[Simh] What is the best machine to run simh?

Thomas Pfau pfau at nbpfaus.net
Fri Jan 25 06:39:34 EST 2008


Marc Chametzky wrote:
>> simh with inline changes and compiled with -O3 -march=athlon64
>> -mfpmath=387
>
> If you have a few moments and can do an experiment, could you try
> another run of PT_VAX with the vax binary built almost the same, but
> with -O2 instead?
>
> I was surprised to see -O3 give worse performance on my machine than
> -O2. I want to find whether it's some anomaly on my system, or if
> there is really something to it.
>
> --Marc
-O2

Simh04» r pt_vax
Emulated VAX CPU comparative performance test, version 1.0
Copyright 2006 Software Resources International.
This test is not suitable to measure overall system performance.
No rights can be derived from the execution or the output of this test
Adjusting test length for fixed and floating point to 10 seconds/unit
Test started on Fri Jan 25 06:24:45 2008, for a VAXserver 3900 Series
CPU with V
MS V7.3   

Sequential test results:                                             

               MIPs       Dhrystones   VUPs
Run 0          14.3        30769       16.4       
Run 1          12.5        27972       16.0       
Run 2          14.3        34188       14.8       
Run 3          12.5        38461       16.0       
---------------------------------------------
Average        13.4        32847       15.8       

Interleaved test results:

               MIPs       Dhrystones   VUPs
Run 0          12.5        30769       16.4       
Run 1          14.3        38461       15.8       
Run 2          14.3        30769       16.2       
Run 3          11.1        34188       15.8       
---------------------------------------------
Average        13.0        33546       16.1       

Test ended on Fri Jan 25 06:28:47 2008, checksum A2FD2277

-O3

Simh04» r pt_vax
Emulated VAX CPU comparative performance test, version 1.0
Copyright 2006 Software Resources International.
This test is not suitable to measure overall system performance.
No rights can be derived from the execution or the output of this test
Adjusting test length for fixed and floating point to 10 seconds/unit
Test started on Fri Jan 25 06:35:15 2008, for a VAXserver 3900 Series
CPU with V
MS V7.3   

Sequential test results:                                             

               MIPs       Dhrystones   VUPs
Run 0          14.3        34722       16.0       
Run 1          14.3        31250       16.0       
Run 2          14.3        31250       16.0       
Run 3          14.3        31250       16.0       
---------------------------------------------
Average        14.3        32118       16.0       

Interleaved test results:

               MIPs       Dhrystones   VUPs
Run 0          14.3        31250       16.0       
Run 1          14.3        31250       16.0       
Run 2          14.3        39062       16.0       
Run 3          14.3        31250       15.8       
---------------------------------------------
Average        14.3        33203       16.0       

Test ended on Fri Jan 25 06:39:07 2008, checksum 3D69FE85




More information about the Simh mailing list