From: Norman Geist (norman.geist_at_uni-greifswald.de)
Date: Fri Jun 10 2011 - 02:56:30 CDT
Hi Danny,
It's not very advisable to use that much cores with one gpu. The speedup
already falls from 1:1 to 2:1 CPU:GPU ratio. The speedup you then see is
mostly to the unaccelerated part of NAMD as it is only the non-bonded
calculations which are made on the gpu. I have a machines with 6-core and
with 2 TeslaC2050, that's far enough cores. I also think, that 48 cpus alone
will do much better itself, and you don't need the gpu here. NAMD scaling
within a machine is pretty great. Maybe you can think about plugging in more
gpus if you can, but such a machine needs no gpus I think. What kind of
machine is that??
Best regards
Norman Geist.
Von: owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] Im Auftrag
von Danny Xu
Gesendet: Freitag, 10. Juni 2011 08:18
An: namd-l
Betreff: namd-l: max number of CPU PEs per GPU?
Hi,
I was testing the namd 2.8b2 cuda binary (
<http://www.ks.uiuc.edu/Development/Download/download.cgi?UserID=&AccessCode
=&ArchiveID=1127> Linux-x86_64-CUDA) on a 48-core opteron/1x GTX480 server.
I got:
...
Pe 41 sharing CUDA device 0 first 0 next 42
Pe 40 sharing CUDA device 0 first 0 next 41
FATAL ERROR: CUDA error cudaStreamCreate on Pe 23 (chili device 0): no
CUDA-capable device is available
------------- Processor 23 Exiting: Called CmiAbort ------------
Reason: FATAL ERROR: CUDA error cudaStreamCreate on Pe 23 (chili device 0):
no CUDA-capable device is available
First I thought there was something wrong with the GTX480 card. Then I tried
12 PE's and the job ran just fine. It turned out the max number core I could
use with one GTX480 is 14. My question is, is this hard coded in namd 2.8?
Or it's a bug in 2.8b2 and has been fixed in 2.8?
It'd be nice if all 48-core could be used with the GPU.
Thanks,
DX
This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:20:24 CST