10/15/04

A student writes:

     Hi Dr. Patt,
     First I'd like to say that I was impressed at how fast
     the tests were graded and returned. 

Thank the TAs. It was a very long team effort, and they deserve more credit than I.

     I hadn't been checking my email lately, but today 
     I went through it and read the one about misgraded problem sets,
     inspiring me to go over my test carefully.

You should always go over your test carefully, regardless whether there was a problem or not. It is an outstanding way to reinforce what you do know and point out what you don't know.

If you only checked the exam carefully because of that email, then I need to thank the student who wrote it!

By the way, the TAs and I grade all exams.

     I came across 3b and I believe I wrote the right answer.

Let's see.

     The question was "What is the largest positive
     normalized number that can be expressed with this 
     8-bit floating point data type? With 8 bits (1 sign bit, 4
     exponent, 3 precision), the largest number would be
     01111111.  Expanded to normalized form, that would be 
     1.111 x 2^(15-9) equaling 1.111 x 2^6.  Not wanted to mix bases 

I would have preferred you mixed bases!

     I rewrote it as 1.111 x 10^(110) and it was counted wrong.
     Have I missed something I'm not seeing?

Yup. In IEEE Floating Point, the largest exponent is used for infinities and other stuff.

     Thanks for your time,
     << name withheld to protect the *************** >>

Yale Patt