How to apply interleaved batch permutation?


I am trying to solve a batch of linear systems using QR factorization.

The steps I follow are: 1) Assemble matrix and right-hand sides, 2) interleave with dge_interleave, 3) A P = QR with dgeqrff_interleave_batch, 4) B := Q^T B with dormqr_interleave_batch, 5) solve R X = B with dtrsm_interleave_batch. 

Now I need to apply the row (?) permutation to get the true X. 

I have tried the following process

    // The solution X is now sitting in the B_p buffer.
    // The interleaved pivot vectors P**i are in jvpt_p.

    std::vector<double> col_B(m*nrhs); // regular column-major array

    for (int i = 0; i < ninter; i++) {

        // Deinterleave
        ARMPL_CHECK(
            armpl_dge_deinterleave(ninter, i,
                m, nrhs, col_B.data(), 1, m,
                B_p, istrd_B, jstrd_B));

        // Permute
        LAPACKE_dlaswp(LAPACK_COL_MAJOR, nrhs, col_B.data(), m,
            0, m-1, jpvt_p, istrd_jpvt);

        // Print the result vector (first right-hand side only)
        for (int row = 0; row < m; row++) {
            std::cout << col_B[row] << '\n';
        }
    }


but it doesn't give me the expected result.

I would be very grateful if you could add an example of using the batch-interleave QR functions in future releases.


Parents
  • Ignore my threading observation. I meant to write libarmpl_mp and not _pl. The outer multi-threading works just fine when I use clang/flang and hence libomp (the same one linked in the performance libraries), not libgomp. 

     Btw, I noticed in the flamegraph that larfg makes some slow calls to dlapy2 and dnrm2. I was able to shave off a few cycle by linking my own versions (albeit less numerically stable). 

    Best,
    Ivan

Reply
  • Ignore my threading observation. I meant to write libarmpl_mp and not _pl. The outer multi-threading works just fine when I use clang/flang and hence libomp (the same one linked in the performance libraries), not libgomp. 

     Btw, I noticed in the flamegraph that larfg makes some slow calls to dlapy2 and dnrm2. I was able to shave off a few cycle by linking my own versions (albeit less numerically stable). 

    Best,
    Ivan

Children
No data