j143 opened a new pull request #890: [SYSTEMML-2121] AutoEncoder test for codegenalg suite URL: https://github.com/apache/systemml/pull/890 This patch adds a test case for AutoEncoder with codegen enabled against a corresponding R script. **Some results:** `512SparseRewritesSparkFuseNoRedundancy` ``` # Instruction Time(s) Count 1 sp_mapmm 12.568 71 2 sp_write 6.837 8 3 sp_map+ 5.662 72 4 sp_uasqk+ 5.156 10 5 sp_spoofRATMP17 2.690 10 6 sp_spoofRATMP64 2.169 10 7 sp_spoofRATMP9 2.152 21 8 sp_spoofOPTMP31 1.541 10 9 sp_spoofRATMP50 1.459 20 10 sp_rblk 1.420 6 ``` `512DenseRewritesCP` ``` # Instruction Time(s) Count 1 ba+* 3.267 71 2 sp_rblk 1.987 6 3 write 1.641 8 4 spoofRATMP135 0.184 40 5 * 0.148 97 6 r' 0.120 110 7 spoofOPTMP141 0.068 10 8 + 0.042 74 9 uack+ 0.042 1 10 spoofCellTMP152 0.037 10 ``` `512DenseRewritesSpark` ``` # Instruction Time(s) Count 1 sp_mapmm 16.587 71 2 sp_write 7.663 8 3 sp_spoofCellTMP262 4.310 10 4 sp_spoofRATMP243 3.250 10 5 sp_spoofRATMP180 2.451 26 6 sp_spoofRATMP210 2.429 10 7 sp_spoofRATMP271 1.700 20 8 sp_map+ 1.654 72 9 sp_spoofRATMP189 1.531 14 10 sp_spoofOPTMP246 1.526 10 ``` Related discusssion: >both R and DML scripts, the implementation of our rand is very different as we need to ensure > that, given a seed, we generate the same data in local and distributed operations. Accordingly, > we derive seeds for each 1K x 1K block from the initial seed. So, we generated the random matrix in the Java test class itself, so that both the scripts consume same random matrices. [Mail archive](https://www.mail-archive.com/dev@systemml.apache.org/msg00765.html)
---------------------------------------------------------------- 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