On Thu, Apr 26, 2018, Kadir Ozcan wrote:
>
> I’m using AmberTools17 with Amber16. Because our institution recently
> purchased Amber16, we will be requesting the upgrade soon so I try
> this process again with AmberTools18 with Amber18. I will try the
> configuration and installation process with Amber18 once we receiver it
> and get back to you. I appreciate all of the help.
Thanks for the information.
Here's what has happened: the pbsa and cpptraj programs in AmberTools
originally included "-lcuda" as a library to be linked when building the
CUDA versions. I don't know if this was needed, but it didn't do any
harm.
More recent versions of the NVIDIA toolkit (certainly the 9.x family)
exit with the error message you saw. We've fixed this in AmberTools18,
but have not (yet?) backported this to earlier versions of AmberTools.
One reason is that we don't have the people power to test older versions
of AmberTools with releases of NVIDIA software that came out
subsequently. In this case, I think it is OK to just remove the "-cuda"
flags, but be aware that there might be other problems.
....regards...dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Apr 27 2018 - 05:30:04 PDT
