Re: Fw: Questions/query about recode / transform in systemML

2016-05-24 Thread Matthias Boehm
thanks for the question Alok - couple of comments: Q1) Independent of the number of columns, we will always do two passes, one to compute the recode maps (distinct values) and one to apply the recode maps to your data. Q2) For the distributed case, we have right now only broadcast-based transfor

Re: [VOTE] Apache SystemML 0.10.0-incubating (RC1)

2016-05-24 Thread Matthias Boehm
+1 In detail, I ran our performance testsuite on Spark 1.6.1 for data sizes {80MB, 800MB, 8GB, 80GB}, sparse/dense, intercept 0/1/2, and the algorithm classes binomial (Mlogreg, L2SVM, MSVM), multinomial (Mlogreg, MSVM, Naive Bayes), regression (LinregCG, LinregDS, GLM poisson-log, GLM gamma-log,

Re: Discussion on GPU backend

2016-05-24 Thread Niketan Pansare
Hi all, Since there is interest in collaborating on GPU backend, I wanted to know what is the preferred way to go ahead with a new feature (i.e. GPU backend) ? This discussion is also generally applicable to other major features (for example: Flink backend, Deep Learning support, OpenCL backend,

Location for release validation checklist?

2016-05-24 Thread Deron Eriksson
Hi, The release candidate validation checklist needs to be easy to modify by SystemML committers/contributors. As a result, it should probably take a 'Wiki' form. This raises the question of where it should go? Seems like the options are: 1) md file in the project itself (not a wiki, not easy to

missing release candidate checksums?

2016-05-24 Thread Deron Eriksson
Hi, I noticed that not all the artifacts at https://dist.apache.org/repos/dist/dev/incubator/systemml/0.10.0-incubating-rc1/ have md5 checksums. Also, the previous release (see https://repo1.maven.org/maven2/org/apache/systemml/systemml/0.9.0-incubating/) featured sha1 checksums but the current r

Re: Location for release validation checklist?

2016-05-24 Thread Luciano Resende
On Tue, May 24, 2016 at 5:16 PM, Deron Eriksson wrote: > Hi, > > The release candidate validation checklist needs to be easy to modify by > SystemML committers/contributors. As a result, it should probably take a > 'Wiki' form. This raises the question of where it should go? Seems like the > opti

Re: missing release candidate checksums?

2016-05-24 Thread Luciano Resende
On Tue, May 24, 2016 at 5:20 PM, Deron Eriksson wrote: > Hi, > > I noticed that not all the artifacts at > > https://dist.apache.org/repos/dist/dev/incubator/systemml/0.10.0-incubating-rc1/ > have md5 checksums. > > Also, the previous release (see > > https://repo1.maven.org/maven2/org/apache/sys

Re: missing release candidate checksums?

2016-05-24 Thread Deron Eriksson
In my opinion, I don't think restarting the vote is needed for this. On Tue, May 24, 2016 at 8:21 PM, Luciano Resende wrote: > On Tue, May 24, 2016 at 5:20 PM, Deron Eriksson > wrote: > > > Hi, > > > > I noticed that not all the artifacts at > > > > > https://dist.apache.org/repos/dist/dev/inc

Re: Discussion on GPU backend

2016-05-24 Thread Matthias Boehm
Generally, I think we should really stick to (3) as done in the past, i.e., bring up major features in the roadmap discussions, create jira epics and try to break them into rather isolated tasks. This works for almost any major/minor feature. The only exception are features, where it is initially

Re: missing release candidate checksums?

2016-05-24 Thread Matthias Boehm
good catch Deron - and I agree, adding the checksums should be fine. Regards, Matthias From: Deron Eriksson To: dev@systemml.incubator.apache.org Date: 05/24/2016 08:42 PM Subject:Re: missing release candidate checksums? In my opinion, I don't think restarting the vote is n

Re: Location for release validation checklist?

2016-05-24 Thread Deron Eriksson
Hi, Personally, I would like if we can go to a single page and see the current validation status as a checklist (that can be updated by any contributors). In the examples I sent, in a couple seconds I can see that 1) no one has run the test suite on the src package, 2) performance suite hasn't bee