|
|
本帖最后由 金鹏 于 2016-1-19 20:08 编辑
强制更新21内核到0.017版本
实测0.017内核比0.016内核同频同机跑同包低20%的PPD(N卡关闭PME差距真大啊)
跑出惨不忍睹的PPD
 by msultan » Tue Jan 19, 2016 12:54 pm
p9167-9172 moving to advancedby msultan » Sat Jan 09, 2016 5:45 am Hello all,
I am introducing 6 new core21 gpu projects. These projects aim to increase our understanding of how imatinib, a blockbuster anti cancer drug, binds to its various different kinase targets. In this series, I am looking at its binding to Abl(1967,1968), Src(1969,1970) and Syk(1971,1972). The bigger of number in all 3 corresponds to kinase configurations that have a larger number of water molecules and thus they have larger number of points. Given the variation in water molecule, I suspect there might be some point variation but they should average out to be within 15% across multiple runs within the same project. Let me know if there are any instabilities or issues. Thanks for all the help everyone and happy new years!
Details for 9167,9169,9171:
PPD:3700
Core:21
OS:Win/Linux
GPU:Nvidia/ATI
Deadline:10
Timeout:8
k-factor:0.75
SystemSize: ~35000-45000
Details for 9168,9170,9172:
PPD:5300
Core:21
OS:Win/Linux
GPU:Nvidia/ATI
Deadline:10
Timeout:8
k-factor:0.75
SystemSize: ~45000-55000
Thanks,
Munee New Core21 Beta projects 9167-9172by msultan » Sat Dec 26, 2015 2:23 pm Hello all,
I am introducing 6 new core21 gpu projects. These projects aim to increase our understanding of how imatinib, a blockbuster anti cancer drug, binds to its various different kinase targets. In this series, I am looking at its binding to Abl(1967,1968), Src(1969,1970) and Syk(1971,1972). The bigger of number in all 3 corresponds to kinase configurations that have a larger number of water molecules and thus they have larger number of points. Given the variation in water molecule, I suspect there might be some point variation but they should average out to be within 15% across multiple runs within the same project. Let me know if there are any instabilities or issues. Thanks for all the help everyone and happy holidays!
Details for 9167,9169,9171:
PPD:3700
Core:18
OS:Win/Linux
GPU:Nvidia/ATI
Deadline:10
Timeout:8
k-factor:0.75
SystemSize: ~35000-45000
Details for 9168,9170,9172:
PPD:5300
Core:21
OS:Win/Linux
GPU:Nvidia/ATI
Deadline:10
Timeout:8
k-factor:0.75
SystemSize: ~45000-55000
Thanks,
Muneeb
ps I am gonna be traveling for the next few days so I wont be able to fix any points issue until i get back around the 31st of Dec 2015. If these projects are unstable, please send an email to msultan@stanford.eduand I will have someone from the pande group remove these projects from assign.
|
|
发表于 2015-12-26 14:58:26
