Errata for year 2000 (s1) notes:

2.3.1 Fixed Point Numbers (excess-n)

The text says to reverse the sign of a 2's complement number to give an excess-n representation. It should be added that this leads to an excess-128 representation only for 8-bit numbers (and in general to excess-2ⁿ format for n-bit binary numbers).

2.3.4 Division

This concerns the division flowchart. There was an initial condition missing, and the condition for A < 0 has now changed.
Please click here for correct information.

2.3.6 Floating Point

The example given for a denormalized number shows an exponent with value 1 when in fact a denormalized number is defined as a number with exponent bitpattern equal to zero (and a non-zero mantissa). Other examples given in the lecture are correct, and should be referred to instead.

2.3.6 Floating Point

The size of an IEEE754 extended double precision number should, of course be 79 and not 64 (as written in the lecture notes). You can find this if you add up the bit numbers given, i.e. 1 bit for the sign, 15 bits for the exponent and 63 bits for the mantissa.

4.4.2 Testing - self-test

For the list of things to check, it should be 'write-then-read'. This is to see if there is any damage to the memory (it's no good to write after you have read....). You need to write a known value to each memory location (maybe a count value), and then scan through memory reading back the count. Your PC probably does this every time it boots up.

Division Flowchart example

For step with count=3, the Q-M calculation is incorrect. The result should be 1001000 and not 10010000. the answer is unaffected. The details on the webpage have been corrected, however, please don't forget to update and copies you printed previously.