| 建置時間 | April 2017 | ||
| 主機架構 | CPU nodes | GPU nodes | Big Memory nodes |
| 硬體規格 | 44組計算節點 1056組計算核心 |
2組計算節點 48組計算核心 4組GPU核心 |
6組計算節點 192組計算核心 |
| 計算節點 | Intel E5-2650 v4 × 2 (2.2 GHz, 12 cores) RAM DDR4-2400 (128 GB) |
Intel E5-2650 v4 × 2 (2.2 GHz, 12 cores) RAM DDR4-2400 (128 GB) nVidia Tesla K40 × 2 |
Intel E5-2697A v4 × 2 (2.6 GHz, 16 cores) RAM DDR4-2400 (128 GB) × 4組節點 RAM DDR4-2400 (1 TB) × 2組節點 |
| 網路連線 | Infiniband FDR (56 Gb/s) | ||
| 作業系統 | Linux (Centos 7.x) | ||
| 排程軟體 | Torque, Maui | ||
| 編譯程式 | GCC, Intel, PGI, Python 2.x/3.x | ||
| 資料庫 | CUDA Intel MKL, LAPACK MySQL NetCDF, PnetCDF NumPy, SciPy OpenMPI, MVAPICH2 |
||
| 安裝軟體 | Amber, CP2K Matlab Quantum Expresso, VASP(使用者需有License) SAS |
||
| 單價(SU) | 教育及學術單位 0.05 NT/core-hour | 企業單位 1.20 NT/core-hour |
HPL 2.2 Parallel Linpack (Software Source from NETLIB)
Machine |
HPL 2.2 (GFLOPS) |
CPU node (44 nodes) |
29,339 |
BIGMEM node (6 nodes) |
6,635 |
GPU node (2 nodes) (CPU contribution) |
1,392 |
HPCG
count |
normal |
rating |
1 |
10 |
10.14834 |
2 |
20 |
20.29668 |
3 |
30 |
30.44502 |
4 |
40 |
40.59336 |
5 |
50 |
50.74170 |
6 |
60 |
60.89005 |
7 |
70 |
71.03839 |
8 |
80 |
81.18673 |
9 |
90 |
91.33507 |
10 |
100 |
101.48341 |
11 |
110 |
111.63175 |
12 |
120 |
121.78009 |
13 |
130 |
131.92843 |
14 |
140 |
142.07677 |
15 |
150 |
152.22511 |
16 |
160 |
162.37345 |
17 |
170 |
172.52180 |
18 |
180 |
182.67014 |
19 |
190 |
192.81848 |
20 |
200 |
202.96682 |
21 |
210 |
213.11516 |
22 |
220 |
223.26350 |
23 |
230 |
233.41184 |
24 |
240 |
243.56018 |
25 |
250 |
253.70852 |
26 |
260 |
263.85686 |
27 |
270 |
274.00520 |
28 |
280 |
284.15355 |
29 |
290 |
294.30189 |
30 |
300 |
304.45023 |
31 |
310 |
314.59857 |
32 |
320 |
324.74691 |
33 |
330 |
334.89525 |
34 |
340 |
345.04359 |
35 |
350 |
355.19193 |
36 |
360 |
365.34027 |
37 |
370 |
375.48861 |
38 |
380 |
385.63695 |
39 |
390 |
395.78530 |
40 |
400 |
405.93364 |
41 |
410 |
416.08198 |
42 |
420 |
426.23032 |
43 |
430 |
436.37866 |
44 |
440 |
446.52700 |
LAMMPS Scaling
MD for bulk water, 37×37×37 A3, (1728 molecules)
Node |
Time(sec) |
Speed-Up |
Efficientcy(%) |
1 |
1120 |
1.00 |
100% |
2 |
634 |
1.77 |
57% |
4 |
407 |
2.75 |
36% |
8 |
311 |
3.60 |
28% |
16 |
310 |
3.61 |
28% |
VASP 5.4.24
Cu 32, PAW_PBE
Cores |
NPAR |
CPU_TIME |
24 |
4 |
138.686 |
48 |
6 |
129.733 |
72 |
8 |
91.410 |
