RE: About CUDA version

From: Vermaas, Joshua (Joshua.Vermaas_at_nrel.gov)
Date: Tue Sep 25 2018 - 15:06:54 CDT

Enabling CUDA support is an explicit switch from what I recall. You activate it with "--with-cuda" when configuring NAMD. A non-CUDA build is just a matter of compiling without having given the flag to configure.

-Josh

On 2018-09-25 13:49:11-06:00 owner-namd-l_at_ks.uiuc.edu wrote:

OK I know that. Binaries have several options. I was thinking about
compiling from source and then use a switch to enable/disable the CUDA
capability. I didn't see that in the manual. Maybe I missed the point.

Regards,
Mahmood

On Tue, Sep 25, 2018 at 10:34 PM Bassam Haddad <bhaddad_at_pdx.edu> wrote:
&gt;
&gt; Don’t use the cuda version of the NAMD.
&gt;
&gt; On Tue, Sep 25, 2018 at 11:27 AM Mahmood Naderan <mahmood.nt_at_gmail.com> wrote:
&gt;&gt;
&gt;&gt; Hi
&gt;&gt; With the following command
&gt;&gt;
&gt;&gt; /opt/NAMD_2.13b1_Linux-x86_64-multicore-CUDA/namd2 +p8 2_md.conf
&gt;&gt;
&gt;&gt; I see that it automatically utilizes the GPU. I want to know, how can
&gt;&gt; I force namd to use CPU instead of GPU?
&gt;&gt;
&gt;&gt;
&gt;&gt; Regards,
&gt;&gt; Mahmood
&gt;&gt;

</mahmood.nt_at_gmail.com></bhaddad_at_pdx.edu>

This archive was generated by hypermail 2.1.6 : Tue Dec 31 2019 - 23:20:14 CST