Default Branch

8c1f4d8108 · 迁移C算子的循环融合和临时量消除 · Updated 2026-03-03 16:20:15 +08:00

Branches

9c44d1c885 · fix(bssn_rhs) · Updated 2026-03-03 16:00:45 +08:00    ianchb

5
3

f1fe9fd443 · 迁移C算子的循环融合和临时量消除 · Updated 2026-03-03 15:57:10 +08:00    gh0s7

7
6

12e1f63d50 · prolong3: 减少Z-pass 冗余计算 · Updated 2026-03-02 21:20:49 +08:00    jaunatisblue

19
4

43975017eb · prolong3 改为先算实际 stencil 窗口;只有窗口触及对称边界时才走全域 symmetry_bd,否则只复制必需窗口。restrict3 同样改成窗口判定,无触边时仅填 ii/jj/kk 必需窗口。 · Updated 2026-03-02 18:10:38 +08:00    ianchb

46
34

e11363e06e · Optimize fdderivs: skip redundant 2nd-order work in 4th-order overlap · Updated 2026-03-02 03:21:21 +08:00    jaunatisblue

16
0
Included

19b0e79692 · 黄老板逆天重写 · Updated 2026-03-01 05:48:40 +08:00    jaunatisblue

75
1

588fb675a0 · 尝试划分4block但是效果不好,转为研究访存 · Updated 2026-02-28 21:17:02 +08:00    jaunatisblue

69
4

e0b5e012df · 引入 PGO 式两遍编译流程,将 Interp_Points 负载均衡优化合法化 · Updated 2026-02-27 15:10:22 +08:00    gh0s7

52
0

f7ada421cf · skip redundant MPI ghost cell syncs for stages 0, 1 & 2 · Updated 2026-02-26 16:16:33 +08:00    ianchb

59
2

f147f79ffa · 修改block划分,对负载高的rank所在block进行划分,添加到空rank,空rank是平移得到的 · Updated 2026-02-26 09:40:46 +08:00    jaunatisblue

69
2

cc06e30404 · Apply async Sync optimization to Z4c_class using Sync_start/finish pattern · Updated 2026-02-20 09:58:26 +08:00    ianchb

69
3

b32675ba99 · 1. Pass 1(357-395行):遍历所有 Patch,对每个 block 计算含ghost zone 的实际体积,存入 block_volumes · Updated 2026-02-12 03:22:46 +08:00    jaunatisblue

87
6

714c6e90c6 · Add OpenMP parallelization to Fortran compute kernels · Updated 2026-02-10 23:40:17 +08:00    gh0s7

76
2

8b68b5d782 · fixup! Fix load explosion: use subprocess for binary data plots to avoid thread conflict · Updated 2026-02-09 23:00:17 +08:00    ianchb

84
7

86704100ec · Only enable OpenMP for TwoPunctures · Updated 2026-02-08 23:36:12 +08:00    jaunatisblue

87
4

3f7e20f702 · 删除diff_new.f90中冗余部分,方便后续工作 · Updated 2026-02-08 00:54:23 +08:00    jaunatisblue

103
2

6796384bf4 · taskset setting updated · Updated 2026-02-07 22:24:02 +08:00    gh0s7

90
2

c6e4d4ab71 · Add OpenMP parallelization to BSSN RHS hot-path stencil routines · Updated 2026-02-07 13:58:55 +08:00    gh0s7

90
1

4eb698f496 · Add MPI+OpenMP hybrid parallelism (48 ranks x 2 threads) for full 96-core utilization · Updated 2026-02-06 15:53:15 +08:00    gh0s7

94
1

223ec17a54 · input updated · Updated 2026-02-06 13:57:48 +08:00    gh0s7

94
0
Included