找回密码
 新注册用户
搜索
查看: 9245|回复: 20

[新闻] Projects 11703-11707 [core21, GPU, LIN/WIN] moving to FAH

[复制链接]
发表于 2016-1-9 09:13:21 | 显示全部楼层 |阅读模式
本帖最后由 金鹏 于 2016-5-4 20:02 编辑

接到11704包,在0.017内核下(关闭了PME)跑出的PPD中规中矩.好怀念0.016内核下的超高PPD


捕获1.PNG

捕获.PNG


1.PNG

捕获.PNG

2.PNG

3.PNG


p11706 and p11707 moving to FAH
by rafwiewiora » Wed May 04, 2016 3:33 am
11706:
Here we are simulating the NSD2 protein methyltransferase, a continuation of the Chodera Lab effort to map the conformational landscapes of all protein lysine methyltransferases - important epigenetic enzymes. This will provide powerful data for validations of existing chemical probes and inhibitors of this protein class, as well as will facilitate new explorations of the chemical space for further drug candidates.

NSD2's mutations and chromosomal translocations are implicated in reorganizing the epigenetic landscape in cancers such as multiple myelomas and pediatric acute lymphoblastic leukemia. [doi:10.1016/j.molcel.2011.08.042; doi:10.1038/ng.2777; doi:/10.1371/journal.pgen.1004566].

On interesting things about the simulation - we're continuing using the 4-dummy atom zinc atom model [http://www.mayo.edu/research/lab ... -atom-cada-approach] to model structural zinc clusters in these proteins.

Stats:
# atoms: 62560
credit: 21503
k-factor: 0.75
timeout: 7d
deadline: 10d

Mixed precision calculations are enforced (DP for motion integrations, SP for force calculations). All GPUs lacking DP support are excluded.

11707:
Stats:
# atoms: 35200
credit: 15965
k-factor: 0.75
timeout: 7d
deadline: 10d

11707: Protein lysine methyltransferase SETD8, we have used Ensembler (http://ensembler.readthedocs.org/en/latest/) to generate models from all chains in all available crystal structures - 23 in total. This is the same protein as project 10478, where we started only from two conformations. While I’m analyzing the 1ms (very nice!) of data we’ve gathered there, we’re starting this also with more heterogenous initial conformation seeding. This is part of a bigger ‘operation SETD8’ - next project coming will be a study of this protein in all states of its catalytic cycle to understand changes it undergoes during enzymatic activity and binding of other species.

We have 23 RUNs here, mixed precision is enforced (no-DP GPUs excluded).

Happy Folding! Thanks!
11706 released to ADVANCED
by rafwiewiora » Thu Apr 14, 2016 4:32 am
Here we are simulating the NSD2 protein methyltransferase, a continuation of the Chodera Lab effort to map the conformational landscapes of all protein lysine methyltransferases - important epigenetic enzymes. This will provide powerful data for validations of existing chemical probes and inhibitors of this protein class, as well as will facilitate new explorations of the chemical space for further drug candidates.

NSD2's mutations and chromosomal translocations are implicated in reorganizing the epigenetic landscape in cancers such as multiple myelomas and pediatric acute lymphoblastic leukemia. [doi:10.1016/j.molcel.2011.08.042; doi:10.1038/ng.2777; doi:/10.1371/journal.pgen.1004566].

On interesting things about the simulation - we're continuing using the 4-dummy atom zinc atom model [http://www.mayo.edu/research/lab ... -atom-cada-approach] to model structural zinc clusters in these proteins.

Stats:
# atoms: 62560
credit: 21503
k-factor: 0.75
timeout: 7d
deadline: 10d

Mixed precision calculations are enforced (DP for motion integrations, SP for force calculations). All GPUs lacking DP support are excluded.
Project 11707 [core21, GPU, LIN/WIN]
by rafwiewiora » Tue Apr 12, 2016 10:06 pm
11707 joining BETA!

Stats:
# atoms: 35200
credit: 15965
k-factor: 0.75
timeout: 7d
deadline: 10d

11707: Protein lysine methyltransferase SETD8, we have used Ensembler (http://ensembler.readthedocs.org/en/latest/) to generate models from all chains in all available crystal structures - 23 in total. This is the same protein as project 10478, where we started only from two conformations. While I’m analyzing the 1ms (very nice!) of data we’ve gathered there, we’re starting this also with more heterogenous initial conformation seeding. This is part of a bigger ‘operation SETD8’ - next project coming will be a study of this protein in all states of its catalytic cycle to understand changes it undergoes during enzymatic activity and binding of other species.

We have 23 RUNs here, mixed precision is enforced (no-DP GPUs excluded).

Thanks everyone for testing!
Project 11706 [core21, GPU, LIN/WIN]
by rafwiewiora » Tue Apr 05, 2016 12:24 pm
Project 11706 is finishing it's time in INTERNAL and will be moving into BETA on Wed, April 6th, 11am EST.

Here we are simulating the NSD2 protein methyltransferase, a continuation of the Chodera Lab effort to map the conformational landscapes of all protein lysine methyltransferases - important epigenetic enzymes. This will provide powerful data for validations of existing chemical probes and inhibitors of this protein class, as well as will facilitate new explorations of the chemical space for further drug candidates.

NSD2's mutations and chromosomal translocations are implicated in reorganizing the epigenetic landscape in cancers such as multiple myelomas and pediatric acute lymphoblastic leukemia. [doi:10.1016/j.molcel.2011.08.042; doi:10.1038/ng.2777; doi:/10.1371/journal.pgen.1004566].

On interesting things about the simulation - we're continuing using the 4-dummy atom zinc atom model [http://www.mayo.edu/research/labs/computer-aided-molecular-design/projects/zinc-protein-simulations-using-cationic-dummy-atom-cada-approach] to model structural zinc clusters in these proteins.

Stats:
# atoms: 62560
credit: 21503
k-factor: 0.75
timeout: 7d
deadline: 10d


Mixed precision calculations are enforced (DP for motion integrations, SP for force calculations). All GPUs lacking DP support should be excluded. Assignment Server is running the following constraints:
CODE: SELECT ALLGPU!=None
(
  [OS=Linux]
  [OS=Win32 CPU=AMD64 OSSpecies != WIN_XP]
)

(
  [GPU=NVIDIA GPUSpecies>=2]
  [GPU=ATI GPUSpecies>=5]
)
MaxPacket>5


Thanks to everyone for your contributions!
Projects 11703-5 [core21, GPU, LIN/WIN] moving to Advanced
by rafwiewiora » Fri Jan 15, 2016 10:10 am
Projects 11703, 11704, 11705 moving into ADV!

These are simulations of epigenetic enzymes - protein lysine methyltransferases NSD1 (11703), NSD3 (11704) and SETD2 (11705) using a new multisite Zn2+ ion model.

Stats:
k-factor is 0.75 for all of them

Project 11703
base credit: 24051
timeout: 7d
deadline: 10d
atoms: 68556

Project 11704
base credit: 17402
timeout: 7d
deadline: 10d
atoms: 52855

Project 11705
base credit: 20645
timeout: 7d
deadline: 10d
atoms: 61463

Thanks a lot everyone!
Projects 11703-5 [core21, GPU, LIN/WIN]
by rafwiewiora » Sat Jan 09, 2016 4:51 am
Projects 11703, 11704, 11705 now in beta!

These are simulations of protein lysine methyltransferases NSD1 (11703), NSD3 (11704) and SETD2 (11705) using a new multisite Zn2+ ion model. Note that 11703 and 11704 are replacing 11700 and 11701, which were previously in beta but have been deprecated due to an equilibration / stability issue.

Stats:
k-factor is 0.75 for all of them

Project 11703
base credit: 24051
timeout: 7d
deadline: 10d

Project 11704
base credit: 17402
timeout: 7d
deadline: 10d

Project 11705
base credit: 20645
timeout: 7d
deadline: 10d

Thanks a lot for testing!

评分

参与人数 1基本分 +15 收起 理由
zflowers + 15 灰常好!

查看全部评分

回复

使用道具 举报

发表于 2016-1-9 10:53:14 | 显示全部楼层
感谢金版主分享,我现在只要不接到9201就很开心了, 神马18/21核都可以。。。
回复

使用道具 举报

发表于 2016-1-9 10:53:48 | 显示全部楼层
金版您说每新出一个运算包就代表斯坦福新研究一种疾病吗
回复

使用道具 举报

发表于 2016-1-9 23:07:40 | 显示全部楼层
跑了一个11703 @1300mhz ,ppd只有57万,tpf 3:24.
回复

使用道具 举报

 楼主| 发表于 2016-1-10 08:57:41 | 显示全部楼层
yimu35 发表于 2016-1-9 23:07
跑了一个11703 @1300mhz ,ppd只有57万,tpf 3:24.

貌似对频率更敏感些,否则即使1330下的TTX也应该上60万+
回复

使用道具 举报

 楼主| 发表于 2016-1-16 22:19:27 | 显示全部楼层


Projects 11703-5 [core21, GPU, LIN/WIN] moving to Advanced
by rafwiewiora » Fri Jan 15, 2016 10:10 am
Projects 11703, 11704, 11705 moving into ADV!

These are simulations of epigenetic enzymes - protein lysine methyltransferases NSD1 (11703), NSD3 (11704) and SETD2 (11705) using a new multisite Zn2+ ion model.

Stats:
k-factor is 0.75 for all of them

Project 11703
base credit: 24051
timeout: 7d
deadline: 10d
atoms: 68556

Project 11704
base credit: 17402
timeout: 7d
deadline: 10d
atoms: 52855

Project 11705
base credit: 20645
timeout: 7d
deadline: 10d
atoms: 61463

Thanks a lot everyone!


回复

使用道具 举报

 楼主| 发表于 2016-4-8 13:35:21 | 显示全部楼层
本帖最后由 金鹏 于 2016-4-8 13:38 编辑



Project 11706 [core21, GPU, LIN/WIN]
by rafwiewiora » Tue Apr 05, 2016 12:24 pm
Project 11706 is finishing it's time in INTERNAL and will be moving into BETA on Wed, April 6th, 11am EST.

Here we are simulating the NSD2 protein methyltransferase, a continuation of the Chodera Lab effort to map the conformational landscapes of all protein lysine methyltransferases - important epigenetic enzymes. This will provide powerful data for validations of existing chemical probes and inhibitors of this protein class, as well as will facilitate new explorations of the chemical space for further drug candidates.

NSD2's mutations and chromosomal translocations are implicated in reorganizing the epigenetic landscape in cancers such as multiple myelomas and pediatric acute lymphoblastic leukemia. [doi:10.1016/j.molcel.2011.08.042; doi:10.1038/ng.2777; doi:/10.1371/journal.pgen.1004566].

On interesting things about the simulation - we're continuing using the 4-dummy atom zinc atom model [http://www.mayo.edu/research/labs/computer-aided-molecular-design/projects/zinc-protein-simulations-using-cationic-dummy-atom-cada-approach] to model structural zinc clusters in these proteins.

Stats:
# atoms: 62560
credit: 21503
k-factor: 0.75
timeout: 7d
deadline: 10d


Mixed precision calculations are enforced (DP for motion integrations, SP for force calculations). All GPUs lacking DP support should be excluded. Assignment Server is running the following constraints:
CODE: SELECT ALLGPU!=None
(
  [OS=Linux]
  [OS=Win32 CPU=AMD64 OSSpecies != WIN_XP]
)

(
  [GPU=NVIDIA GPUSpecies>=2]
  [GPU=ATI GPUSpecies>=5]
)
MaxPacket>5


Thanks to everyone for your contributions!

回复

使用道具 举报

发表于 2016-4-8 14:19:25 | 显示全部楼层
看来Pascal出的真是时候。。。FP64以后FAH也需要了?
回复

使用道具 举报

 楼主| 发表于 2016-4-8 15:20:39 | 显示全部楼层
yimu35 发表于 2016-4-8 14:19
看来Pascal出的真是时候。。。FP64以后FAH也需要了?

也许在传说中的CORE22内核中使用,配合帕斯卡能飞出百万+PPD?
回复

使用道具 举报

发表于 2016-4-8 16:55:38 | 显示全部楼层
金鹏 发表于 2016-4-8 15:20
也许在传说中的CORE22内核中使用,配合帕斯卡能飞出百万+PPD?

一百万真心不高啊,980ti 都能8x万。。。 希望起码能上120+
回复

使用道具 举报

发表于 2016-4-8 17:06:45 | 显示全部楼层
yimu35 发表于 2016-4-8 16:55
一百万真心不高啊,980ti 都能8x万。。。 希望起码能上120+

现在能上80万的包很少了,平均在70W~75W就很不错了,这两天core18又冒出来很多,好在确实温度低不少,夏天会好过些
回复

使用道具 举报

发表于 2016-4-29 17:06:36 | 显示全部楼层
11707  1202mhz 55万左右
回复

使用道具 举报

 楼主| 发表于 2016-4-30 01:02:57 | 显示全部楼层
yimu35 发表于 2016-4-29 17:06
11707  1202mhz 55万左右

是TTX跑的?
看来是吃高频的包
回复

使用道具 举报

发表于 2016-9-8 12:12:40 | 显示全部楼层
11707 40万左右

评分

参与人数 1基本分 +20 收起 理由
金鹏 + 20 辛苦了!

查看全部评分

回复

使用道具 举报

 楼主| 发表于 2016-9-8 14:26:35 | 显示全部楼层

可以小超一下到1400-1450,那样PPD肯定上50万
回复

使用道具 举报

您需要登录后才可以回帖 登录 | 新注册用户

本版积分规则

论坛官方淘宝店开业啦~

Archiver|手机版|小黑屋|中国分布式计算总站 ( 沪ICP备05042587号 )

GMT+8, 2024-3-29 05:41

Powered by Discuz! X3.5

© 2001-2024 Discuz! Team.

快速回复 返回顶部 返回列表