To be honest it's not a great test, the latency on each (essential) round trip is going to kill the throughput measured in this way.
My point is this: If you increase the size of the messages the throughput will increase as less overall time (sender has to wait for less replies) spent waiting for reply before sending next message. If you reduce the size of the message the throughput will drop as more waiting/latency. On Thu, Sep 6, 2018 at 7:14 PM Jasper Jaspers <jaspers01...@gmail.com> wrote: > Sorry I had a typo in my reported throughput. When I run the request > reply application I'm getting a throughput of ~ 3.7Gbit /sec. Would still > like to get this closer to 10Gbit/sec if possible. > > > On Wed, Sep 5, 2018 at 2:37 PM Marcus Diogo <mvdiog...@gmail.com> wrote: > >> Sorry i cant follow you. >> 50000000000 total size >> 50000000000÷1024 = 48828125kbytes >> 48828125÷1024=47683,715820313mbytes >> 47683,715820313÷1024=46,566128731gbyte >> 10gbits/s /8 = 1,5gbytes/s >> 46,566128731gbyte÷1,5gbyte/s=31,044085821 seconds >> I guest some thing is not ok but should be like that i think. >> >> >> >> >> >> Em qua, 5 de set de 2018 às 14:59, Jasper Jaspers <jaspers01...@gmail.com> >> escreveu: >> >>> In my setup I have two nodes connected via a 10 gigabit switch. I ran >>> iperf and zeroMQ throughput (local_thr and remote_thr) to prove that I >>> could achieve full bandwidth and I'm able to: >>> iperf: >>> [ ID] Interval Transfer Bandwidth >>> [ 3] 0.0-10.0 sec 11.5 GBytes 9.90 Gbits/sec >>> >>> zeroMQ: >>> message size: 500000 [B] >>> message count: 10000 >>> mean throughput: 2474 [msg/s] >>> mean throughput: 9896.000 [Mb/s] >>> >>> I then wrote a simple zeroMQ request reply test application to measure >>> the "throughput" using REQ/REP socket. In this application the client runs >>> on one node and sends N (10000) messages each of size M (500000) and waits >>> for the reply for each message. The server running on another node >>> receives the messages and sends a simple acknowledgment reply message. The >>> client calculates the "throughput" based on how long it took to send all >>> messages and receive all replies. I'm seeing that it takes about 11 seconds >>> to complete which gives a throughput of 0.12 Gbit /sec. Are these results >>> expected due to the nature of the request reply sockets? Should I expect >>> to be anywhere near the full 10Gbit bandwidth with this test? >>> >>> -Jasper >>> >>> >>> >>> >>> >>> >>> >>> >>> _______________________________________________ >>> zeromq-dev mailing list >>> zeromq-dev@lists.zeromq.org >>> https://lists.zeromq.org/mailman/listinfo/zeromq-dev >>> >> _______________________________________________ >> zeromq-dev mailing list >> zeromq-dev@lists.zeromq.org >> https://lists.zeromq.org/mailman/listinfo/zeromq-dev >> > _______________________________________________ > zeromq-dev mailing list > zeromq-dev@lists.zeromq.org > https://lists.zeromq.org/mailman/listinfo/zeromq-dev >
_______________________________________________ zeromq-dev mailing list zeromq-dev@lists.zeromq.org https://lists.zeromq.org/mailman/listinfo/zeromq-dev
