| Input:A, ,x, N, x_norm, r_norm |
| Output:x |
| (1) | fork (A, ,x, N, x_norm, r_norm) |
| (2) | Th_id, Tt_id (Thread id and Total Thread by OS) |
| (3) | St_RowId ← Ed_RowId ← 0 {Start and end ids} |
| (4) | IfTh_id ＜ N mod Tt_ththen |
| (5) | St_RowID ← Th_id∗N/Tt/th+Th_id |
| (6) | Ed_RowID ← St_RowID+N/Tt_th+1 |
| (7) | else |
| (8) | St_RowID ← Th_id ∗ N/Tt_th+Tt_th |
| (9) | Ed_RowID ← St_RowID+N/Tt_th |
| (10) | end_if |
| (11) | r {Local Residual vector for allocated nodes} |
| (12) | error ← 1 |
| (13) | Whileerror > Thresholddo |
| (14) | r_tmp ← x_tmp ← 0 {Local Norms for thread} |
| (15) | foreachi ← St_RowID, Ed,RowIDdo |
| (16) | repeat |
| (17) | r[i-St_RowID]←b[i]-A[i,j]∗x[j] |
| (18) | until For each entry j for row i of A |
| (19) | r_tmp ← r_tmp + r[i-St_RowID]2 |
| (20) | end for |
| (21) | x_norm ← r_norm ← 0 [Global Norms] |
| (22) | barrier |
| (23) | foreachi ← St_RowID, Ed_RowIDdo |
| (24) | x[i] ← x[i] + r[i-St_RowID]/A[i,i] |
| (25) | x_tmp ← x_tmp+x[i]2 |
| (26) | end for |
| (27) | lock |
| (28) | r_norm ←r_norm+r_tmp |
| (29) | x_norm←x_norm+x_tmp |
| (30) | unlock |
| (31) | barrier |
| (32) | error ← |
| (33) | end while |
| (34) | join |
