https://pfspipe.ipmu.jp/jira This file is an XML representation of an issue en-us 8.3.4 803005 13-09-2019 https://pfspipe.ipmu.jp/jira/browse/PIPE2D-862 DRP 2-D Pipeline <p>The 2d psf generating code is considerable slower when running under LSST imports, when compared to when being run with same libraries outside LSST environment. Testing reveals that to generate a 20x20 images in focus,  same code under LSST env takes around 1.3 seconds, and around 0.85 seconds outside LSST enviroment.</p> <p>Profliing is in support of my suspicion and previous result that the main slowdown is due to slower FFT calculations. Profiling results are attached as ``LSST_package'' and ``my_package''. They are complex to read, but careful comparison shows that that FFT calculations take larger fraction of time under LSST enviroment. This is almost certainly due to the fact that LSST env does not support MKL. During Monday meeting, it has benn clarified that this is due to legacy support which is not needed any more, and it should be relatively easy and painfree to turn MKL support back on.</p> PIPE2D-862

Make pipeline compatible with MKL libraries

Task Normal Done Done price ncaplar Tue, 22 Jun 2021 18:23:47 +0000 Fri, 28 Jan 2022 17:16:27 +0000 Tue, 29 Jun 2021 19:47:26 +0000 0 4 <p><a href="https://pfspipe.ipmu.jp/jira/secure/ViewProfile.jspa?name=price" class="user-hover" rel="price">price</a> when you get back next week, what would it take to provide a version of numpy with MKL if we ignore the sincPhotometry problems? I bet we could fix that if needs be, but I don't expect to need to run that algorithm in PFS.</p> <p><a href="https://pfspipe.ipmu.jp/jira/secure/ViewProfile.jspa?name=ncaplar" class="user-hover" rel="ncaplar">ncaplar</a>, could you please provide a test script that I can run that will time the operations you're interested in? Then I can run that script with and without the changes I'm thinking of to see if they are effective.</p> <p>Yes, of course. Give me a bit of time. </p> <p>I have placed the smallest script at /tigress/ncaplar/Images/<a href="https://pfspipe.ipmu.jp/jira/browse/PIPE2D-862" title="Make pipeline compatible with MKL libraries " class="issue-link" data-issue-key="PIPE2D-862"><del>PIPE2D-862</del></a>, called test.py. It import a single array in and runs fft on it, with imports scipy.fftpack. Is that what you had in mind? (I also placed a single image showing that the code execute faster in ``my environment'', and slower in LSST env)</p> <p>Activating MKL in a Docker container on my MacBook resulted in about a 2x speedup for the FFT operation:</p> <div class="code panel" style="border-width: 1px;"><div class="codeContent panelContent"> <pre class="code-java"> ncalls tottime percall cumtime percall filename:lineno(function) 10 0.000 0.000 2.217 0.222 /opt/pfs/python/miniconda3-4.5.12/envs/lsst-scipipe-1172c30/lib/python3.7/site-packages/scipy/fftpack/basic.py:370(fft2) </pre> </div></div> <p>vs:</p> <div class="code panel" style="border-width: 1px;"><div class="codeContent panelContent"> <pre class="code-java"> 10 0.000 0.000 5.029 0.503 /opt/pfs/python/miniconda3-4.5.12/envs/lsst-scipipe-1172c30/lib/python3.7/site-packages/scipy/fftpack/basic.py:673(fft2) </pre> </div></div> <p>I'll update our install scripts and do a bit more testing to be sure this doesn't break anything else.</p> <p>Integration test passed, I think we can deploy this.</p> <p>Did you also perhaps test the execution times with the test code I have provided?</p> <p>Changes made approved. But check comment from Neven above.</p> <p>Yes, same result as before.</p> <p>I meant regarding the timing, did you also get a speed up making it comparable to non-LSST code? (I can see that for myself once when deployed)</p> <p>Yes, I got the same timing results as before.</p> <p>Hm, so no speed up when using MKL libraries within LSST code with test.py? Am I understanding this correctly?</p> <blockquote> <p>about a 2x speedup for the FFT operation</p></blockquote> <p>Ok, I understand. I thought that ``same result as before'' was to be interpreted  as ``same result as before the changes'', and I was super confused. </p> <div class="code panel" style="border-width: 1px;"><div class="codeContent panelContent"> <pre class="code-java"> (lsst-scipipe) pprice@tiger2-sumire:/tigress/ncaplar/Images/PIPE2D-862 $ time python test.py test ended Time spent in user mode : 0.582 sec Time spent in kernel mode : 0.235 sec Total time : 0.820 sec CPU utilisation : 99.60% </pre> </div></div> <p>Merged and deployed.</p> Development Rank 0|02qpq3:r2o Reviewers hassan