Alex Granovsky
gran@classic.chem.msu.su
The sample input file on CUDing page was carefully tuned for
the specific computer system for which it was intended.
So, generally, it cannot be used as is without modification.
Indeed, there is a warning message in the output:
DGEMM will use 4 threads. CUDA aware parts of program will use 1 devices. Creating thread pool to serve up to 64 threads. Warning: number of DGEMM threads exceeds number of allotted CPU cores.
The more general input for dual-core system with no Hyper-Threading and single CUDA device could be as follows as Gena suggested:
$contrl scftyp=rhf mplevl=4 runtyp=energy icharg=-1 $end $system mwords=200 mklnp=2 np=1 freshf=0 memf=1 flush=0 async=0 $end $smp cuda=.t. dianp=1 $end $cuda cuflgs=1 events=1 $end $basis gbasis=n311 ngauss=6 npfunc=2 ndfunc=2 diffsp=.t. $end $scf dirscf=1 $end $mp4 sdtq=1 trpmet=1 mtio=1 $end $data C1 CL 17.0 -0.3520333657 -0.1650980028 -0.0471638329 H 1.0 0.7972862956 -1.4281273262 -1.2294694043 H 1.0 -1.5889984100 1.0741294712 -1.1576600606 H 1.0 -1.5392863051 -1.2871081468 1.2265482694 H 1.0 0.7428387888 0.9385578775 1.0969088438 F 9.0 1.3109647391 -2.0051605917 -1.7641455975 F 9.0 -2.1531089472 1.6362757807 -1.6569055973 F 9.0 -2.0787314212 -1.7943015141 1.8058474018 F 9.0 1.2619562758 1.4885572294 1.6876526453 H 1.0 2.8342698106 1.9875432625 1.3734980540 F 9.0 3.7058425393 2.3127319602 1.2848892783 $end
You may find the following threads on the forum useful:
http://classic.chem.msu.su/cgi-bin/ceilidh.exe/gran/gamess/forum/?C38c6e6626R9R-7726-474-00.htm
http://classic.chem.msu.su/cgi-bin/ceilidh.exe/gran/gamess/forum/?C347e378b7OrJ-7644-1248-00.htm
as well as some on disabling TDR and core parking features.
Kind regards,
Alex Ganovsky
On Wed Jul 20 '11 5:04pm, Alain Igban wrote
-------------------------------------------
>Thank you!
>Yes my pc crashed and i restarted it. I dont know about bugs but this happened twice (Black screen, unresponsive pc).
>Anyway thanks for pointing the difference between HF and the MP4 calculation. Now I know :]. I am an undergrad chem student and almost everything here is way over my head.
>Well I really wanted to use Firefly for my next project with CUDA enabled, so I did some test runs which you have seen in the first message.
>Thank you for pointing that (MKLNP) out
>