[
https://issues.apache.org/jira/browse/MAHOUT-1885?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15824821#comment-15824821
]
ASF GitHub Bot commented on MAHOUT-1885:
----------------------------------------
Github user andrewpalumbo commented on the issue:
https://github.com/apache/mahout/pull/261
Interesting.. After fixing a few small bugs in the `viennacl-omp` module
(00fbe3e), all tests pass with the exception of `sparse mmul
microbenchmark`, which fails with a
``` java.lang.OutOfMemoryError: Java heap space
at it.unimi.dsi.fastutil.ints.Int2DoubleOpenHashMap.rehash
```
if I chage the dimensions of the matrices to :
```
val m = 300
val n = 300
val s = 10
```
(one order of magnitude lower on `m` and `n` two lower on `s`, all tests
will pass:
```
Discovery completed in 296 milliseconds.
Run starting. Expected test count is: 6
ViennaCLSuiteOMP:
- row-major viennacl::matrix
[INFO] Creating org.apache.mahout.viennacl.opencl.GPUMMul solver
[WARN] Unable to create class GPUMMul: attempting OpenMP version
[INFO] Creating org.apache.mahout.viennacl.openmp.OMPMMul solver
[INFO] Successfully created org.apache.mahout.viennacl.openmp.OMPMMul solver
jvmRWRW
ompRWCW
log4j:WARN No appenders could be found for logger
(org.apache.mahout.viennacl.openmp.OMPMMul$).
log4j:WARN Please initialize the log4j system properly.
log4j:WARN See http://logging.apache.org/log4j/1.2/faq.html#noconfig for
more info.
- mmul microbenchmark
+ Mahout multiplication time: 11444 ms.
+ ViennaCL/cpu/OpenMP multiplication time: 2820 ms.
- trans
[INFO] Creating org.apache.mahout.viennacl.opencl.GPUMMul solver
[WARN] Unable to create class GPUMMul: attempting OpenMP version
[INFO] Creating org.apache.mahout.viennacl.openmp.OMPMMul solver
[INFO] Successfully created org.apache.mahout.viennacl.openmp.OMPMMul solver
ompSparseRWRW
- sparse mmul microbenchmark
+ Mahout Sparse multiplication time: 65 ms.
+ ViennaCL/cpu/OpenMP Sparse multiplication time: 5 ms.
- VCL Dense Matrix %*% Dense vector - no OpenCl
+ Mahout dense matrix %*% dense vector multiplication time: 30 ms.
+ ViennaCL/cpu/OpenMP dense matrix %*% dense vector multiplication time:
16 ms.
[INFO] Creating org.apache.mahout.viennacl.opencl.GPUMMul solver
[WARN] Unable to create class GPUMMul: attempting OpenMP version
[INFO] Creating org.apache.mahout.viennacl.openmp.OMPMMul solver
[INFO] Successfully created org.apache.mahout.viennacl.openmp.OMPMMul solver
ompSparseRowRWRW
- Sparse %*% Dense mmul microbenchmark
+ Mahout multiplication time: 2583 ms.
+ ViennaCL/cpu/OpenMP multiplication time: 2639 ms.
Run completed in 21 seconds, 602 milliseconds.
Total number of tests run: 6
Suites: completed 2, aborted 0
Tests: succeeded 6, failed 0, canceled 0, ignored 0, pending 0
All tests passed.
[INFO]
------------------------------------------------------------------------
[INFO] BUILD SUCCESS
[INFO]
------------------------------------------------------------------------
[INFO] Total time: 40.433 s
[INFO] Finished at: 2017-01-16T17:27:49-08:00
[INFO] Final Memory: 21M/327M
[INFO]
------------------------------------------------------------------------
```
What's confusing to me is why they would pass on the GPU module, and not on
the OpenMP module, which use almost Identical methods to convert back and forth
to interface with ViennaCL operations. As well, this same test is run for
comparison in the GPU module. What an annoying little bug. Must be a maven
test setting.
> 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)
