The latest IPP code fails to properly load the following JPEG when using a DCT scaling like JD_1_4:
It seems like it's a problem with 411 JPEGs, as the same JPEG with 211 encoding doesn't show the problem.
I think something is wrong with the fourth h-sample when calling ppiDCTQuantInv8x8To2x2LS_JPEG_16s8u_C1R inside ReconstructMCURowBL8x8To2x2, but I'm not certain what.
Thanks for any help.
Is Intel IPP still supported by Intel? Is there some other avenue we should pursue for bugs?
This thread also has an open IPP bug we need help with (we got great help from Vladimir Dudnikin that thread to begin with, but that was some time ago):
Sure, Intel IPP is supported by our development and technical support teams. You can submit your issue reports directly to Intel Premier Support where they will fall to our data base and will be tracked on regular base.
I'm sorry for the delay, our JPEG expert was on vacation. Now we are investigating your issue report and I'll update your with status of this soon. Thank you for your interest to Intel Integrated Performance Primitives product.