Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly

Jeff Johns
jjeff@kku.ac.th

Running the standard MP4 test I got:
$system mwords=140 mklnp=4 np=3 freshf=0 memf=1 flush=0 async=0 $end
$smp csmtx=1 cuda=.t. tpool=64 $end                                            
$cuda cuflgs=1 events=1 nocpu=1 $end                                          
$cuda cumask=1 cuaff=5  $end                                                  

1 GTX460 GPU’s (not overclocked)
Intel Core2/ Win32  Firefly version running under Windows NT (Jeff’s)
Running on Intel CPU:  Brand ID  0, Family  6, Model  23, Stepping 10
CPU Brand String    :  Intel(R) Core(TM)2 Quad CPU    Q8400  @ 2.66GHz
CPU Features        :  CMOV, MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, MWAIT, EM64T
Data cache size     :  L1 32 KB, L2 2048 KB, L3     0 KB
CPU frequency is    :  3281.570 MHz
max    # of   cores/package :   4
max    # of threads/package :   4
max     cache sharing level :   2
 0    GeForce GTX 460      Yes      Yes      Yes       No      Yes
......END OF NBO ANALYSIS......
CPU        TIME:   STEP =      0.00 ,  TOTAL =     2386.6 SECONDS (   39.8 MIN)
WALL CLOCK TIME:   STEP =      0.04 ,  TOTAL =     2169.1 SECONDS (   36.2 MIN)
CPU UTILIZATION:   STEP =      0.00%,  TOTAL =     110.03%
                        I/O STATISTICS:
DATA READ TOTAL =     1702.750 MB,  DATA WRITTEN TOTAL =      914.661 MB
  125174229 WORDS OF    DYNAMIC MEMORY USED
EXECUTION OF FIREFLY TERMINATED NORMALLY  9:54:00 30-DEC-2010    
----------------------------------------------------------------

CPU only 4 cores
Intel Core2/ Win32  Firefly version running under Windows NT (Jeff’s)
Running on Intel CPU:  Brand ID  0, Family  6, Model  23, Stepping 10
CPU Brand String    :  Intel(R) Core(TM)2 Quad CPU    Q8400  @ 2.66GHz
CPU Features        :  CMOV, MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, MWAIT, EM64T
Data cache size     :  L1 32 KB, L2 2048 KB, L3     0 KB
CPU frequency is    :  3281.570 MHz
max    # of   cores/package :   4
max    # of threads/package :   4
max     cache sharing level :   2
......END OF NBO ANALYSIS......
CPU        TIME:   STEP =      0.00 ,  TOTAL =     4680.5 SECONDS (   78.0 MIN)
WALL CLOCK TIME:   STEP =      0.00 ,  TOTAL =     4697.1 SECONDS (   78.3 MIN)
CPU UTILIZATION:   STEP =      0.00%,  TOTAL =      99.65%
                        I/O STATISTICS:
DATA READ TOTAL =    91005.658 MB,  DATA WRITTEN TOTAL =      999.917 MB
   18950703 WORDS OF    DYNAMIC MEMORY USED
EXECUTION OF FIREFLY TERMINATED NORMALLY 11:42:14 30-DEC-2010    

-----------------------------------------------------------------------

Previous data reported on this forum:

2 GTX280 GPU’s
Core i7    / Win32  Intel(R) Core(TM) i7 CPU         940  @ 2.93GHz
CPU        TIME:   STEP =      0.00 ,  TOTAL =     1818.2 SECONDS (   30.3 MIN)
WALL CLOCK TIME:   STEP =      0.00 ,  TOTAL =     1526.3 SECONDS (   25.4 MIN)

As above, no GPU
CPU        TIME:   STEP =      0.00 ,  TOTAL =     4284.7 SECONDS (   71.4 MIN)
WALL CLOCK TIME:   STEP =      0.00 ,  TOTAL =     4291.8 SECONDS (   71.5 MIN)

The GTX 460 ran straight out of the box (less than 10 minutes to install incl software), no
installation hassles, runs cool (45-50 C), and is readily available. The GTX460 provides about a factor of two performance over Quad CPU Q8400 @3.38GHz, and about the same as two GTX280's. You can overclock the 460 about 25-30% as well if desired. The GTX460 showed much better performance over CPU only for OpenMM/Gromacs (5 to 100 times, increasing with increasing no. of atoms) and about 50% of the performance of the GTX480/same as a Tesla1070 running HOOMD-Blue (as expected). It might be a good alternative for those not wanting to invest in the cost of a Tesla.

[ This message was edited on Wed Jan 5 '11 at 7:52pm by the author ]