About Multi threaded Benchmark Example，the test you showed is very different from our test's result

[smallwhitehh]

The test result as follows:
except that we change the ip address from 2.2.2.3 to 10.0.0.4,there is nothing change In parameter list.
./runJBenchmarkTest.py -c -a 10.0.0.4 -i 1024 -o 0
PYTHONPATH is /usr/bin/python
LD library is: /export/home/rdma2014/JXIO/examples
Compiling JAVA files....
------ Running Client Test Application -----
running client task number 1
gathering results from tasks
processing results
average_TPS = 400084.28, average_RX_BW = 390.7073046875 MB, average_TX_BW = 0.0 MB, in_msg_size = 1024 Bytes, out_msg_size = 0 Bytes

Maybe it is because cluster server is different.

But when I sends 100MB payload requests to the server ，average_RX_BW = 11372.419921875 MB，it is obviously wrong.

Can you explain why? thank you.

[katyakats]

Hi,
Can you send me the command line and the output for 100MB?
I tried to reproduce using

1. ./runJBenchmarkTest.py -s -a 36.0.0.116 -i 104857600 -o 0. I found a bug and I am working on a fix. I saw the following output:
   ------ Running Server Test Application -----
   2014-10-26 11:29:06,022 main WARN transport/rdma/xio_rdma_verbs.c:375:xio_alloc() Contig pages allocation failed. (errno=12 Cannot allocate memory)
   2014-10-26 11:29:06,022 main ERROR transport/rdma/xio_rdma_verbs.c:381:xio_alloc() memalign failed. sz:18446744072778416128
   2014-10-26 11:29:06,022 main WARN MsgPool[0x7f1d1825c0b0]:43:MsgPool() there was an error while allocating & registering memory via huge pages
   2014-10-26 11:29:06,022 main WARN MsgPool[0x7f1d1825c0b0]:44:MsgPool() You should work with Mellanox OFED 2.0 or newer
   2014-10-26 11:29:06,022 main WARN MsgPool[0x7f1d1825c0b0]:45:MsgPool() attempting to allocate&registering memory. THIS COULD HURT PERFORMANCE!!!!!
   2014-10-26 11:29:06,022 main ERROR Bridge.cc:415:Java_com_mellanox_jxio_impl_Bridge_createMsgPoolNative() failure on new MsgPool(msg_num+114, in_size=104857600, out_size=0)
   2014-10-26 11:29:06,023 main FATAL MsgPool:81 there was an error creating the MsgPool
   2014-10-26 11:29:06,025 main ERROR EventQueueHandler:535 jxio.EQH[7f1d1824ffc0]: msgPool provided is null or id is wrong. Can not bind
   Starting server portal workers
   Server worker number 1 is up and waiting for requests
2. ./runJBenchmarkTest.py -a 36.0.0.120 -i 100MB -o 0. This is a wrong way to configure the command and as expected I saw the following error which results from wrong parsing:
   ------ Running Server Test Application -----
   Exception in thread "main" java.lang.NumberFormatException: For input string: "100MB"
   at java.lang.NumberFormatException.forInputString(NumberFormatException.java:65)
   at java.lang.Integer.parseInt(Integer.java:492)
   at java.lang.Integer.parseInt(Integer.java:527)
   at com.mellanox.jxio.tests.benchmarks.DataPathTest.parseCommandArgs(DataPathTest.java:81)
   at com.mellanox.jxio.tests.benchmarks.DataPathTest.(DataPathTest.java:51)
   at com.mellanox.jxio.tests.benchmarks.DataPathTestServer.(DataPathTestServer.java:47)
   at com.mellanox.jxio.tests.benchmarks.DataPathTestServer.main(DataPathTestServer.java:103)
   Thanks,
   Katya

[yuyang733]

I also have encountered this situation.

[katyakats]

Hi CrazyYu!
Thank you for your comment. Do you see lower performance or are you unable to run the test at all?
Katya