Intel® Integrated Performance Primitives
Deliberate problems developing high-performance vision, signal, security, and storage applications.

ippsSortAscend_32f_I goes into infinite loop

thorsan
Beginner
811 Views

Hi,

In my signal processing routine i sort a vector of 32f numbers using ippsSortAscend_32f_I(). But in some cases the vector is all zeros(or all equal numbers) and then the function call never returns. Ofcourse I could check for the vector to be non-zero, but I just feel like it should be reported. It is also the case if many numbers are the same, the routine is very slow.

Thanks,

Thor Andreas

0 Kudos
5 Replies
Vladimir_Dudnik
Employee
811 Views
Thanks for your report. Could you please provide a little bit more info on this? What is your IPP version, what is your test platform (processor, OS)? What is length of vector which cause the issue?
Regards,
Vladimir

0 Kudos
thorsan
Beginner
811 Views

Hi,

My IPP version is 5.3 ia32 and my CPU is U2500 @ 1.20 GHz running Windows XP. I have experienced this on another processor as well, but I dont remember the type of CPU.

The length of my vectors are in the range 1800-2300 samples.

Do you need any more information?

Thanks,

Thor Andreas

0 Kudos
Vladimir_Dudnik
Employee
811 Views
Thanks, yes, additional questions are
do you link IPP DLLs or static libraries (threaded/non threaded)?
Are you vectors aligned on data type size (4 bytes for Ipp32f)?
Vladimir

0 Kudos
thorsan
Beginner
811 Views

Hi,

Im linking using DLL's (ipps.lib) and I use ippsMalloc_32f allocate memory.

Thanks,

Thor Andreas

0 Kudos
Vladimir_Dudnik
Employee
811 Views
Thanks, we will try to reproduce the issue
Vladimir

0 Kudos
Reply