First update

Well, that was fast. Only a few days since I’ve uploaded the 0.8 version of the simulator and here’s an update.

As it turns out, the IOS uses the size of the COS image to determine the location to upload the PARAMETER file during boot. If I get the COS image file size wrong (which is what happened) than IOS loads the PARAMETER file to the wrong location and consequently COS will not find it.

The new, v-0.81 release fixes this problem so now the parameter file settings are effective. Here’s the update on the boot process:

Once you’ve logged in to the Cray Station console (see the previous article on how to get there), type in:


This gives you an ‘interactive’ communication channel with the booting COS. Eventually you should get the following message:


Here, you reply by typing in:


Some time after, you get another message:


This message is telling you that your PARAMETER file works and you will be performing an ‘install’ operation, which is what you want. So, reply again:


The next message you get is this:


And, well, this is where I get stuck now. I’m not sure what’s going wrong, but something obviously does. Curiously, I also get a lot of errors on the IOP3 (XIOP) console:


This is almost certainly because of my next to non-existent BMX channel and device simulation – COS is most likely trying to talk to the IBM tape drives here. But are the two problems correlated? Who knows. I guess I’ll have to square up my BMX simulation before I will.

Till then, the newest release is here:

Print Friendly, PDF & Email

7 thoughts on “First update

    • Thanks!

      I wasn’t aware of this repository of documentation (I’ve been on the CrayWiki but didn’t find this page). I’ll look around!

  1. Hello,
    i posted 10mins ago a comment… but I cannot see it.
    I found a website with some old version of cray manuals (

    In the system programmer manuals, there’s a part regarding the boot process and ZLOG. It says the startup program, called Z, has a log. You can access to this log in this way:


    I tried and I have an output but I don’t know if this could be useful. It should contain an error message, if this is the case… and the error message is in ASCII format.


    • Fausto!

      Thanks for the info! Your comment didn’t appear on the site due to moderation (you can’t believe how much comment spam I get).

      I’m not sure which version did you download, but V-0.81 parses the parameter file correctly and proceeds into the install process (where it bails with an error):

      After this message appears, I can type in what you’ve suggested, and get the following dump:

      This looks like an exchange packet dump, but the S registers don’t seem to contain an error string, which might be because parameter parsing (the ZY routine) succeeded. I’ll keep looking.

      You pointed me to a nice debugging facility, I can keep poking at, so this is useful!

      Thanks again,

      • Hi Andras,

        cool! 🙂
        Yes I have the same error, and I hoped to find something in the S register, but your analisys seems to be correct, so maybe there’s another log to check…

        But if I understood well the INSTALL should format the disk… maybe the program cannot access it in a proper way.

        Just my two cents… 🙂


  2. I wouldn’t be too supprised if it was loading installation software from tape.. after all its not a common task and would probably be quite large so a tape would be ideal for that task.

    • That would be a reasonable assumption, but all I’ve read about the install procedure seems to indicate that no additional SW is loaded from tape. I’ve also gotten the block multiplexer simulation fixed more or less (I should do an update about it), but it didn’t fix the problem. I’m now under the belief that the two errors were unrelated, and just happened at the same time.

Leave a Reply

Your email address will not be published. Required fields are marked *