Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.programming.threads > #2527 > unrolled thread

About scimark2

Back to article view | Back to comp.programming.threads

Contents

  About scimark2 Ramine <ramine@1.1> - 2014-06-28 15:30 -0700
    Re: About scimark2 Ramine <ramine@1.1> - 2014-06-28 15:34 -0700

#2527 — About scimark2

Hello,


About the scimark2 benchmark that i did on FreePascal and Delphi and GCC,
i have used the following command to compiler:

gcc -O1 -S


i have used the -S flag to generate the assembler source code,
and i have looked at the assembler source code and i have noticed that 
effectivly GCC do auto-vectorize the code, it uses SSE2 instructions to 
vectorize the floating point calculations even with only the -O1 
optimization, this is why GCC is 2X times faster than FreePascal and 
Delphi on some scimark2 benchmarks that use floating point calculations.


Thank you,
Amine Moulay Ramdane.

[toc] | [next] | [standalone]

#2528

Hello,

Auto-vectorization has not been implemented yet on FreePascal
and Delphi, but it has been implemented on for example GCC and Intel 
compilers...



Thank you,
Amine Moulay Ramdane.


On 6/28/2014 3:30 PM, Ramine wrote:
>
> Hello,
>
>
> About the scimark2 benchmark that i did on FreePascal and Delphi and GCC,
> i have used the following command to compiler:
>
> gcc -O1 -S
>
>
> i have used the -S flag to generate the assembler source code,
> and i have looked at the assembler source code and i have noticed that
> effectivly GCC do auto-vectorize the code, it uses SSE2 instructions to
> vectorize the floating point calculations even with only the -O1
> optimization, this is why GCC is 2X times faster than FreePascal and
> Delphi on some scimark2 benchmarks that use floating point calculations.
>
>
> Thank you,
> Amine Moulay Ramdane.

[toc] | [prev] | [standalone]

Back to top | Article view | comp.programming.threads

csiph-web