systemml-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Matthias Boehm <mboe...@gmail.com>
Subject Re: [VOTE] Apache SystemML 1.0.0 (RC2)
Date Tue, 12 Dec 2017 23:26:26 GMT
that's a good observation - the reasons for this behavior are (1) a new
instruction scheduling algorithm that significantly improved memory
efficiency, and reduced evictions from CPU/GPU memory, as well as (2) new
rewrites for merging sequences of statement blocks into a single DAG. Since
there are no data dependencies or functions with side effects (like
writes), theses prints are reordered.

I don't think this should block the release, but once the release is out,
I'll extend the new instruction scheduling to sort prints by their original
line numbers.

Regards,
Matthias

On Tue, Dec 12, 2017 at 2:43 PM, Mike Dusenberry <dusenberrymw@gmail.com>
wrote:

> I tried running the `nn` tests using `git checkout
> 7b44380e49425ac20912c2d42de8603a06fa55a1; mvn clean package; cd
> scripts; ../bin/systemml nn/test/run_tests.dml`, and while the tests pass,
> they run in reverse order from how they are written in the file.  I.e., if
> you run it, it will first run the print statements at the bottom of the
> file, then run the `test::...` functions one-by-one starting from the
> bottom, etc.  That doesn't look right to me.  I've pasted the output of the
> tests below.
>
> $ ../bin/systemml nn/test/run_tests.dml
> ============================================================
> ====================
> ============================================================
> ====================
> 17/12/12 14:38:42 INFO api.DMLScript: BEGIN DML run 12/12/2017 14:38:42
> 17/12/12 14:38:44 WARN util.NativeCodeLoader: Unable to load native-hadoop
> library for your platform... using builtin-java classes where applicable
> ---
> Other tests complete -- look for any ERRORs or WARNINGs.
>
>
> Testing the tanh forward function.
> Testing the tanh forward function.
> Testing the 2D softmax function.
> 17/12/12 14:38:45 INFO Configuration.deprecation: session.id is
> deprecated.
> Instead, use dfs.metrics.session-id
> 17/12/12 14:38:45 INFO jvm.JvmMetrics: Initializing JVM Metrics with
> processName=JobTracker, sessionId=
> Testing the top_k2d function.
> Testing the top_k function.
>  - Testing top_1.
>  - Testing top_2.
>  - Testing top_All.
> Testing the top_k_row function.
> Testing transpose_NCHW_to_CNHW function.
> Testing the threshold function.
> Testing the tanh forward function.
> Testing the padding and unpadding functions.
> Testing the 2D max pooling functions.
>  - Testing w/ padh=0 & padw=0.
>  - Testing w/ padh=0 & padw=1.
>  - Testing w/ padh=0 & padw=2.
>  - Testing w/ padh=0 & padw=3.
>  - Testing w/ padh=1 & padw=0.
>  - Testing w/ padh=1 & padw=1.
>  - Testing w/ padh=1 & padw=2.
>  - Testing w/ padh=1 & padw=3.
>  - Testing w/ padh=2 & padw=0.
>  - Testing w/ padh=2 & padw=1.
>  - Testing w/ padh=2 & padw=2.
>  - Testing w/ padh=2 & padw=3.
>  - Testing w/ padh=3 & padw=0.
>  - Testing w/ padh=3 & padw=1.
>  - Testing w/ padh=3 & padw=2.
>  - Testing w/ padh=3 & padw=3.
>  - Testing for correct behavior against known answer w/ pad=0.
>  - Testing for correct behavior against known answer w/ pad=1.
>  - Testing for correct behavior against known answer w/ all negative matrix
> w/ pad=0.
>  - Testing for correct behavior against known answer w/ all negative matrix
> w/ pad=1.
> Testing the im2col and col2im functions.
> Testing the 2D cross-entropy loss function.
> Testing the cross-entropy loss function with zero-valued predictions.
> Testing the 2D depthwise transpose convolution function.
> Testing the 2D transpose convolution function.
> Testing the 2D depthwise convolution function.
> Testing the 2D convolution functions.
> Testing the 2D (spatial) batch normalization function.
> Testing the 1D batch normalization function.
>
>
> ---
> Grad checks complete -- look for any ERRORs or WARNINGs.
> If any tests involving ReLUs failed, try a few times to ensure that they
> were not false negatives due to kinks being crossed.
>
>
> Starting other tests.
> ---
> Grad checking the tanh nonlinearity layer with L2 loss.
> Grad checking the 2D softmax layer with L2 loss.
> Grad checking the softmax layer with L2 loss.
> Grad checking the sigmoid nonlinearity layer with L2 loss.
> Grad checking the 2D scale & shift layer with L2 loss.
>  - Grad checking X.
>  - Grad checking gamma.
>  - Grad checking beta.
> Grad checking the 1D scale & shift layer with L2 loss.
>  - Grad checking X.
>  - Grad checking gamma.
>  - Grad checking beta.
> Grad checking the simple RNN layer with L2 loss.
>  - Grad checking with return_seq = TRUE
>    - Grad checking X.
>    - Grad checking W.
>    - Grad checking b.
>    - Grad checking out0.
>  - Grad checking with return_seq = FALSE
>    - Grad checking X.
>    - Grad checking W.
>    - Grad checking b.
>    - Grad checking out0.
> Grad checking the ReLU nonlinearity layer with L2 loss.
> Grad checking the simple reference 2D max pooling layer with L2 loss.
>  - Grad checking w/ pad=0.
>  - Grad checking w/ pad=1.
> Grad checking the built-in 2D max pooling layer with L2 loss.
>  - Grad checking w/ pad=0.
>  - Grad checking w/ pad=1.
> Grad checking the 2D max pooling layer with L2 loss.
>  - Grad checking w/ pad=0.
>  - Grad checking w/ pad=1.
> Grad checking the LSTM layer with L2 loss.
>  - Grad checking with return_seq = TRUE
>    - Grad checking X.
>    - Grad checking W.
>    - Grad checking b.
>    - Grad checking out0.
>    - Grad checking c0.
>  - Grad checking with return_seq = FALSE
>    - Grad checking X.
>    - Grad checking W.
>    - Grad checking b.
>    - Grad checking out0.
>    - Grad checking c0.
> Grad checking the (inverted) dropout layer with L2 loss.
> Grad checking the 2D depthwise transpose convolutional layer with L2 loss.
>  - Grad checking X.
>  - Grad checking W.
>  - Grad checking b.
>  - Grad checking X.
> Grad checking the 2D transpose convolutional layer with L2 loss.
>  - Grad checking W.
>  - Grad checking b.
> Grad checking the 2D depthwise convolutional layer with L2 loss.
>  - Grad checking X.
>  - Grad checking W.
>  - Grad checking b.
> Grad checking the simple reference 2D convolutional layer with L2 loss.
>  - Grad checking X.
>  - Grad checking W.
>  - Grad checking b.
>  - Grad checking X.
> Grad checking the built-in 2D convolutional layer with L2 loss.
>  - Grad checking W.
>  - Grad checking b.
> Grad checking the `im2col` 2D convolutional layer with L2 loss.
>  - Grad checking X.
>  - Grad checking W.
>  - Grad checking b.
> Grad checking the 2D (spatial) batch normalization layer with L2 loss.
>  - Grad checking the 'train' mode.
>    - Grad checking X.
>    - Grad checking gamma.
>    - Grad checking beta.
>  - Grad checking the 'test' mode.
>    - Grad checking X.
>    - Grad checking gamma.
>    - Grad checking beta.
> Grad checking the 1D batch normalization layer with L2 loss.
>  - Grad checking the 'train' mode.
>    - Grad checking X.
>    - Grad checking gamma.
>    - Grad checking beta.
>  - Grad checking the 'test' mode.
>    - Grad checking X.
>    - Grad checking gamma.
>    - Grad checking beta.
> Grad checking the affine layer with L2 loss.
>  - Grad checking X.
>  - Grad checking W.
>  - Grad checking b.
>
> Grad checking the log loss function.
> Grad checking the L2 regularization function.
> Grad checking the L2 loss function.
> Grad checking the L1 regularization function.
> Grad checking the L1 loss function.
> Grad checking the 2D cross-entropy loss function.
> Grad checking the cross-entropy loss function.
>
> Starting grad checks.
> ---
> SystemML Statistics:
> Total execution time: 36.865 sec.
> Number of executed MR Jobs: 0.
>
> 17/12/12 14:39:22 INFO api.DMLScript: END DML run 12/12/2017 14:39:22
>
>
> --
>
> Mike Dusenberry
> GitHub: github.com/dusenberrymw
> LinkedIn: linkedin.com/in/mikedusenberry
>
> On Tue, Dec 12, 2017 at 1:35 PM, Krishna Kalyan <krishnakalyan3@gmail.com>
> wrote:
>
> > +1
> > Ran performance tests upto 80GB on 3 node  m3.xlarge AWS instance without
> > issues.
> >
> > Regards,
> > Krishna
> >
> >
> > On Tue, Dec 12, 2017 at 10:26 PM, Frederick R Reiss <frreiss@us.ibm.com>
> > wrote:
> >
> > > +1
> > >
> > > Fred
> > >
> > > [image: Inactive hide details for "Niketan Pansare" ---12/11/2017
> > 03:05:28
> > > PM---+1. Ran GPU tests as well as sanity test for Python API]"Niketan
> > > Pansare" ---12/11/2017 03:05:28 PM---+1. Ran GPU tests as well as
> sanity
> > > test for Python APIs. Thanks,
> > >
> > > From: "Niketan Pansare" <npansar@us.ibm.com>
> > > To: dev@systemml.apache.org
> > > Date: 12/11/2017 03:05 PM
> > > Subject: Re: [VOTE] Apache SystemML 1.0.0 (RC2)
> > > ------------------------------
> > >
> > >
> > >
> > > +1. Ran GPU tests as well as sanity test for Python APIs.
> > >
> > > Thanks,
> > >
> > > Niketan Pansare
> > > IBM Almaden Research Center
> > > E-mail: npansar At us.ibm.com
> > > *http://researcher.watson.ibm.com/researcher/view.php?
> person=us-npansar*
> > > <http://researcher.watson.ibm.com/researcher/view.php?
> person=us-npansar>
> > >
> > > "Glenn Weidner" ---12/11/2017 09:49:48 AM---+1 I ran Linear Regression,
> > > Logistic Regression, SVM, Naive Bayes Python tests
> > >
> > > From: "Glenn Weidner" <gweidner@us.ibm.com>
> > > To: dev@systemml.apache.org
> > > Date: 12/11/2017 09:49 AM
> > > Subject: Re: [VOTE] Apache SystemML 1.0.0 (RC2)
> > > ------------------------------
> > >
> > >
> > >
> > > +1
> > >
> > > I ran Linear Regression, Logistic Regression, SVM, Naive Bayes Python
> > > tests with Spark 2.1 in cloud environment. All tests passed with Python
> > 2.7.
> > >
> > > Note I also attempted to repeat the tests with Python 3.5
> 'experimental'
> > > kernel but encountered issue:
> > >
> > > .local/lib/python3.5/site-packages/systemml/mllearn/estimators.py",
> line
> > > 887
> > >  def __init__(self, sparkSession, keras_model, input_shape,
> > > transferUsingDF=False, weights=None, labels=None):
> > > SyntaxError: import * only allowed at module level"
> > >
> > > I don't believe this is release blocking issue but may want to discuss
> if
> > > other problems reported during RC2 validation.
> > >
> > > Regards,
> > > Glenn
> > >
> > > Matthias Boehm ---12/09/2017 03:13:43 PM---+1 I ran the perftest suite
> > > with the artifact on Spark 2.2 up to 80GB without
> > >
> > > From: Matthias Boehm <mboehm7@gmail.com>
> > > To: dev@systemml.apache.org
> > > Date: 12/09/2017 03:13 PM
> > > Subject: Re: [VOTE] Apache SystemML 1.0.0 (RC2)
> > > ------------------------------
> > >
> > >
> > >
> > > +1
> > >
> > > I ran the perftest suite with the artifact on Spark 2.2 up to 80GB
> > without
> > > any failures or performance issues. On earlier versions, I also ran the
> > > perftest suite with Spark 2.1 and 2.2, w/ and w/o codegen, and w/ auto
> > > compression up to 800GB without remaining issues.
> > >
> > > As a minor nitpick (to be resolved in any additional RCs or the next
> > > release), the root directory of the artifact includes a
> > SystemML-config.xml
> > > file that only has a small subset of configuration parameters. We might
> > > want to either delete this or pull in a copy of
> > > SystemML-config.xml.template from the conf directory.
> > >
> > > Regards,
> > > Matthias
> > >
> > > On Sat, Dec 9, 2017 at 1:53 AM, Luciano Resende <luckbr1975@gmail.com>
> > > wrote:
> > >
> > > > On Fri, Dec 8, 2017 at 12:15 PM, Berthold Reinwald <
> > reinwald@us.ibm.com>
> > > > wrote:
> > > >
> > > > > Please vote on releasing the following candidate as Apache SystemML
> > > > > version 1.0.0
> > > > >
> > > > > The vote is open for at least 72 hours and passes if a majority of
> at
> > > > > least 3 +1 PMC votes are cast.
> > > > >
> > > > > [ ] +1 Release this package as Apache SystemML 1.0.0
> > > > > [ ] -1 Do not release this package because ...
> > > > >
> > > > > To learn more about Apache SystemML, please see
> > > > >
> > > *https://urldefense.proofpoint.com/v2/url?u=http-
> > 3A__systemml.apache.org_&d=DwIBaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=
> > we2L9Q2NOeniL5PbfdrO3CmQPNy_vnhoDpPDviJpnrU&m=Ud62VUVZNsp0I0PlFMRr-
> > OOvxUC5pssmMZdUB4I3g5w&s=KKHMDbNiFLfVSqUYXqIZLGxx-yif5PAGAQQLnnMfXR8&e=*
> > > <https://urldefense.proofpoint.com/v2/url?u=http-
> > 3A__systemml.apache.org_&d=DwIBaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=
> > we2L9Q2NOeniL5PbfdrO3CmQPNy_vnhoDpPDviJpnrU&m=Ud62VUVZNsp0I0PlFMRr-
> > OOvxUC5pssmMZdUB4I3g5w&s=KKHMDbNiFLfVSqUYXqIZLGxx-yif5PAGAQQLnnMfXR8&e=>
> > > > >
> > > > > The tag to be voted on is v1.0.0-rc2 (
> > > > > 7b44380e49425ac20912c2d42de8603a06fa55a1):
> > > > >
> > > *https://urldefense.proofpoint.com/v2/url?u=https-
> 3A__github.com_apache_
> > systemml_commit_7b44380e49425ac20912c2d42de860
> &d=DwIBaQ&c=jf_iaSHvJObTbx-
> > siA1ZOg&r=we2L9Q2NOeniL5PbfdrO3CmQPNy_vnhoDpPDviJpnrU&m=
> > Ud62VUVZNsp0I0PlFMRr-OOvxUC5pssmMZdUB4I3g5w&s=k8A_
> > OU7p9Wa4sKxpiY5v9gzbLrJg8rGTSiSEZ-hFgTA&e=*
> > > <https://urldefense.proofpoint.com/v2/url?u=https-
> 3A__github.com_apache_
> > systemml_commit_7b44380e49425ac20912c2d42de860
> &d=DwIBaQ&c=jf_iaSHvJObTbx-
> > siA1ZOg&r=we2L9Q2NOeniL5PbfdrO3CmQPNy_vnhoDpPDviJpnrU&m=
> > Ud62VUVZNsp0I0PlFMRr-OOvxUC5pssmMZdUB4I3g5w&s=k8A_
> > OU7p9Wa4sKxpiY5v9gzbLrJg8rGTSiSEZ-hFgTA&e=>
> > > > > 3a06fa55a1
> > > > >
> > > > >
> > > > Looks like the tag hash should be
> > > > 6519f3383f8fd43af93ecba8742ae4b2b28d4b35
> > > >
> > > > And the link
> > > >
> > > *https://urldefense.proofpoint.com/v2/url?u=https-
> 3A__github.com_apache_
> > systemml_tree_6519f3383f8fd43af93ecba8742ae4&d=DwIBaQ&c=jf_iaSHvJObTbx-
> > siA1ZOg&r=we2L9Q2NOeniL5PbfdrO3CmQPNy_vnhoDpPDviJpnrU&m=
> > Ud62VUVZNsp0I0PlFMRr-OOvxUC5pssmMZdUB4I3g5w&s=ntv-qbRObkRBbs_a2UlTXf-
> > Oe1zcYp7itwRbclOi0d4&e=*
> > > <https://urldefense.proofpoint.com/v2/url?u=https-
> 3A__github.com_apache_
> > systemml_tree_6519f3383f8fd43af93ecba8742ae4&d=DwIBaQ&c=jf_iaSHvJObTbx-
> > siA1ZOg&r=we2L9Q2NOeniL5PbfdrO3CmQPNy_vnhoDpPDviJpnrU&m=
> > Ud62VUVZNsp0I0PlFMRr-OOvxUC5pssmMZdUB4I3g5w&s=ntv-qbRObkRBbs_a2UlTXf-
> > Oe1zcYp7itwRbclOi0d4&e=>
> > > > b2b28d4b35
> > > >
> > > >
> > > > >
> > > > > The release artifacts can be found at:
> > > > >
> > > *https://urldefense.proofpoint.com/v2/url?u=https-
> > 3A__dist.apache.org_repos_dist_dev_systemml_1.0.0-2Drc2_
> > &d=DwIBaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=we2L9Q2NOeniL5PbfdrO3CmQPNy_
> > vnhoDpPDviJpnrU&m=Ud62VUVZNsp0I0PlFMRr-OOvxUC5pssmMZdUB4I3g5w&s=
> > vAMEHElSXjrqufLuWtyr235fwADLB3q0dYRx2w8DDRU&e=*
> > > <https://urldefense.proofpoint.com/v2/url?u=https-
> > 3A__dist.apache.org_repos_dist_dev_systemml_1.0.0-2Drc2_
> > &d=DwIBaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=we2L9Q2NOeniL5PbfdrO3CmQPNy_
> > vnhoDpPDviJpnrU&m=Ud62VUVZNsp0I0PlFMRr-OOvxUC5pssmMZdUB4I3g5w&s=
> > vAMEHElSXjrqufLuWtyr235fwADLB3q0dYRx2w8DDRU&e=>
> > > > >
> > > > >
> > > > > The maven release artifacts, including signatures, digests, etc.
> can
> > be
> > > > > found at:
> > > > >
> > > *https://urldefense.proofpoint.com/v2/url?u=https-
> > 3A__repository.apache.org_content_repositories_&d=
> DwIBaQ&c=jf_iaSHvJObTbx-
> > siA1ZOg&r=we2L9Q2NOeniL5PbfdrO3CmQPNy_vnhoDpPDviJpnrU&m=
> > Ud62VUVZNsp0I0PlFMRr-OOvxUC5pssmMZdUB4I3g5w&s=_
> > 6Aa2MkhOCYdwj5euVEuxQn8w6zaa42kObUUTFF_vSM&e=*
> > > <https://urldefense.proofpoint.com/v2/url?u=https-
> > 3A__repository.apache.org_content_repositories_&d=
> DwIBaQ&c=jf_iaSHvJObTbx-
> > siA1ZOg&r=we2L9Q2NOeniL5PbfdrO3CmQPNy_vnhoDpPDviJpnrU&m=
> > Ud62VUVZNsp0I0PlFMRr-OOvxUC5pssmMZdUB4I3g5w&s=_
> > 6Aa2MkhOCYdwj5euVEuxQn8w6zaa42kObUUTFF_vSM&e=>
> > > > > orgapachesystemml-1026/org/apache/systemml/systemml/1.0.0/
> > > > >
> > > > >
> > > > >
> > > > > ===========================
> > > > > == Apache Release policy ==
> > > > > ===========================
> > > > >
> > > *https://urldefense.proofpoint.com/v2/url?u=http-
> > 3A__www.apache.org_legal_release-2Dpolicy.html&d=
> DwIBaQ&c=jf_iaSHvJObTbx-
> > siA1ZOg&r=we2L9Q2NOeniL5PbfdrO3CmQPNy_vnhoDpPDviJpnrU&m=
> > Ud62VUVZNsp0I0PlFMRr-OOvxUC5pssmMZdUB4I3g5w&s=
> > czTEYvafHC0iIo3EU86chnYqtnqf5jSssAFMspZE698&e=*
> > > <https://urldefense.proofpoint.com/v2/url?u=http-
> > 3A__www.apache.org_legal_release-2Dpolicy.html&d=
> DwIBaQ&c=jf_iaSHvJObTbx-
> > siA1ZOg&r=we2L9Q2NOeniL5PbfdrO3CmQPNy_vnhoDpPDviJpnrU&m=
> > Ud62VUVZNsp0I0PlFMRr-OOvxUC5pssmMZdUB4I3g5w&s=
> > czTEYvafHC0iIo3EU86chnYqtnqf5jSssAFMspZE698&e=>
> > > > >
> > > > > =======================================
> > > > > == How can I help test this release? ==
> > > > > =======================================
> > > > > If you are a SystemML user, you can help us test this release by
> > taking
> > > > an
> > > > > existing Algorithm or workload and running on this release
> candidate,
> > > > then
> > > > > reporting any regressions.
> > > > >
> > > > > ================================================
> > > > > == What justifies a -1 vote for this release? ==
> > > > > ================================================
> > > > > -1 votes should only occur for significant stop-ship bugs or legal
> > > > related
> > > > > issues (e.g. wrong license, missing header files, etc). Minor bugs
> or
> > > > > regressions should not block this release.
> > > > >
> > > > >
> > > >
> > > >
> > > > --
> > > > Luciano Resende
> > > >
> > > *https://urldefense.proofpoint.com/v2/url?u=http-
> > 3A__twitter.com_lresende1975&d=DwIBaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=
> > we2L9Q2NOeniL5PbfdrO3CmQPNy_vnhoDpPDviJpnrU&m=Ud62VUVZNsp0I0PlFMRr-
> > OOvxUC5pssmMZdUB4I3g5w&s=KPnK6-jyZxLeNWGTHI4xIlUsTq_bQ2mQk_DlpRTX4ZY&e=*
> > > <https://urldefense.proofpoint.com/v2/url?u=http-
> > 3A__twitter.com_lresende1975&d=DwIBaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=
> > we2L9Q2NOeniL5PbfdrO3CmQPNy_vnhoDpPDviJpnrU&m=Ud62VUVZNsp0I0PlFMRr-
> > OOvxUC5pssmMZdUB4I3g5w&s=KPnK6-jyZxLeNWGTHI4xIlUsTq_bQ2mQk_DlpRTX4ZY&e=>
> > > >
> > > *https://urldefense.proofpoint.com/v2/url?u=http-
> > 3A__lresende.blogspot.com_&d=DwIBaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=
> > we2L9Q2NOeniL5PbfdrO3CmQPNy_vnhoDpPDviJpnrU&m=Ud62VUVZNsp0I0PlFMRr-
> > OOvxUC5pssmMZdUB4I3g5w&s=GgeilBeHqQtVhhfRNkcKvk0gRIDPkvlZu15YRs1XxLk&e=*
> > > <https://urldefense.proofpoint.com/v2/url?u=http-
> > 3A__lresende.blogspot.com_&d=DwIBaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=
> > we2L9Q2NOeniL5PbfdrO3CmQPNy_vnhoDpPDviJpnrU&m=Ud62VUVZNsp0I0PlFMRr-
> > OOvxUC5pssmMZdUB4I3g5w&s=GgeilBeHqQtVhhfRNkcKvk0gRIDPkvlZu15YRs1XxLk&e=>
> > > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> >
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message