[
https://issues.apache.org/jira/browse/MAHOUT-1885?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15582746#comment-15582746
]
ASF GitHub Bot commented on MAHOUT-1885:
----------------------------------------
Github user andrewpalumbo commented on the issue:
https://github.com/apache/mahout/pull/261
I'd really opened this PR so that we could discuss architecture and the
correct entry points for the SolverFactory, hence the ridiculously verbose and
confusing debugging output. I'd like to get the arch right before we commit.
At its core it works as is, but there are alot of changes that need to be made.
Eg. We may be able to get away with only one Viennacl module by setting an
environment variable and checking for it in Context.scala if it exists we
"#define VIENNACL_WITH_OPENCL" which will make the module OpenMP only. The
Problem With This Is That We Can't Test Gpu in this module.. without checking
for the environment var.
The other big thing is that I don't know if the placement if the
SolverFactory as it is is correct.. we may want to examine this, as well as
cache a XXXMMUL Of both types the first time this class is hit so that we are
not dynamically loading a class each time %*% is hit.
These are a couple of things that I'd like to get figured out before
committing.
We've already identified a few pain points in the testing that everybody
did over the weekend. So let's get it right before committing.
Sent from my Verizon Wireless 4G LTE smartphone
-------- Original message --------
From: Dmitriy Lyubimov <notificati...@github.com>
Date: 10/17/2016 8:51 AM (GMT-08:00)
To: apache/mahout <mah...@noreply.github.com>
Cc: Andrew Palumbo <ap....@outlook.com>, Author <aut...@noreply.github.com>
Subject: Re: [apache/mahout] MAHOUT-1885 [WIP][FOR COMMENT]: Inital
implementation of VCL bindings for mahout math. (#261)
IMO this needs more work and committing this is premature at this point due
to architecture considerations. Not only it misses some things, but also IMO
has things that should not be done the way they currently are.
-
You are receiving this because you authored the thread.
Reply to this email directly, view it on
GitHub<https://github.com/apache/mahout/pull/261#issuecomment-254248310>, or
mute the
thread<https://github.com/notifications/unsubscribe-auth/AHU2HUmvxvJwfx7bWgzjiKJuNdmtei49ks5q05mMgaJpZM4KSKUm>.
> Inital Implementation of VCL Bindings
> -------------------------------------
>
> Key: MAHOUT-1885
> URL: https://issues.apache.org/jira/browse/MAHOUT-1885
> Project: Mahout
> Issue Type: Improvement
> Components: Math
> Affects Versions: 0.12.2
> Reporter: Andrew Palumbo
> Assignee: Andrew Palumbo
> Fix For: 0.13.0
>
>
> Push a working experimental branch of VCL bindings into master. There is
> still a lot of work to be done. All tests are passing, At the moment there
> am opening this JIRA mostly to get a number for PR and to test profiles
> against on travis.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
