NAMD with CUDA

From: Prakash Velayutham (Prakash.Velayutham_at_cchmc.org)
Date: Thu May 07 2009 - 13:06:04 CDT

• Next message: David Hardy: "Re: water models and some questions"
• Previous message: Peter Freddolino: "Re: water models and some questions"
• Next in thread: Prakash Velayutham: "Re: NAMD with CUDA"
• Reply: Prakash Velayutham: "Re: NAMD with CUDA"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]

Hi,

I was wondering if CHARM needs to be built with CUDA option for NAMD
(with --with-cuda option) to make use of GPUs.

I compiled NAMD 2.7b1 like following:

Built Charm 6.1.2 as below:

/build charm++ multicore-linux64

Built NAMD as below:

/config Linux-x86_64 --with-cuda --charm-arch multicore-linux64
cd Linux-x86_64
make

But the test apoa1 code runs slower using GPU than using the CPU
itself. Just wanted to know what the reason could be.

Thanks,
Prakash

• Next message: David Hardy: "Re: water models and some questions"
• Previous message: Peter Freddolino: "Re: water models and some questions"
• Next in thread: Prakash Velayutham: "Re: NAMD with CUDA"
• Reply: Prakash Velayutham: "Re: NAMD with CUDA"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:52:45 CST