|
|
本帖最后由 金鹏 于 2015-12-19 13:25 编辑
暂时不建议兄弟们尝试,我先来做做小白鼠,降低了性能提升稳定性?
FahCore_21 Version 0.0.14.rar
(2.34 MB, 下载次数: 27)
新的0.014版21内核稳定性提高,PPD降低约10+%
core21是64位内核,效率很高,是2016年的主力内核,
现有17/18内核包的后续项目,可能会直接采用21内核(或者增强版的22内核)计算,而不是沿用17/18内核,
总之丢弃幻想,迎接core21(或者增强版的22内核)全面普及 - 980TI在1470MHZ下跑9630包,用0.012内核跑TPF=1:06,用0.014内核跑TPF=1:14, PPD从72万降到60万,降低约15%(性能损失远大于8.5%)
- 980TI在1370MHZ下跑9643包,用0.012内核跑TPF=1:09,用0.014内核跑TPF=1:19, PPD从67万降到55万,降低约18%(性能损失远大于8.5%)
Core21 v0.0.14 has been released!We rolled out Core21 v0.0.14 last week.
Main features are:
* Bad State errors fixed: Improvements that drastically reduce the high rate of Bad State errors we were seeing with earlier versions of the core, especially with NVIDIA cards.
These Bad State errors were ironically caused by a couple of bugs in the code that checks the simulation for integrity every time a checkpoint file is written. The simulation integrity was solid, but there was a high false positive rate for errors. This should be greatly reduced with 0.0.14.
For NVIDIA cards, this unfortunately comes with a 10% performance regression when PME is in use, but we are compensating by rebenchmarking affected core 21 projects. We hope to release an update that will undo this performance regression soon.
* More debugging info: When Bad State errors *do* occur, we now bring back more information to help us diagnose these issues.
* Early NaN detection eliminates slowdowns: We previously had some reports of WU slowdowns that were traced to NaNs appearing in the simulation and slowing down the integration loop drastically. Previously, these were not detected until a checkpoint was written. Now, checks are performed much more frequently, hopefully eliminating these slowdowns.
* Minor improvements: We have also made a number of other minor improvements that give us a bit more flexibility and control over what simulations can be run.
Thanks to everyone who helped make this possible, especially Peter Eastman (OpenMM developer extraordinaire!) and all of the testers who helped kick the tires, including (but by no means limited to): New Core 21 build 0.0.13 live in BETA!
15 posts • Page 1 of 1
New Core 21 build 0.0.13 live in BETA!We have a new Core 21 build (0.0.13) live in BETA as of today.
Many thanks to donors on Slack (many of which are on Folding Forum too) who helped us test the core before making it live, especially (but not limited to!):
CODE: SELECT ALLchristianvirtual
davidcoton
bollix47
grandpa
bruce
jimerickson
frodothehobbit
This core should fix the vast majority of Bad State errors we have been seeing with Core 21, and when errors do occur in the future, more information to help us debug the cause will be brought back to our servers.
The current hotfix comes at the cost of ~8.5% performance degradation on NVIDIA cards for systems that use PME. We'll be working hard to push out an updated core that fixes the performance regression in the coming weeks.
Please bear with us as we rebenchmark projects that will be impacted by this performance change.
Please report any issues you see with core build 0.0.13 here!
|
评分
-
查看全部评分
|
发表于 2015-12-3 23:49:52
