Re: [AMBER] pmemd.cuda test possible failure

From: David A Case via AMBER <amber.ambermd.org>
Date: Tue, 28 Feb 2023 08:38:52 -0500

On Thu, Feb 23, 2023, Freddy Bernal via AMBER wrote:
>
>I have compiled Amber22 with CUDA support but when running the tests
>(test.cuda.serial), I got some possible failures. The manual
>says that minor differences can be ignored, but what is minor? I got the
>following range in relative errors. Are those in red
>still ok? If not, what should I do? Attached the diff file.
>
>### Maximum relative error in matching lines = 2.97e+00 at line 199 field 3
>### Maximum relative error in matching lines = 4.03e-01 at line 197 field 3
>### Maximum relative error in matching lines = 5.99e+01 at line 221 field 3

These look like quite serious errors. Can you provide (a pointer to) the
actual log files? What GPU card do you have, and what OS and compiler
versions are you using?

...thanks...dac


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Feb 28 2023 - 06:00:02 PST
Custom Search