Hany

Hi, I am using Block LDPC codes to encode and decode WiGig packets. To do that, I generate parity-check matrix from a base- matrix. Then, I pass the generated parity-check matrix to Block LDPC generator BLDPC_Generator which in turn is passed to the LDPC_Code object. Based on the online documentation, the -1 in the base-check matrix presents submatrices with all zero entries. Here is a snippet of my code: imat codeMatrix = "40 -1 38 -1 13 -1 5 -1 18 -1 -1 -1 -1 -1 -1 -1;" "34 -1 35 -1 27 -1 -1 30...

So if I use the BLDPC _Generator class directly to generate the enocded codeword and just use the LDPC_Code to decode the received codeword i.e. without passing the generator, the code works. However, it seems there are always errors in the decoded codeword even if I use a noiseless channel. So probably that is why perform_integrity_check fails? Here is my code: #include <itpp/itcomm.h> using namespace std; using namespace itpp; int main (int argc, char *argv[]) { imat codeMatrix = "40 -1 38 -1 13...

Hi, I am using Block LDPC codes to encode and decode WiGig packets. To do that, I generate Block LDPC parity matrix from a base matrix. Then, I pass the Block LDPC parity matrix to Block LDPC generator which in turn is passed to the LDPC_Code object. Based on the desciption, -1 will present submatrices with all zero entries. Here is my code: imat codeMatrix = "40 -1 38 -1 13 -1 5 -1 18 -1 -1 -1 -1 -1 -1 -1;" "34 -1 35 -1 27 -1 -1 30 2 1 -1 -1 -1 -1 -1 -1;" "-1 36 -1 31 -1 7 -1 34 -1 10 41 -1 -1 -1...

Hi, I am using Block LDPC codes to encode and decode WiGig packets. To do that, I generate Block LDPC parity matrix from a base matrix. Then, I pass the Block LDPC parity matrix to Block LDPC generator which in turn is passed to the LDPC_Code object. Based on the desciption, -1 will present submatrices with all zero entries. Here is my code: ivec codeMatrix = "40 -1 38 -1 13 -1 5 -1 18 -1 -1 -1 -1 -1 -1 -1;" "34 -1 35 -1 27 -1 -1 30 2 1 -1 -1 -1 -1 -1 -1;" "-1 36 -1 31 -1 7 -1 34 -1 10 41 -1 -1 -1...

Hi, I am using Block LDPC codes to encode and decode WiGig packets. To do that, I generate Block LDPC parity matrix from a base matrix. Then, I pass the Block LDPC parity matrix to Block LDPC generator which in turn is passed to the LDPC_Code object. Based on the desciption, -1 will present submatrices with all zero entries. Here is my code: ivec codeMatrix = "40 -1 38 -1 13 -1 5 -1 18 -1 -1 -1 -1 -1 -1 -1;" "34 -1 35 -1 27 -1 -1 30 2 1 -1 -1 -1 -1 -1 -1;" "-1 36 -1 31 -1 7 -1 34 -1 10 41 -1 -1 -1...

Thanks Bodgan, I've posted my question on Intel MKL Forum. I wanted to see if anyone had expereinced the same problem with the latest MKL. What you mentioned regaeding IT++ not optimally using MKL API makes sense as ITPP is not being updated and Intel MKL has been quite modified since the last stable release of IT++. One of the issues that I ran into during linking with MKL is that one of the domains is outdated by MKL. So instead of using MKL_FFT in ITPP, you have to use MKL_DOMAIN_FFT.

Hi, I've written a PHY stack that encodes and decodes 802.11 frames using the ITPP library. Now I am doing some performance evaluation for my implementation and trying different optimization options to speed up calculations and execution time. Using standard BLAS and LAPACK libraries in Linux with FFTW 3.8, my program can generate and decode 1K packets per 180 seconds. However, when I switch to the latest version of Intel MKL, the execution time increases to 210 seconds which is not expected. These...