| ²é¿´: 560 | »Ø¸´: 4 | ||
| ¡¾½±Àø¡¿ ±¾Ìû±»ÆÀ¼Û3´Î£¬×÷ÕßxirainbowÔö¼Ó½ð±Ò 1.5 ¸ö | ||
| µ±Ç°Ö÷ÌâÒѾ´æµµ¡£ | ||
xirainbowľ³æ (ÕýʽдÊÖ)
|
[×ÊÔ´]
¡¾·ÖÏí¡¿Reduce memory usage
|
|
|
ÎÄÕÂÀ´Ô´£ºhttp://blog.sina.com.cn/s/blog_5f15ead20100eiyv.html There are a few measures you can take to reduce the memory usage: 1. if you are using pools, reduce the number of pools, it requires more communication between nodes, but reduces the memory usage. 2. if you are using norm-conserving pseudopotentials try ultrasoft or PAW instead; more in general try to do more accurate test on convergence with respect to wfc and rho cutoff 3. if you have a slab geometry you may still have good results reducing the amount of vacuum (you'll have to test the convergence wrt the amount of vacuum) 4. if you are using a computation cluster, run the code on more processors 5. if you are using a PC, buy more RAM 6. us cg instead of david diagonalization, but only as a last measure |
» ²ÂÄãϲ»¶
²ÄÁÏ¿ÆÑ§Ó빤³Ì320Çóµ÷¼Á£¬080500
ÒѾÓÐ4È˻ظ´
0703µ÷¼Á£¬Ò»Ö¾Ô¸Ìì½ò´óѧ319·Ö
ÒѾÓÐ17È˻ظ´
Ò»Ö¾Ô¸211£¬0703»¯Ñ§305·ÖÇóµ÷¼Á
ÒѾÓÐ5È˻ظ´
0703»¯Ñ§µ÷¼Á 348·Ö
ÒѾÓÐ6È˻ظ´
273Çóµ÷¼Á
ÒѾÓÐ11È˻ظ´
308Çóµ÷¼Á
ÒѾÓÐ5È˻ظ´
²ÄÁÏÓ뻯¹¤363ÇóÍÆ¼ö
ÒѾÓÐ11È˻ظ´
²ÄÁϵ÷¼Á
ÒѾÓÐ12È˻ظ´
ÉúÎïѧµ÷¼Á ¿Éµ÷¼Áµ½ÉúÎïÓëÒ½Ò©
ÒѾÓÐ3È˻ظ´
304Çóµ÷¼Á
ÒѾÓÐ11È˻ظ´
2Â¥2009-09-14 11:54:09
enola
¾èÖú¹ó±ö (ÕýʽдÊÖ)
- 1STÇ¿Ìû: 2
- Ó¦Öú: 13 (СѧÉú)
- ½ð±Ò: 1485.9
- Ìû×Ó: 847
- ÔÚÏß: 102.5Сʱ
- ³æºÅ: 947564
5Â¥2010-04-13 18:44:09
¼òµ¥»Ø¸´
zxzj053Â¥
2009-09-17 09:50
»Ø¸´


Ö§³Ö£¡£¡£¡
freshgirl4Â¥
2009-09-17 10:11
»Ø¸´














»Ø¸´´ËÂ¥