systemml-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Niketan Pansare (JIRA)" <j...@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
Date Tue, 17 Oct 2017 17:40:00 GMT

    [ https://issues.apache.org/jira/browse/SYSTEMML-1965?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16208004#comment-16208004
] 

Niketan Pansare commented on SYSTEMML-1965:
-------------------------------------------

My two cents: I think both engine improvement as well as good DML practice go hand in hand.
If `init` is not the right place, then we should add `compute_height` and `compute_width`
functions. Having a loop-invariant code (as part of reusable function) in the for loop is
not advisable even if we fix the constant propagation portion of the engine. 

> 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)

Mime
View raw message