ippsAddProductC_32fc and ippsAddProductC_64fc

[Mike45]

Will ippsAddProductC_32fc and ippsAddProductC_64fc be added to the IPP library anytime soon?  I am currently using Eigen for this operation because the only way I have found to do it in IPP is with two operations which is much slower with large datasets.

[Ruqiu_C_Intel]

Thank you for raising the new APIs request for the complex dataset.

We will consider the request carefully and put them into our backlog. For better evaluating the request, can you help introduce your projection briefly, including your target platforms, common dataset sizes, OS, target date, etc.

[Mike45]

Thank you, I have two target processors (13th Gen Intel(R) Core(TM) i7-13800HRE and  Intel(R) Core(TM) i7-9850HE both running Ubuntu 22 (Jammy) compiling with gcc 11.4.  The application will create a single Ipp32fc* as the output then iterate across multiple Ipp32fc* applying a separate constant(Ipp32fc) to each.  The product of each sample in the input Ipp32fc* and the constant will be added to the output Ipp32fc* at the same sample.  Each array will have 2 million to 6 million samples.

I do have a workaround using Eigen today so this isn't a high priority but I am very curious to see if IPP can beat out Eigen or not for this function.  Eigen only appears to be using SSE2 instructions.  I am hoping there are additional SIMD instructions IPP can make use of.

[Ruqiu_C_Intel]

Thank you for the reply.

We are considering the API expanding to support complex data type.

Regards,

Ruqiu

[Mike45]

Do you have any estimate for when this may be included?  Thanks

[Aleksandra_K]

Hi, it's planned for the next release.

Regards,

Alex

[Aleksandra_K]

Hi,

The new APIs are included the in the new release, see Intel® Integrated Performance Primitives Release Notes for Intel®...

Regards,

Alex