[MITgcm-devel] [MITgcm-cvs] MITgcm/doc CVS Commit

[Jean-Michel Campin]

Hi Martin,

Thanks for doing these tests.
My take out from this is:
1) this reinforce the choice to have W2_USE_E2_SAFEMODE defined as the
 default (whereas with the old version, the slow down was significant, 
 and this default was more problematic).
2) it's still faster with W2_USE_E2_SAFEMODE #undef, so I guess it's worth
to keep this option (we could have decided to remove this option and to
keep the part of the code corresponding to #define W2_USE_E2_SAFEMODE).

Thanks,
Jean-Michel

On Fri, Oct 19, 2012 at 04:19:04PM +0200, Martin Losch wrote:
> I can now add this:
> 
> I ran the CS150 with 6 cpu (on per face) with exch2_put_rx2.template version 1.4 (SAFEMODE defined) and 1.3 (SAFEMODE undefined) with a simple ftrace analysis and am getting this:
> FREQUENCY  EXCLUSIVE       AVER.TIME    MOPS  MFLOPS V.OP  AVER.   VECTOR I-CACHE O-CACHE    BANK  PROG.UNIT           TIME[sec](  % )    [msec]                 RATIO V.LEN    TIME   MISS    MISS      CONF
> 
>  1311552   271.006(  2.7)     0.207  6305.2   166.1 95.55  99.5  268.974  1.1314  0.1703  0.0397  exch2_put_rl2 (r1.4)
>  1311552    60.982(  0.6)     0.046  2552.4   738.3 57.95  15.5   59.130  1.0930  0.0497  0.0649  exch2_put_rl2 (r1.4 W2_USE_E2_SAFEMODE undefined)
>   1311552  6055.632( 36.8)     4.617   133.6     7.6  0.07 204.0    4.536  3.2848 4629.3812 1190.0043  exch2_put_rl2 (r1.3 W2_USE_E2_SAFEMODE defined)
>  1311552    54.211(  0.6)     0.041  2867.8   830.5 57.91  15.5   52.802  0.2240  0.4724  0.0663  exch2_put_rl2 (r1.3 W2_USE_E2_SAFEMODE undefined)
> 
> 
> The total timing is (includes more effects than just that of exch2_put_rl2.F I assume)
> r1.4 W2_USE_E2_SAFEMODE undefined: 1675.324
> r1.4 W2_USE_E2_SAFEMODE   defined: 1735.251
> r1.3 W2_USE_E2_SAFEMODE undefined: 1657.051
> r1.3 W2_USE_E2_SAFEMODE   defined: 2789.741
> 
> So there's a clear improvement with r1.4 when W2_USE_E2_SAFEMODE is defined
> 
> M.
> 
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel