diff options
| -rw-r--r-- | main.c | 123 |
1 files changed, 87 insertions, 36 deletions
@@ -22,8 +22,8 @@ typedef struct aos_tri { } aos_tri; typedef struct soa_tri { - UINT32 x[3]; - UINT32 y[3]; + UINT16 x[3]; + UINT16 y[3]; } soa_tri; aos_tri aos_triangles[] = { @@ -33,7 +33,8 @@ aos_tri aos_triangles[] = { .p3 = { 512, 384 }, }, }; -soa_tri soa_triangles[] = { + +_Alignas(UINT32) soa_tri soa_triangles[] = { { .x = { 128, 320, 512 }, .y = { 384, 96, 384 }, @@ -293,53 +294,103 @@ void RenderTriangleSIMDSimplified(void) { max_y += max_y % 2; min_y -= min_y % 2; - __m128i v1v0y = _mm_set1_epi32(t.y[0] - t.y[1]); - __m128i v0v1x = _mm_set1_epi32(t.x[1] - t.x[0]); + INT32 s_v1v0y = t.y[0] - t.y[1]; + INT32 s_v0v1x = t.x[1] - t.x[0]; + __m128i v1v0y = _mm_set1_epi32(s_v1v0y); + __m128i v0v1x = _mm_set1_epi32(s_v0v1x); - __m128i v2v1y = _mm_set1_epi32(t.y[1] - t.y[2]); - __m128i v1v2x = _mm_set1_epi32(t.x[2] - t.x[1]); + INT32 s_v2v1y = t.y[1] - t.y[2]; + INT32 s_v1v2x = t.x[2] - t.x[1]; + __m128i v2v1y = _mm_set1_epi32(s_v2v1y); + __m128i v1v2x = _mm_set1_epi32(s_v1v2x); - __m128i v0v2y = _mm_set1_epi32(t.y[2] - t.y[0]); - __m128i v2v0x = _mm_set1_epi32(t.x[0] - t.x[2]); + INT32 s_v0v2y = t.y[2] - t.y[0]; + INT32 s_v2v0x = t.x[0] - t.x[2]; + __m128i v0v2y = _mm_set1_epi32(s_v0v2y); + __m128i v2v0x = _mm_set1_epi32(s_v2v0x); __m128i dt1_C = _mm_set1_epi32(t.x[0] * t.y[1] - t.y[0] * t.x[1]); __m128i dt2_C = _mm_set1_epi32(t.x[1] * t.y[2] - t.y[1] * t.x[2]); __m128i dt3_C = _mm_set1_epi32(t.x[2] * t.y[0] - t.y[2] * t.x[0]); - __m128i zero = _mm_set1_epi32(-1); + __m128i m_one = _mm_set1_epi32(-1); + + __m128i vx = _mm_set_epi32(min_x + 1, min_x, min_x + 1, min_x); + __m128i vy = _mm_set_epi32(min_y + 1, min_y + 1, min_y, min_y); + __m128i dt1_A = _mm_mullo_epi32(v1v0y, vx); + __m128i dt1_B = _mm_mullo_epi32(v0v1x, vy); + + __m128i dt2_A = _mm_mullo_epi32(v2v1y, vx); + __m128i dt2_B = _mm_mullo_epi32(v1v2x, vy); + + __m128i dt3_A = _mm_mullo_epi32(v0v2y, vx); + __m128i dt3_B = _mm_mullo_epi32(v2v0x, vy); + + __m128i dt1_1 = _mm_add_epi32(dt1_A, dt1_B); + __m128i dt2_1 = _mm_add_epi32(dt2_A, dt2_B); + __m128i dt3_1 = _mm_add_epi32(dt3_A, dt3_B); + + __m128i x0_dt1 = _mm_add_epi32(dt1_1, dt1_C); + __m128i x0_dt2 = _mm_add_epi32(dt2_1, dt2_C); + __m128i x0_dt3 = _mm_add_epi32(dt3_1, dt3_C); + + __m128i diffx1 = _mm_set_epi32(2 * s_v1v0y, 2 * s_v1v0y, 2 * s_v1v0y, 2 * s_v1v0y); + __m128i diffx2 = _mm_set_epi32(2 * s_v2v1y, 2 * s_v2v1y, 2 * s_v2v1y, 2 * s_v2v1y); + __m128i diffx3 = _mm_set_epi32(2 * s_v0v2y, 2 * s_v0v2y, 2 * s_v0v2y, 2 * s_v0v2y); + + __m128i diffy1 = _mm_set_epi32(2 * s_v0v1x, 2 * s_v0v1x, 2 * s_v0v1x, 2 * s_v0v1x); + __m128i diffy2 = _mm_set_epi32(2 * s_v1v2x, 2 * s_v1v2x, 2 * s_v1v2x, 2 * s_v1v2x); + __m128i diffy3 = _mm_set_epi32(2 * s_v2v0x, 2 * s_v2v0x, 2 * s_v2v0x, 2 * s_v2v0x); for (UINT32 y = min_y; y < max_y + 1; y += 2) { UINT32 row_1 = y * g_width; UINT32 row_2 = (y + 1) * g_width; - for (UINT32 x = min_x; x < max_x + 1; x += 2) { - __m128i vx = _mm_set_epi32(x + 1, x, x + 1, x); - __m128i vy = _mm_set_epi32(y + 1, y + 1, y, y); + // __m128i vx = _mm_set_epi32(min_x + 1, min_x, min_x + 1, min_x); + // __m128i vy = _mm_set_epi32(y + 1, y + 1, y, y); + // __m128i dt1_A = _mm_mullo_epi32(v1v0y, vx); + // __m128i dt1_B = _mm_mullo_epi32(v0v1x, vy); + + // __m128i dt2_A = _mm_mullo_epi32(v2v1y, vx); + // __m128i dt2_B = _mm_mullo_epi32(v1v2x, vy); + + // __m128i dt3_A = _mm_mullo_epi32(v0v2y, vx); + // __m128i dt3_B = _mm_mullo_epi32(v2v0x, vy); + + // __m128i dt1_1 = _mm_add_epi32(dt1_A, dt1_B); + // __m128i dt2_1 = _mm_add_epi32(dt2_A, dt2_B); + // __m128i dt3_1 = _mm_add_epi32(dt3_A, dt3_B); + + // __m128i dt1 = _mm_add_epi32(dt1_1, dt1_C); + // __m128i dt2 = _mm_add_epi32(dt2_1, dt2_C); + // __m128i dt3 = _mm_add_epi32(dt3_1, dt3_C); - __m128i dt1_A = _mm_mullo_epi32(v1v0y, vx); - __m128i dt1_B = _mm_mullo_epi32(v0v1x, vy); + // __m128i diff1 = _mm_set_epi32(2 * s_v1v0y, 2 * s_v1v0y, 2 * s_v1v0y, 2 * s_v1v0y); + // __m128i diff2 = _mm_set_epi32(2 * s_v2v1y, 2 * s_v2v1y, 2 * s_v2v1y, 2 * s_v2v1y); + // __m128i diff3 = _mm_set_epi32(2 * s_v0v2y, 2 * s_v0v2y, 2 * s_v0v2y, 2 * s_v0v2y); - __m128i dt2_A = _mm_mullo_epi32(v2v1y, vx); - __m128i dt2_B = _mm_mullo_epi32(v1v2x, vy); + x0_dt1 = _mm_add_epi32(x0_dt1, diffy1); + x0_dt2 = _mm_add_epi32(x0_dt2, diffy2); + x0_dt3 = _mm_add_epi32(x0_dt3, diffy3); + __m128i dt1 = x0_dt1; + __m128i dt2 = x0_dt2; + __m128i dt3 = x0_dt3; - __m128i dt3_A = _mm_mullo_epi32(v0v2y, vx); - __m128i dt3_B = _mm_mullo_epi32(v2v0x, vy); + __m128i dt1_or_dt2 = _mm_or_si128(dt1, dt2); + __m128i or = _mm_or_si128(dt1_or_dt2, dt3); + __m128i gtz = _mm_cmpgt_epi32(or, m_one); - __m128i dt1_1 = _mm_add_epi32(dt1_A, dt1_B); - __m128i dt2_1 = _mm_add_epi32(dt2_A, dt2_B); - __m128i dt3_1 = _mm_add_epi32(dt3_A, dt3_B); + pixels[row_1 / 2] = _mm_extract_epi64(gtz, 0); + pixels[row_2 / 2] = _mm_extract_epi64(gtz, 1); - __m128i dt1 = _mm_add_epi32(dt1_1, dt1_C); - __m128i dt2 = _mm_add_epi32(dt2_1, dt2_C); - __m128i dt3 = _mm_add_epi32(dt3_1, dt3_C); + for (UINT32 x = min_x + 2; x < max_x + 1; x += 2) { + dt1 = _mm_add_epi32(dt1, diffx1); + dt2 = _mm_add_epi32(dt2, diffx2); + dt3 = _mm_add_epi32(dt3, diffx3); - // __m128i t1_gtz = _mm_cmpgt_epi32(d_t1, zero); - // __m128i t2_gtz = _mm_cmpgt_epi32(d_t2, zero); - // __m128i t1_and_t2 = _mm_and_si128(t1_gtz, t2_gtz); - // __m128i t3_gtz = _mm_cmpgt_epi32(d_t3, zero); __m128i dt1_or_dt2 = _mm_or_si128(dt1, dt2); __m128i or = _mm_or_si128(dt1_or_dt2, dt3); - __m128i gtz = _mm_cmpgt_epi32(or, zero); + __m128i gtz = _mm_cmpgt_epi32(or, m_one); pixels[(row_1 + x) / 2] = _mm_extract_epi64(gtz, 0); pixels[(row_2 + x) / 2] = _mm_extract_epi64(gtz, 1); @@ -404,9 +455,9 @@ LRESULT CALLBACK WndProc(HWND window, UINT msg, WPARAM wparam, LPARAM lparam) { break; } case WM_PAINT: { - RenderFullscren(); - RenderTriangle(); - RenderTriangleSIMD(); + // RenderFullscren(); + // RenderTriangle(); + // RenderTriangleSIMD(); RenderTriangleSIMDSimplified(); frame_count++; @@ -454,9 +505,9 @@ int mainCRTStartup(void) { DispatchMessage(&msg); } - RenderFullscren(); - RenderTriangle(); - RenderTriangleSIMD(); + // RenderFullscren(); + // RenderTriangle(); + // RenderTriangleSIMD(); RenderTriangleSIMDSimplified(); frame_count++; } |