No announcement yet.

cpu failure , program to test?

  • Filter
  • Time
  • Show
Clear All
new posts

  • cpu failure , program to test?

    Hello everybody,
    Now that we are in the heat of july, it reminds me of some problems i have had with processors
    over heating and causing damage to data files.

    A few years back i had two computers where the cpu over heated and caused damaage to
    some of my random access files. Both cpu's fans slowed and i was lucky in that it showed its
    face when a random access files were retrieving records, the records were not the same as what
    the records asked for using the record pointer during a file read statement.
    By the way, i could not tell looking at the cpu fan on one machine, that it was not turning the speed
    that it should be turning.

    Well in short i was wondering if anybody has worked on a routine to check the cpu for accuracy.
    I believe this would go well in a screen saver, and just about anywhere else if your data is important to you.

    I know i could do some simple math and check the outcome, but is there better way.
    Maybe the formula would have to be complicated?
    Maybe i need to use the math coprocessor, or maybe not use it?
    Maybe i need to do some other instructions that do some kind of function?

    I feel this kind of code would be helpful, now that i am using pb/dos and have room for code that
    i did not have before.

    I did not list the code, i could, but i do not know what purpose it would sever unless somebody wanted
    to know exactly what happened, but the code did have some interger multiplication in it.
    I was able to duplicate the problem when it occurred by simple pulling the cpu fan power source off while
    my test program would read index files to a random access file and check the variables in the returned record to
    see if key matched what was in the index.

    I did not experience screen problems, everything seemed normal except the records were not the ones requested.
    Of course this was during the pre overheating, when only on one computer did i have total computer failure.

    thanks in advance

    [This message has been edited by paul d purvis (edited July 15, 2003).]
    p purvis

  • #2

    I think if a CPU starts failing, it's too late to detect by a code the cpu runs itself. Why should the fault be on the ALU? If the fail is in the program counter, or in the cache, or wherever else, the CPU will just stop, or cause a GPF, or ... This is to say, I think a method the cpu self-detects a failure can't be reliable; it can work on some instances, but not always.


    [This message has been edited by Aldo Cavini (edited July 16, 2003).]
    Rgds, Aldo