Turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

- Intel Community
- Software
- Software Development SDKs and Libraries
- Intel® Integrated Performance Primitives
- Max size for arrays/matrices

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Mute
- Printer Friendly Page

apolo74

Beginner

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

07-19-2011
01:06 AM

42 Views

Max size for arrays/matrices

independently of the amount of RAM, is there any other constraint for the maximum size of IPP arrays or matrices???

I've read that IPP Small Matrix should be used for matrices in the range of 3x3 to 20x20 and IPP for larger number of elements, but is there an upper limit for IPP elements without considering the amount of RAM? Another related question would be about the advantages of using MKL compared to IPP libraries or vice versa... and to complicate a bit more my confusion, I think there is a new player on the field called AVX which could/should be used for vector/matrix operations???

I'm trying to build an application that uses convolutions on 1D vectors or 2D matrices and also 3D matrices with elements ranging from 20 to 500 in each dimension. So in the extreme case I would need a pointer to 500x500x500 elements, is this possible?

Boris

Link Copied

1 Reply

Naveen_G_Intel

Employee

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

07-28-2011
11:45 PM

42 Views

Hi,

IPP provide small matrix operation functionsin ippm.pdf,Unlike MKL, whichis usually forengineeringand scientificcomputation, ippm functionsare optimizated for small matrix. So the **preferred size generallyranges from 3x3 to 20x20.**

We are optimized many IPP functions using AVX instructions. Refer** to list to get details.**

My recommendation is to refer to an article on intel site **Benefits of Intel AVX For Small Matrices**

Regards,

Naveen Gv

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

For more complete information about compiler optimizations, see our Optimization Notice.