + input_height_stride * (output_y_start * 2 - … in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_1x1() 41 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_1x1() 42 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_1x1() 306 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_1x1() 307 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_1x1()
|
D | 3x3s2p1c3x4-neonfma-2x2.c | 34 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() local 40 …const float* i0 = (const float*) ((uintptr_t) input + input_height_stride * (output_y_start * 2 - … in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 41 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 42 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 43 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 44 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 649 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 650 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 651 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2() 652 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neonfma_2x2()
|
D | 3x3s2p1c3x4-neon-2x2.c | 34 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() local 40 …const float* i0 = (const float*) ((uintptr_t) input + input_height_stride * (output_y_start * 2 - … in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 41 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 42 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 43 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 44 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 645 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 646 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 647 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2() 648 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__neon_2x2()
|
D | 3x3s2p1c3x4-scalar-1x1.c | 32 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__scalar_1x1() local 38 …const float* i0 = (const float*) ((uintptr_t) input + input_height_stride * (output_y_start * 2 - … in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__scalar_1x1() 39 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__scalar_1x1() 40 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__scalar_1x1() 666 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__scalar_1x1() 667 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__scalar_1x1()
|
D | 3x3s2p1c3x4-wasmsimd-2x2.c | 34 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() local 40 …const float* i0 = (const float*) ((uintptr_t) input + input_height_stride * (output_y_start * 2 - … in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 41 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 42 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 43 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 44 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 648 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 649 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 650 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2() 651 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__wasmsimd_2x2()
|
D | 3x3s2p1c3x4-sse-2x2.c | 34 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() local 40 …const float* i0 = (const float*) ((uintptr_t) input + input_height_stride * (output_y_start * 2 - … in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 41 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 42 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 43 const float* i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 44 const float* i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 649 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 650 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 651 i3 = (const float*) ((uintptr_t) i2 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2() 652 i4 = (const float*) ((uintptr_t) i3 + input_height_stride); in xnn_f32_conv_hwc2chw_ukernel_3x3s2p1c3x4__sse_2x2()
|
/external/tensorflow/tensorflow/lite/kernels/internal/optimized/ |
D | depthwiseconv_uint8_3x3_filter.h | 292 static_assert(offsetof(DepthwiseConvDotProdParams, input_height_stride) == 6078 const int input_height_stride = function_params->input_height_stride; 6268 input_block_data += input_height_stride; 6309 const int input_height_stride = function_params->input_height_stride; 6360 input_block_data += input_height_stride; 6697 input_block_data += input_height_stride; 6741 const int input_height_stride = function_params->input_height_stride; 6776 input_block_data += input_height_stride; 6803 input_height_stride - width_block_number * input_width_micro_repeats); 6805 TFLITE_DCHECK_GE(copy_size, input_height_stride - 1); [all …]
|
D | depthwiseconv_3x3_filter_common.h | 367 int32 input_height_stride; 568 const int input_height_stride = function_params->input_height_stride; 580 row_ptr += input_height_stride;
|
D | depthwiseconv_uint8_transitional.h | 452 const int input_height_stride = function_params.input_height_stride; 508 input_block_data += input_height_stride; 542 input_data = input_block_data + k_height * input_height_stride + 651 const int input_height_stride = function_params->input_height_stride; 693 input_block_data += input_height_stride; 718 input_height_stride - width_block_number * input_width_micro_repeats); 720 TFLITE_DCHECK_GE(copy_size, input_height_stride - 1); 726 input_data = input_block_data + k_height * input_height_stride; 775 const int input_height_stride = function_params->input_height_stride; 802 input_block_data += input_height_stride; [all …]
|
/external/XNNPACK/src/f32-conv-hwc/ |
D | 3x3s2p1c3x4-scalar-1x1.c | 32 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__scalar_1x1() local 39 …const float* i0 = (const float*) ((uintptr_t) input + input_height_stride * (output_y_start * 2 - … in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__scalar_1x1() 40 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__scalar_1x1() 41 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__scalar_1x1() 680 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__scalar_1x1() 681 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p1c3x4__scalar_1x1()
|
D | 3x3s2p0p1c3x4-scalar-1x1.c | 32 const size_t input_height_stride = input_width * 3 /* channels */ * sizeof(float); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__scalar_1x1() local 39 …const float* i0 = (const float*) ((uintptr_t) input + input_height_stride * (output_y_start * 2 - … in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__scalar_1x1() 40 const float* i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__scalar_1x1() 41 const float* i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__scalar_1x1() 680 i1 = (const float*) ((uintptr_t) i0 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__scalar_1x1() 681 i2 = (const float*) ((uintptr_t) i1 + input_height_stride); in xnn_f32_conv_hwc_ukernel_3x3s2p0p1c3x4__scalar_1x1()
|
/external/tensorflow/tensorflow/lite/kernels/internal/optimized/integer_ops/ |
D | depthwise_conv_hybrid.h | 139 const int input_height_stride = input_shape.Dims(3) * input_shape.Dims(2); in DepthwiseConvHybridGeneral() local 140 const int input_batch_stride = input_height_stride * input_shape.Dims(1); in DepthwiseConvHybridGeneral() 198 input_data + in_y * input_height_stride + b * input_batch_stride, in DepthwiseConvHybridGeneral()
|