中国分布式计算论坛

 找回密码
 新注册用户
搜索
查看: 4789|回复: 55

[注意] ========= Changes / Issues with FahCore21 ==========

[复制链接]
发表于 2015-11-23 22:42:07 | 显示全部楼层 |阅读模式
本帖最后由 金鹏 于 2015-11-28 09:22 编辑
大意:
我们发现FahCore21兼容性不太好,尤其是Maxwell二代。所以我们把FahCore21的任务类型改为Advanced了,以免普通用户收到他们。

详见V版的新闻贴   http://www.equn.com/forum/forum.php?mod=redirect&goto=findpost&ptid=25057&pid=549044

鉴于导致FahCore21内核包出错因素不仅仅是频率,尤其是Maxwell二代出错概率较高,不建议980TI/980/970/960/950之类卡跑;

对于Kepler的卡特别是780TI/780显卡跑FahCore21内核包有着惊人的提升,PPD直逼GTX970,建议试试
A卡不受影响,自主选择是否跑FahCore21内核包
以上都需注意做好显卡散热/降频/降压等保护功课
November 20, 2015 by Vijay Pande ·
We’ve discovered that the new FahCore_21 is producing more errors than we consider acceptable for some clients.  The error rate seems to depend on several factors but most noticeable is that it doesn’t work well with second generation Maxwell GPUs.  A few projects have made their way through Advanced testing have been distributed to everyone under the default “FAH” client-type setting.  To allow donors to limit this exposure, those projects have been reclassified as “Advanced” which is appropriate for a FahCore that is still under development.
As has always been the case, the “Advanced” setting will give you access to newer projects which may have a higher error rate.  It is our intention to provide only the safest assignments with the default setting or you can choose to configure your system to run these advanced projects depending on how frequently you encounter these errors.
These conditions are expected to improve as new projects, new versions of that FahCore, or new versions of the drivers incorporate whatever fixes are required.  In the meantime, Work Units which are completed successfully allow scientific research to progress toward even more challenging projects than we’ve done so far.


Filed Under: Uncategorized ·

mTOR: Projects 10491-10499
November 9, 2015 by Steven Albanese ·
In projects 10491-10499, the Chodera lab takes a look at mTOR, a serine/threonine kinase. The MTOR gene was originally discovered in yeast in 1991 and named TOR1/2 because it was the target of rapamycin, an anti-fungal small molecule isolated from the soil of Easter Island in the 1970s. In 1994, the mammalian target of rapamycin (mTOR) was discovered by Drs. Sabatini, Snyder, Abraham, and Schreiber.
mTOR integrates multiple signal inputs to control processes such as cell growth and metabolism, among others. Due to its role in controlling a number of cellular processes, mTOR has clinical significance in neurodegenerative diseases, diabetes and cancer. In the Chodera lab, we are working with the Hsieh lab at MSKCC to understand mTOR’s role in cancer and the development of new and better therapeutics that target it.
Currently, the FDA has approved treatment for metastatic clear cell Renal Cell Carcinoma (ccRCC) that includes mTOR inhibitors such as Everolimus and Temsirolimus. An effort to understand the patient-to-patient variation in response to these drugs by studying how extraordinary responders lead to the characterization of mTOR activating missense mutations in these patients. These mutants cluster in two domains of mTOR: the kinase and FAT domains. These projects will allow us to generate a model of the conformations available to mTOR and ultimately to investigate how these clinically relevant mutations might influence the protein’s structure.
Both the mTOR kinase domain and the larger construct including the FAT domain are very large systems, exploiting the latest OpenMM GPU core (0×21) and push the capabilities of latest-generation GPUs to their full extent.

p9712, p9704 demoted to adv
by mpharrigan » Sat Nov 21, 2015 9:26 am
Hi,
Until more core21 issues can be resolved, these projects are being demoted to adv.

Thanks to all the donors who persevered and produced some great data

Note: If you didn't see a high error rate with these projects, consider setting your client to ADV


发表于 2015-11-23 23:08:49 | 显示全部楼层
对A卡有影响吗,之前Core_21包貌似正常! 现在接了一大堆OX17的包

点评

辛苦费 (「・ω・): 5.0
金版的意思应该是core21对A卡没影响,而且是甜点,A卡应该加Advanced参数跑  发表于 2015-11-24 09:28
辛苦费 (「・ω・): 5
A卡没影响,21内核包是A卡的甜点,注意散热  发表于 2015-11-23 23:10
发表于 2015-11-24 07:31:42 | 显示全部楼层
果然全部不见了~
发表于 2015-11-24 07:43:49 | 显示全部楼层
這對N卡用戶來說很好,不加參數就能完全免接21包了。這樣一來core可以拉高很多。

点评

给力哟!: 5.0 精辟!: 5.0
精辟!: 5
避免混跑造成混乱  发表于 2015-11-24 09:21
给力哟!: 5 精辟!: 5
N卡福音,换头像了哦  发表于 2015-11-24 09:15
发表于 2015-11-24 09:14:16 | 显示全部楼层
难怪上周5接了两个9704之后就没有了,我机智的在周6晚把频率拉到1481,PPD大涨

点评

给力哟!: 5.0
给力哟!: 5
我全加ADV参数专跑21包,看看PPD上不上算  发表于 2015-11-24 09:22
发表于 2015-11-24 10:03:35 | 显示全部楼层
怪不得又能接到9201了,我还是喜欢9201这个包
发表于 2015-11-24 17:32:39 | 显示全部楼层
780Ti早上加了adv参数,接的2个包都是18的:(我喜欢9704
发表于 2015-11-24 17:59:55 | 显示全部楼层
这是啥意思啊,是麦克斯韦卡加了client-type、advanced之后就能收到21的包了吗

点评

精辟!: 5.0
精辟!: 5
你的980TI还是不加参数艹的高,目前21内核包容易莫明奇妙出错  发表于 2015-11-24 21:22
发表于 2015-11-24 18:04:17 | 显示全部楼层
玩家共和国 发表于 2015-11-24 17:59
这是啥意思啊,是麦克斯韦卡加了client-type、advanced之后就能收到21的包了吗

是的,想算core21的就要加advanced参数,像我不想算的就不加参数
发表于 2015-11-24 19:19:06 | 显示全部楼层
Lynt 发表于 2015-11-24 18:04
是的,想算core21的就要加advanced参数,像我不想算的就不加参数

CORE 21好,PPD能到60多万。现在连续算了几个CORE 18的,显卡一直80%几负载,PPD才47万。日了狗。

点评

给力哟!: 5.0
给力哟!: 5
确保不出错,否则爆包得不偿失  发表于 2015-11-24 21:24
发表于 2015-11-25 08:49:07 | 显示全部楼层
玩家共和国 发表于 2015-11-24 19:19
CORE 21好,PPD能到60多万。现在连续算了几个CORE 18的,显卡一直80%几负载,PPD才47万。日了狗。
...

这个要看个人情况:显卡、系统、驱动等,找到合适自己的就好,像我的970在Linux下跑core21,小超到1417还偶尔爆点,并且显卡温度高,很不省心,PPD平均在31~2万,跑core18就能超到1481稳定跑,温度低10度左右,PPD平均能到35万左右。
发表于 2015-11-25 08:59:15 | 显示全部楼层
我的几张卡,没了Core21,PPD提升不少。还是let it go吧~
 楼主| 发表于 2015-11-25 09:30:49 | 显示全部楼层
Lynt 发表于 2015-11-25 08:49
这个要看个人情况:显卡、系统、驱动等,找到合适自己的就好,像我的970在Linux下跑core21,小超到1417还 ...

我的980TI/980/970*2/780TI/780都加了ADV参数跑21内核包(,960和770不加参数跑非21内核包)
在调低频率确保稳定前提下,理论上跑出200万+PPD,比跑非21内核包高出30万左右PPD
可惜不能一直接21内核包,那个9430毒包总来搅局

PS:个人感觉980TI/980跑21内核包PPD都是60+/40+万节奏,特别是780TI/780居然都是33+/30+万节奏,直接触摸970的节奏

@guihuo

点评

看来9430在LINUX下甜包,在WIN下毒包  发表于 2015-11-25 10:28
9430是我喜欢的包之一,PPD36万+,一个包算很久,上传下载少  发表于 2015-11-25 09:39

评分

参与人数 1基本分 +10 收起 理由
yimu35 + 10 感謝分享

查看全部评分

发表于 2015-11-25 10:37:55 | 显示全部楼层
金鹏 发表于 2015-11-25 09:30
我的980TI/980/970*2/780TI/780都加了ADV参数跑21内核包(,960和770不加参数跑非21内核包)
在调低频率确保 ...

我这边还有A卡顶半边天,17 18 21混跑的时候,总PPD上摸920K,下探710K,好虐心,总以为那张卡卡死了,可看看又没有。
现在没了core21,基本上860K~920K。这样算下来,core21还是不要不要的好呢~

点评

精辟!: 5.0
精辟!: 5
21内核包是A卡甜点啊!是不是超过头假死卡住了?  发表于 2015-11-25 10:41
发表于 2015-11-25 10:56:44 | 显示全部楼层
GM200跑Adv現在什麼Core 頻率比較合適? 1400Mhz?
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~

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

GMT+8, 2020-4-4 13:45

Powered by Discuz! X3.4

© 2001-2017 Comsenz Inc.

快速回复 返回顶部 返回列表