Mailing-List: contact issues-help@systemml.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@systemml.apache.org
Date: Wed, 18 Oct 2017 02:59:00 +0000 (UTC)
From: "Matthias Boehm (JIRA)" <jira@apache.org>
To: issues@systemml.apache.org
Message-ID: <JIRA.13109848.1508202290000.22980.1508295540079@Atlassian.JIRA>
In-Reply-To: <JIRA.13109848.1508202290000@Atlassian.JIRA>
References: <JIRA.13109848.1508202290000@Atlassian.JIRA> <JIRA.13109848.1508202290246@jira-lw-us.apache.org>
Subject: [jira] [Commented] (SYSTEMML-1965) Refactor nn layers to move the
 computation in forward/backward function known at compile time to init
 function
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
archived-at: Wed, 18 Oct 2017 02:59:05 -0000


    [ https://issues.apache.org/jira/browse/SYSTEMML-1965?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16208746#comment-16208746 ] 

Matthias Boehm commented on SYSTEMML-1965:
------------------------------------------

I agree with [~dusenberrymw] here, let's take this as a use case and see why the multiple rounds of IPA (incl constant propagation and folding) are not sufficient here. Furthermore, code motion (moving loop-invariant operators) automatically outside loops, is on our TODO list for a while now - from a mechanic perspective, it's trivial but on a first attempt, we encountered scenarios where it was counter-productive due to moving (and materializing) intermediates that otherwise would not have been computed at all.

> Refactor nn layers to move the computation in forward/backward function known at compile time to init function
> --------------------------------------------------------------------------------------------------------------
>
>                 Key: SYSTEMML-1965
>                 URL: https://issues.apache.org/jira/browse/SYSTEMML-1965
>             Project: SystemML
>          Issue Type: Bug
>            Reporter: Niketan Pansare
>
> Ideally, we should move the computation known at compile time to init layer, rather than keep in the forward function. This reduces recompilation time and also potentially unnecessary instructions. Here is an example snippet from our conv2d layer:
> {code}
> Hout = as.integer(floor((Hin + 2*padh - Hf)/strideh + 1))
> Wout = as.integer(floor((Win + 2*padw - Wf)/stridew + 1))
> {code}
> [~prithvi_r_s] [~dusenberrymw] [~reinwald] do you have any comments or concerns ?
> [~dusenberrymw] do you have free cycles to take this over ?


--
This message was sent by Atlassian JIRA
(v6.4.14#64029)