[
https://issues.apache.org/jira/browse/SPARK-20446?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15981648#comment-15981648
]
Nick Pentreath commented on SPARK-20446:
----------------------------------------
By "compare to DataFrame implementation" I mean the current "recommendAll"
methods in master/branch-2.2 for {{ALSModel}}. Did you compare against that? If
so what was the result?
Reason being, conceptually that is doing something fairly similar (computing
the vector dot products rather than matrix-matrix multiply, followed by a
priority queue aggregator for top-k). The idea was that this SparkSQL approach
would be more efficient. In practice I didn't find this to be the case for
large data sizes, when comparing to my approach with BLAS 3 (though granted yes
there is potential for more GC pressure).
Also, there is really no point in doing the "blockify" operation in this case
right? As you're not using BLAS 3 anyway, so blocking is unnecessary and the
block size param is irrelevant.
> Optimize the process of MLLIB ALS recommendForAll
> -------------------------------------------------
>
> Key: SPARK-20446
> URL: https://issues.apache.org/jira/browse/SPARK-20446
> Project: Spark
> Issue Type: Improvement
> Components: ML, MLlib
> Affects Versions: 2.3.0
> Reporter: Peng Meng
>
> The recommendForAll of MLLIB ALS is very slow.
> GC is a key problem of the current method.
> The task use the following code to keep temp result:
> val output = new Array[(Int, (Int, Double))](m*n)
> m = n = 4096 (default value, no method to set)
> so output is about 4k * 4k * (4 + 4 + 8) = 256M. This is a large memory and
> cause serious GC problem, and it is frequently OOM.
> Actually, we don't need to save all the temp result. Suppose we recommend
> topK (topK is about 10, or 20) product for each user, we only need 4k * topK
> * (4 + 4 + 8) memory to save the temp result.
> I have written a solution for this method with the following test result.
> The Test Environment:
> 3 workers: each work 10 core, each work 30G memory, each work 1 executor.
> The Data: User 480,000, and Item 17,000
> BlockSize: 1024 2048 4096 8192
> Old method: 245s 332s 488s OOM
> This solution: 121s 118s 117s 120s
>
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
