LabWindows/CVI

cancel
Showing results for 
Search instead for 
Did you mean: 

CVI help (CVI 2009) on external compilers outdated?

Solved!
Go to solution

Hello NI world,

 

the CVI help on 'creating optimized code' says that

 

Note  You cannot compile 64-bit applications using external optimizing compilers.

 

 

Contrary to this there is a template configuration file for Intel C++ 11.1 64 bit...

 

So am I correct in assuming that the help is outdated? 

 

By the way, I am happy to find that ICL 11.1 is now supported by a template:-)

 

Wolfgang

0 Kudos
Message 1 of 5
(4,428 Views)
Solution
Accepted by topic author Wolfgang

Hey Wolfgang - 

 

You are correct.  I've filed a bug report (199382) to make sure this documentation is updated.

 

NickB

National Instruments 

0 Kudos
Message 2 of 5
(4,414 Views)

So then,

 

If I buy CVI 2009 and install it on WinXP Pro, and use the Intel 11.1 compiler to build a 32 bit application,  will I be able to use those Intel Compiler optimizations that require 16 byte alignment (i.e. vectorizing optimizations) ?

 

This is more or less a question about the 32 bit linker in CVI 2009 - is it the same as the one in CVI 9.0 or did it get redone?

 

Thanks.

 

Menchar

0 Kudos
Message 3 of 5
(4,400 Views)

Hey Menchar -

 

We were unable to make any byte alignment changes for CVI 2009.  We have made plans to look at changing our byte alignment for a future version of CVI. 

 

NickB

National Instruments 

0 Kudos
Message 4 of 5
(4,367 Views)

That's a real bummer, but at least NI seems to understand that they should try to solve the problem.

 

I'd recommend that no one buy the Intel compiler (icl) for use with CVI (now that I did buy it and know what I know now).  You have to have Visual Studio installed to use icl anyway, and as it turns out if you're not using vectorizing optimizations the MSVCPP compiler is actually faster than the icl by my experience.  I benchmarked Intel 11.1 icl Vs. MSVCPP 8.0 and MSVCPP clearly outperforms the icl, at least for my application, and I've never seen a byte alignment problem.

 

This wasn't true with older versions of the MSVCPP compiler - the icl 9.x easily outperformed  MSVCPP 6.0 at any optimization level from what I saw.

 

MSVCPP 10.0 due out next spring looks like it has auto-parallelization built into it (icl does this now) so it might become even more attractive.

 

 

0 Kudos
Message 5 of 5
(4,359 Views)