Intel® Integrated Performance Primitives
Community support and discussions relating to developing high-performance vision, signal, security, and storage applications.
Announcements
This community is designed for sharing of public information. Please do not share Intel or third-party confidential information here.

ippiConvValid_32f -9 mem error

Ryan2
Beginner
150 Views
Hi,

Assuming all images are sqaure for this example and all kernel sizes are odd.
I = input image size before padding
k = kernel size (always odd)
We pad I by k.
Src1 forippiConvValid is of size I + k - 1
Scr2 forippiConvValid is of size k
According to the documentation,
dest of ippiConvValid_32f is of size ( I + k -1 ) - k + 1 = I
Everything work perfect!!...until k > I
When k > I,ippiConvValid_32f outputs black.
Can someone please lookinto this for me. I need a solution to this problem as k is often greater than I.
Is there a workaround?
Thanks,
Ryan
0 Kudos
4 Replies
Gennady_F_Intel
Moderator
150 Views
but what is the return Value of ippiConvValid when k > i?

Ryan2
Beginner
150 Views
We installed IPP 7.0.7 The error is now solved!!!
We're able to convolve all the way up until we get a -9 memory error:
ippStsMemAllocErr
/* Memory allocated for the operation is not enough. */
Thanks!
Ryan2
Beginner
150 Views
Hi,
I'm testing on a Win7 64bit. Intel Xeon 24 thread machine with 32GB of RAM.
I'm getting: ippStsMemAllocErr = -9, /* Memory allocated for the operation is not enough. */ withippiConvValid_32f_C1R
When I increase the size of the kernel forippiConvValid_32f_C1R incrementally until I get the error -9. On input sizes just prior to the -9 error, my system is just barely using 6GB.
I am often doing large convoultion and need to be able to utilize my 32GB of system memory. Help please?
Best,
Ryan
Ryan2
Beginner
150 Views
Anyone?
Reply