Homework, 5.32

[TNKS]

Hi again,

Okay, so it's getting a little later than I would prefer. I've given
5.32 a bit of thought, but I think I'm missing something. The problem
seems to assert that it wants a solution by *only* altering the state
diagram, and not by installing any new datapaths/hardware in the
circuit. lui inherently requires a 16-bit shift. I've stared at the
multi-cycle datapath, and I'm having a really hard time figuring out how
to shoehorn a 16-bit shift using the pre-existing design (by just
altering control and possibly many cycles).

My attempts till now have been to try and use the 2-bit shift to
accumulate a 16-bit shift. I also had the idea to try and use the ALU
for a multiply. However, I haven't had any success yet.

The next problem (5.33) seems to let me mess with the datapath, which is
a lot more freedom, so I think I'm okay there, but 5.32 is frustrating me.

Am I interpretting this problem correctly? If someone has a benign
hint. I'm open to it.

- Sukant


/************************************************
* P.S. - THE E-MAIL ADDRESS I PROVIDE IS BOGUS *
* TO PROTECT ME FROM SPAMMERS. PLEASE *
* JUST POST BACK TO THE GROUP. *
************************************************/

[...]

Post by TNKS
Am I interpretting this problem correctly? If someone has a benign
hint. I'm open to it.

I think so, and ... maybe. The intentions are benign, anyway.

Post by TNKS
My attempts till now have been to try and use the 2-bit shift to
accumulate a 16-bit shift. I also had the idea to try and use the ALU
for a multiply. However, I haven't had any success yet.

I just assumed the ALU was capable of a 16-bit shift, since we haven't
actually seen the specs on the ALU, and should therefore be free to
speculate within reason. Correct me if I'm wrong (about not seeing the
ALU specs - are they in here?), but it's gonnna take a stiff argument to
convince me to change anything this early in the morning.

[Alan Lee]

If you want to use the same datapath, you can use the ALU to do the shift.

Post by ...

Post by TNKS
Am I interpretting this problem correctly? If someone has a benign
hint. I'm open to it.

I think so, and ... maybe. The intentions are benign, anyway.

Post by TNKS
My attempts till now have been to try and use the 2-bit shift to accumulate
a 16-bit shift. I also had the idea to try and use the ALU for a multiply.
However, I haven't had any success yet.

I just assumed the ALU was capable of a 16-bit shift, since we haven't
actually seen the specs on the ALU, and should therefore be free to speculate
within reason. Correct me if I'm wrong (about not seeing the ALU specs - are
they in here?), but it's gonnna take a stiff argument to convince me to
change anything this early in the morning.

[David Grohmann]

haha since the book was allowed to draw nice shift bubbles, I kjust drew
my own 16 bit version.

:-)

______________

David Grohmann

Post by Alan Lee
If you want to use the same datapath, you can use the ALU to do the shift.

Post by ...

Post by TNKS
Am I interpretting this problem correctly? If someone has a benign
hint. I'm open to it.

I think so, and ... maybe. The intentions are benign, anyway.

Post by TNKS
My attempts till now have been to try and use the 2-bit shift to
accumulate a 16-bit shift. I also had the idea to try and use the
ALU for a multiply. However, I haven't had any success yet.

I just assumed the ALU was capable of a 16-bit shift, since we haven't
actually seen the specs on the ALU, and should therefore be free to
speculate within reason. Correct me if I'm wrong (about not seeing the
ALU specs - are they in here?), but it's gonnna take a stiff argument
to convince me to change anything this early in the morning.

[...]

Post by TNKS
/************************************************
* P.S. - THE E-MAIL ADDRESS I PROVIDE IS BOGUS *
* TO PROTECT ME FROM SPAMMERS. PLEASE *
* JUST POST BACK TO THE GROUP. *
************************************************/

BTW, it's a bit late for this. Check out the headers on Al's post,
"Succeeding (surviving?) in CS352".