[LLVMdev] Polyhedron 2005 regressions

[Chris Lattner]

On Sep 20, 2010, at 7:40 AM, Jack Howarth wrote:

>  Comparing the Polyhedron 2005 benchmark results for gfortran from
> llvm-gcc-4.2 of April 7th, 2010 and September 18th, 2010 (from
> the rc2 2.8 release branch), we seem to be regressing in performance
> for this release....

It looks like some wins, but bigger losses.  Would you be willing to reduce capacita or ac to a smallish example that shows the performance regression?

-Chris

> 
> ================================================================================
> Date & Time     :  7 Apr 2010 22:24:16
> Test Name       : llvm_gfortran_lin_p4
> Compile Command : llvm-gfortran -ffast-math -funroll-loops -msse3 -O3 %n.f90 -o %n
> Benchmarks      : ac aermod air capacita channel doduc fatigue gas_dyn induct linpk mdbx nf protein rnflow test_fpu tfft
> Maximum Times   :     2000.0
> Target Error %  :      0.100
> Minimum Repeats :    10
> Maximum Repeats :   100
> 
>   Benchmark   Compile  Executable   Ave Run  Number   Estim
>        Name    (secs)     (bytes)    (secs) Repeats   Err %
>   ---------   -------  ----------   ------- -------  ------
>          ac      0.34       10000     12.85      10  0.0067
>      aermod     20.01       10000     15.42      10  0.0076
>         air      1.52       10000      7.50      12  0.0934
>    capacita      0.58       10000     45.67      10  0.0152
>     channel      0.41       10000      2.19      10  0.0365
>       doduc      2.29       10000     29.08      10  0.0140
>     fatigue      1.02       10000      8.83      10  0.0582
>     gas_dyn      0.77       10000     13.12      21  0.0909
>      induct      2.54       10000     24.22      10  0.0096
>       linpk      0.28       10000     15.67      10  0.0343
>        mdbx      0.79       10000     12.09      10  0.0070
>          nf      0.33       10000     30.10      12  0.0751
>     protein      1.10       10000     42.02      10  0.0100
>      rnflow      1.53       10000     32.00      10  0.0214
>    test_fpu      1.28       10000     11.73      10  0.0446
>        tfft      0.24       10000      2.16      13  0.0988
> 
> Geometric Mean Execution Time =      14.07 seconds
> 
> ================================================================================
> 
> ================================================================================
> Date & Time     : 18 Sep 2010 21:28:47
> Test Name       : llvm_gfortran_lin_p4
> Compile Command : llvm-gfortran -ffast-math -funroll-loops -msse3 -O3 %n.f90 -o %n
> Benchmarks      : ac aermod air capacita channel doduc fatigue gas_dyn induct linpk mdbx nf protein rnflow test_fpu tfft
> Maximum Times   :     2000.0
> Target Error %  :      0.100
> Minimum Repeats :    10
> Maximum Repeats :   100
> 
>   Benchmark   Compile  Executable   Ave Run  Number   Estim
>        Name    (secs)     (bytes)    (secs) Repeats   Err %
>   ---------   -------  ----------   ------- -------  ------
>          ac      0.29       10000     14.20      10  0.0147
>      aermod     18.69       10000     15.23      10  0.0412
>         air      1.13       10000      7.51      16  0.0570
>    capacita      0.51       10000     48.51      10  0.0506
>     channel      0.30       10000      2.20      10  0.0697
>       doduc      1.97       10000     30.31      10  0.0084
>     fatigue      0.81       10000      8.55      10  0.0098
>     gas_dyn      0.67       10000     13.00      15  0.0740
>      induct      2.04       10000     24.40      10  0.0089
>       linpk      0.25       10000     15.68      10  0.0734
>        mdbx      0.77       10000     12.02      10  0.0120
>          nf      0.33       10000     31.94      10  0.0909
>     protein      1.00       10000     40.53      10  0.0118
>      rnflow      1.42       10000     32.31      12  0.0642
>    test_fpu      1.26       10000     11.31      13  0.0964
>        tfft      0.22       10000      2.15      13  0.0167
> 
> Geometric Mean Execution Time =      14.20 seconds
> 
> ================================================================================
> 
>                  % delta for 7 Apr 2010 vs 18 Sep 2010
> ac                    -10.5
> aermod                 +1.2
> air                    -0.1
> capacita               -6.2
> channel                -0.5
> doduc                  -1.4
> fatigue                +3.2
> gas_dyn                +0.9
> induct                 -0.7
> linpk                  -0.1
> mdbx                   +0.6
> nf                     -6.1
> protein                +3.5
> rnflow                 -1.0
> test_fpu               +3.6
> tfft                   +0.5
> 
> mean                   -0.8
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev