0
101 Sep 12, 2006 at 19:34

Hi gus, i need help in answering the below question:

A multicomputer with 256 CPUs is organized as 16*16 grid. What is the worst-case delay (in hops) that a message might have to take?
Now consider a 256-CUP hypercube. What is the worst-case delay here, again in hops?

#### 7 Replies

0
165 Sep 12, 2006 at 20:00

0
101 Sep 13, 2006 at 13:42

And it’s not even dificult to answer…

Imagine the 2D case. Google for “Manhattan distance”. Do the same in 3D. Apply the general formula you discovered to the 4D case. Done.

0
101 Sep 13, 2006 at 14:49

Well that is not my homework question. I got this issue from 1 of the reference book and i do not know how to answer. That’s why search for help. And also, i thought we can shared anything if we find any difficulty or interesting in computer related factor issue. It seem like i have misunderstanding.

0
101 Sep 13, 2006 at 16:30

It does not seem you find it interesting or difficult. It just seems like you want the answer without thinking a bit for yourself.

Did you google for manhattan distance?

0
101 Sep 14, 2006 at 03:25

Well that is not my homework question. I got this issue from 1 of the reference book and i do not know how to answer. That’s why search for help. And also, i thought we can shared anything if we find any difficulty or interesting in computer related factor issue. It seem like i have misunderstanding.

The problem may have been just the way you posed the question. You should have described what you did to solve the problem and what you couldn’t figure out yourself. That would have looked less like you just want us to solve your homework.

0
101 Sep 14, 2006 at 08:42

Draw a 16 x 16 grid. Count the number of hops from one corner to the next??
I dont even need to draw it, its pretty simple the longest distance is 15 across and 15 down, ie 30 hops.

This is so simple, it would appear you havent even tried.

0
101 Sep 15, 2006 at 16:26

i not even know it can be google for manhattan distance until you have notified. However, i have figured out the issue even for hypercude topology. Thanks anyway.