Intel® Fortran Compiler
Build applications that can scale for the future with optimized code designed for Intel® Xeon® and compatible processors.
29251 Discussions

Execution speed on different servers

cr362
Beginner
360 Views

Dear All,

I'm compiling identical fortran code on two similar servers, but when I run the code, the runtime takes much longer on one server (3 min), compared to the other (0.5 min). Is there any reason why this might occur?

 

When I compile my code on one server I get the following messages:

f_skyview.for(33) : (col. 9) remark: LOOP WAS AUTO-PARALLELIZED.

 

But on the other sever I get all of these messages:

docells.inc(23): (col. 10) remark: LOOP WAS VECTORIZED.
setup.inc(15): (col. 2) remark: LOOP WAS VECTORIZED.
setup.inc(49): (col. 9) remark: LOOP WAS VECTORIZED.
setup.inc(66): (col. 14) remark: PARTIAL LOOP WAS VECTORIZED.
resetsnow.inc(19): (col. 3) remark: LOOP WAS VECTORIZED.
resetsnow.inc(5): (col. 12) remark: PARTIAL LOOP WAS VECTORIZED.
resetsnow.inc(5): (col. 12) remark: PARTIAL LOOP WAS VECTORIZED.
resetsnow.inc(5): (col. 12) remark: PARTIAL LOOP WAS VECTORIZED.
main.for(521): (col. 4) remark: LOOP WAS VECTORIZED.
main.for(522): (col. 4) remark: LOOP WAS VECTORIZED.
main.for(523): (col. 4) remark: LOOP WAS VECTORIZED.
main.for(524): (col. 4) remark: LOOP WAS VECTORIZED.
main.for(598): (col. 9) remark: LOOP WAS VECTORIZED.
main.for(602): (col. 9) remark: LOOP WAS VECTORIZED.
f_skyview.for(25): (col. 12) remark: LOOP WAS VECTORIZED.
subsurface.for(2138): (col. 2) remark: LOOP WAS VECTORIZED.
subsurface.for(2225): (col. 6) remark: LOOP WAS VECTORIZED.
subsurface.for(2234): (col. 5) remark: LOOP WAS VECTORIZED.
subsurface.for(2273): (col. 5) remark: LOOP WAS VECTORIZED.
subsurface.for(756): (col. 4) remark: LOOP WAS VECTORIZED.
subsurface.for(757): (col. 4) remark: LOOP WAS VECTORIZED.
subsurface.for(783): (col. 5) remark: LOOP WAS VECTORIZED.
subsurface.for(819): (col. 5) remark: PARTIAL LOOP WAS VECTORIZED.
subsurface.for(819): (col. 5) remark: PARTIAL LOOP WAS VECTORIZED.

 

Does anyone know why this is? The options I'm using are:

ifort -extend-source -fp-model precise -i_dynamic -mcmodel=large -O3 -parallel

 

Many thanks!

0 Kudos
1 Reply
TimP
Honored Contributor III
360 Views
Quoting - cr362

Dear All,

I'm compiling identical fortran code on two similar servers, but when I run the code, the runtime takes much longer on one server (3 min), compared to the other (0.5 min). Is there any reason why this might occur?

Something you're not telling us, such as using different compilers on the two systems, without consistent options (e.g. one 32-bit, with vectorization defaulting off, and the other 64-bit).

0 Kudos
Reply