Dear Developers,
I am currently running REST2 simulations using GROMACS patched with PLUMED and have observed inconsistent replica exchange acceptance ratios across different GPU cards. Below are the details:
GPUs Used & Software Versions Used:
1. Tesla P100-PCIE : GROMACS 2022 patched with PLUMED 2.9.0
2. Ada RTX 6000 : GROMACS 2024 patched with PLUMED 2.9.3
3. GeForce RTX 3090 : GROMACS 2022 patched with PLUMED 2.9.0
With Tesla P100 an exchange frequency of 800, the acceptance ratio is around 40–50% for all replicas. While using Ada RTX 6000 card, the exchange frequency of replicas is 0, do not exchange. Only at a frequency of 1 does a replica exchange occur. Exchange frequency with GeForce RTX 3090 above 390, some replicas exchange while others do not. With frequencies below 390, the acceptance ratio is around 40–50% for all replicas.
Since the same simulation setup and input files are being used, I am not getting why the behavior differs significantly across GPUs and GROMACS/PLUMED versions.
Best regards,
Munazzah
PhD student
On 29 Jul 2025, at 06:40, Munazzah Ansari <ansarimu...@gmail.com> wrote:
--
You received this message because you are subscribed to the Google Groups "PLUMED users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/plumed-users/CAGqfTa0oXwt6XODvbd6gMW0_W%3DzFQaSUKP2_8oL4NKX9Sp%2BHCg%40mail.gmail.com.