You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Seth Hendrickson (JIRA)" <ji...@apache.org> on 2017/06/20 16:16:00 UTC
[jira] [Commented] (SPARK-21152) Use level 3 BLAS operations in
LogisticAggregator
[ https://issues.apache.org/jira/browse/SPARK-21152?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16056025#comment-16056025 ]
Seth Hendrickson commented on SPARK-21152:
------------------------------------------
cc [~dbtsai] [~mlnick] [~srowen]
BTW, I've been working on this. DB, you and I discussed the caching issue in the past. Here's a comment from DB for reference:
"In the old mllib implementation, I just decided to have a copy of entire standardized dataset and had it cached for simplicity. After talking to couple people for their use cases, many times, they're training models on the same cached dataset for different regularizations, and then the old mllib will cache them again and again which will result pressure on GC and waste some memory space."
> Use level 3 BLAS operations in LogisticAggregator
> -------------------------------------------------
>
> Key: SPARK-21152
> URL: https://issues.apache.org/jira/browse/SPARK-21152
> Project: Spark
> Issue Type: Sub-task
> Components: ML
> Affects Versions: 2.1.1
> Reporter: Seth Hendrickson
>
> In logistic regression gradient update, we currently compute by each individual row. If we blocked the rows together, we can do a blocked gradient update which leverages the BLAS GEMM operation.
> On high dimensional dense datasets, I've observed ~10x speedups. The problem here, though, is that it likely won't improve the sparse case so we need to keep both implementations around, and this blocked algorithm will require caching a new dataset of type:
> {code}
> BlockInstance(label: Vector, weight: Vector, features: Matrix)
> {code}
> We have avoided caching anything beside the original dataset passed to train in the past because it adds memory overhead if the user has cached this original dataset for other reasons. Here, I'd like to discuss whether we think this patch would be worth the investment, given that it only improves a subset of the use cases.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org