macro

2 files changed, 24 insertions, 39 deletions

diff --git a/cachelab-handout/trans.c b/cachelab-handout/trans.c
index 8ce61ac..8f347aa 100644
--- a/cachelab-handout/trans.c
+++ b/cachelab-handout/trans.c
@@ -120,34 +120,14 @@ void transpose_submit(int M, int N, int A[N][M], int B[M][N])
                 }
 
                 // A00 -> B00
-                j2 = 0;
-                t4 = 0;
-                for (i2 = 0; i2 < CASE_64_64_BLOCK_SIZE; i2++)
-                {
-                    t0 = A[i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2][j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 0];
-                    t1 = A[i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2][j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 1];
-                    t2 = A[i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2][j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 2];
-                    t3 = A[i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2][j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 3];
-                    B[j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 0][i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2] = t0;
-                    B[j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 1][i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2] = t1;
-                    B[j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 2][i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2] = t2;
-                    B[j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 3][i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2] = t3;
-                }
+                TRANSPOSE_4_BY_4(i*8 + 0*4, j*8 + 0*4, 
+                    j*8 + 0*4, i*8 + 0*4, i2);
+
 
                 // A01 -> temp B01
-                j2 = 0;
-                t4 = 4;
-                for (i2 = 0; i2 < CASE_64_64_BLOCK_SIZE; i2++)
-                {
-                    t0 = A[i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2][j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 0];
-                    t1 = A[i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2][j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 1];
-                    t2 = A[i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2][j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 2];
-                    t3 = A[i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2][j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 3];
-                    B[j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + 0][i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + i2] = t0;
-                    B[j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + 1][i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + i2] = t1;
-                    B[j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + 2][i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + i2] = t2;
-                    B[j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + 3][i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + i2] = t3;
-                }
+                TRANSPOSE_4_BY_4(i*8 + 0*4, j*8 + 1*4, 
+                    j*8 + 0*4, i*8 + 1*4, i2);
+
 
                 // B01 -> temp var, A10 -> B01, temp var -> B10
                 j2 = 4, t4 = 0;
@@ -193,19 +173,8 @@ void transpose_submit(int M, int N, int A[N][M], int B[M][N])
                 }
 
                 // A11 -> B11
-                j2 = 4;
-                t4 = 4;
-                for (i2 = 0; i2 < CASE_64_64_BLOCK_SIZE; i2++)
-                {
-                    t0 = A[i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2][j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 0];
-                    t1 = A[i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2][j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 1];
-                    t2 = A[i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2][j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 2];
-                    t3 = A[i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2][j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 3];
-                    B[j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 0][i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2] = t0;
-                    B[j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 1][i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2] = t1;
-                    B[j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 2][i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2] = t2;
-                    B[j * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + t4 + 3][i * CASE_64_64_MAGIC_NUMBER * CASE_64_64_BLOCK_SIZE + j2 + i2] = t3;
-                }
+                TRANSPOSE_4_BY_4(i*8 + 1*4, j*8 + 1*4, 
+                    j*8 + 1*4, i*8 + 1*4, i2);
             }
         }
     }
diff --git a/cachelab-handout/trans_64.h b/cachelab-handout/trans_64.h
index 542866e..1f6fed5 100644
--- a/cachelab-handout/trans_64.h
+++ b/cachelab-handout/trans_64.h
@@ -1,5 +1,21 @@
 #pragma once
 
+#define TRANSPOSE_4_BY_4(base_i_A, base_j_A, base_i_B, base_j_B, tmp) \
+do \
+{ \
+for (tmp = 0; tmp < 4; tmp++) \
+{ \
+    t0 = A[base_i_A + tmp][base_j_A + 0]; \
+    t1 = A[base_i_A + tmp][base_j_A + 1]; \
+    t2 = A[base_i_A + tmp][base_j_A + 2]; \
+    t3 = A[base_i_A + tmp][base_j_A + 3]; \
+    B[base_i_B + 0][base_j_B + tmp] = t0; \
+    B[base_i_B + 1][base_j_B + tmp] = t1; \
+    B[base_i_B + 2][base_j_B + tmp] = t2; \
+    B[base_i_B + 3][base_j_B + tmp] = t3; \
+} \
+} while (0)
+
 #define CP_4_BY_8(base_i, base_j, tmp) \
 do \
 { \