yma11 commented on a change in pull request #27546: [SPARK-30773][ML]Support
NativeBlas for level-1 routines
URL: https://github.com/apache/spark/pull/27546#discussion_r384968081
##########
File path: mllib-local/src/main/scala/org/apache/spark/ml/linalg/BLAS.scala
##########
@@ -27,15 +27,35 @@ private[spark] object BLAS extends Serializable {
@transient private var _f2jBLAS: NetlibBLAS = _
@transient private var _nativeBLAS: NetlibBLAS = _
+ @transient private val vectorSizeThreshold: Int = 256
Review comment:
@srowen I think @kiszk make a good point. With properly tuned based on
specific workload and runtime environment, nativeBLAS may can bring much
benefit than f2jBLAS even vector size < 256. Making it configurable can
provide user a simpler way to use nativeBLAS in Spark ML when vector size is
not so big. Do you think so? if so, we need to add spark-core dependency in
spark-mllib-local package.
Thanks @kiszk for your comments.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
