GPU acceleration using K40 and k80

From: Atanu Maity (atanuchem48_at_gmail.com)
Date: Wed Sep 14 2016 - 00:32:00 CDT

• Next message: Nicolas Martin: "Running REMD"
• Previous message: Kevin C Chan: "Re: Umbrella sampling colvars DistanceZ deviates a lot"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]

Dear all,

I have few queries regarding the acceleration of NAMD simulation using K40
and K80 tesla GPU cards --

i) How faster will be a NAMD job by upgrading from K40 to K80 when same
number of CPU cores are used? Is there any available benchmark ?

ii) What is the optimal CPU:GPU ratio to be used with K40 or K80 to get
maximum speed up. For example using K20, 4:1 CPU:GPU ratio is optimal
(*https://hpc.nih.gov/apps/namd/
<https://hpc.nih.gov/apps/namd/>*). The same benchmark we have found with
C2075 also.

Any suggestion will help us to equip our workstation with proper number of
GPU.

With regards,
Atanu Maity
Research Scholar
Bioinformatics Center
Bose Institute

• Next message: Nicolas Martin: "Running REMD"
• Previous message: Kevin C Chan: "Re: Umbrella sampling colvars DistanceZ deviates a lot"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]

This archive was generated by hypermail 2.1.6 : Sun Dec 31 2017 - 23:20:43 CST