15/09/2024 - 21/09/2024

18/09/2024 04:06

Cockpit is a web interface that shows server information (really computer information) about the machine with cockpit installed. It seems to have "come with" alma9 linux. It may be useful.

To activate, I did the following:

systemctl enable --now cockpit.socket
systemctl enable --now cockpit.socket

This opens a webpage on localhost:9090. You have to port forward an ssh connection to do this, for example:

ssh -L 9090:localhost:9090 root@10.163.102.46
ssh -L 9090:localhost:9090 root@10.163.102.46

If you need to login as a root user, you need to edit the file /etc/cockpit/disallowed-users. Simply remove the line with root in the file, save and exit using your favorite text editor (for example vi /etc/cockpit/disallowed-users). Then you can login as root.

19/09/2024 05:07

While making the pcie_readout test frontend, I had an error due to midas writing the the events to the ODB.

[DataSimulator,ERROR] [odb.cxx:559:realloc_data,ERROR] cannot malloc_data(1048576), called from db_set_data1
[DataSimulator,ERROR] [odb.cxx:6999:db_set_data1,ERROR] Cannot reallocate "/Equipment/Data Simulator/Variables/CR00" with new size 1048576 bytes, online database full
[DataSimulator,ERROR] [midas.cxx:17635:cm_write_event_to_odb,ERROR] cannot write bank "CR00" to ODB, db_set_data1() status 310
[DataSimulator,ERROR] [odb.cxx:559:realloc_data,ERROR] cannot malloc_data(1048576), called from db_set_data1
[DataSimulator,ERROR] [odb.cxx:6999:db_set_data1,ERROR] Cannot reallocate "/Equipment/Data Simulator/Variables/CR00" with new size 1048576 bytes, online database full
[DataSimulator,ERROR] [midas.cxx:17635:cm_write_event_to_odb,ERROR] cannot write bank "CR00" to ODB, db_set_data1() status 310

This can be removed by editting the equipment setting RO_ODB and RO_TRANSITIONS from being added to the "Read on" mask:

EQUIPMENT equipment[] = {
    {"Data Simulator",
        {2, 0,
            "SYSTEM",
            EQ_POLLED,
            0,
            "MIDAS",
            TRUE,
            RO_RUNNING, //remove RO_TRANSITIONS, remove RO_ODB,
            1, //poll time
            0,
            0,
            TRUE,
            "", "", "",},
        read_trigger_event
    },

    {""}
};
EQUIPMENT equipment[] = {
    {"Data Simulator",
        {2, 0,
            "SYSTEM",
            EQ_POLLED,
            0,
            "MIDAS",
            TRUE,
            RO_RUNNING, //remove RO_TRANSITIONS, remove RO_ODB,
            1, //poll time
            0,
            0,
            TRUE,
            "", "", "",},
        read_trigger_event
    },

    {""}
};

19/09/2024 05:35

dbdcecac03ce2401dfb2a0110f4c907f.png

This is as fast as I can get the PCIe data readout to work. Basically, it's just reading data from the PCIe (not writing).

19/09/2024 08:07

There is some mystery going on with the transfer rates. Upon updating the operating system (and consequently needing to update the driver), I am now achieving lower data rates.

First, here is a plot with serveral benchmarks overlayed:

1447b61cf9d2aa28a5017368c6b9a231.png

It's a bit hard to see, but the Xilinx tools reads out perform my C++ library reads for some reason. Also, the shapes are very different.

The shapes are more clear in these individual plots

2e689cc6771810c94323f6e747b5aeaf.png

And here are individual test some comparison plots:

1360487a218ec84df8e71f7082a45639.png

The shape of the XDMA tools is more what I expect, however the data rates are significantly lower than what I saw before updating the operating system (and using the older driver) as seen below.

16dec392019296dc0746c0774a7d9214.png

In any event, these plots explain why the midas frontend is capped at around 600 MB/s right now; because that's the fastest reads can be made using my C++ library (for 1 channel, at least).

19/09/2024 14:52

Some more data on the distriubtion because I don't trust the error bars.

3d03221d6fb52dd4dcead2bace82e941.png