xuyang1706 commented on issue #8631: [FLINK-12745][ml] add sparse and dense
vector class, and dense matrix class with basic operations.
URL: https://github.com/apache/flink/pull/8631#issuecomment-521635674
> > Do you have any benchmark or performance tests? it would be helpful for
understanding performance this logic
>
> I would like to compare old lineal algebra primitives (e.g. from
https://github.com/apache/flink/tree/47db1ab7ce523390bd957380eea1f6153ffdc7c2/flink-libraries/flink-ml/src/main/scala/org/apache/flink/ml/math)
and this implementation.
>
> Also I cannot find any discussion of design and use cases for classes. Do
we have?
> and related to performance
>
> > Did you look linear algebra entities implementations from other
libraries? I understand the risks to use other libraries in flink, but we can
get well tested and performance-optimized implementation of vectors and matrix
as an improvement .
@ex00 we compared with the old flinkml lineal algebra's Vector
implementation, same vector data and repeat 1 times.
| vector op | org.apache.flink.ml.math |
com.alibaba.alink.common.linalg |
| -- | |
--- |
| DenseVector dot | 12.4s| 8.2s
|
| DenseVector axpy | 10.8s | 9.9s
|
| DenseVector scale | 7.5s | 5.9s
|
| SparseVector dot |15.5s |11.5s
|
| SparseVector scale | 0.67s| 0.40s
|
Environment:
*machine*
- 2.2 GHz Intel Core i7
- 16 GB 1600 MHz DDR3
- mac os
*jvm*
java version "1.8.0_121"
Java(TM) SE Runtime Environment (build 1.8.0_121-b13)
Java HotSpot(TM) 64-Bit Server VM (build 25.121-b13, mixed mode)
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:
us...@infra.apache.org
With regards,
Apache Git Services