Using FP registers as additional GP registers 
Author Message
 Using FP registers as additional GP registers

I was coding a load/store loop of double words and thought,
instead of using:

         LM    R00,R01,source
         STM   R00,R01,target

I could instead use:

         LD    R00,source
         STD   R00,target

Since there is no validation of the contents of these registers for loads and
stores, there can be used for transient storage. It also has an advantage over
LM/STM of allowing index registers, as well as giving you some "extra" registers
to play with.

Does anyone out there use FP registers in this manner?

--
Robert Ngan
CSC Financial Services Group



Mon, 23 Jun 2003 03:19:42 GMT  
 Using FP registers as additional GP registers
| I was coding a load/store loop of double words and thought,
| instead of using:
|
|          LM    R00,R01,source
|          STM   R00,R01,target
|
| I could instead use:
|
|          LD    R00,source
|          STD   R00,target
|
| Since there is no validation of the contents of these registers for loads
and
| stores, there can be used for transient storage. It also has an advantage
over
| LM/STM of allowing index registers, as well as giving you some "extra"
registers
| to play with.
|
| Does anyone out there use FP registers in this manner?

Close.  I used to use the floating point (FP) registers all the time for
tokenization under CMS  (8 characters per token) ---- I used the FP
registers in pairs.  It really speeded up my parsing code (for various
options/parameters).

Gerard S.



Mon, 23 Jun 2003 04:40:18 GMT  
 Using FP registers as additional GP registers
I've seen code that initializes them with 0, and then uses stores for
a 'fast clear' of the appropriate amount of storage.

Don't do that myself tho'.

Guy



Quote:

> | I was coding a load/store loop of double words and thought,
> | instead of using:
> |
> |          LM    R00,R01,source
> |          STM   R00,R01,target
> |
> | I could instead use:
> |
> |          LD    R00,source
> |          STD   R00,target
> |
> | Since there is no validation of the contents of these registers for
loads
> and
> | stores, there can be used for transient storage. It also has an
advantage
> over
> | LM/STM of allowing index registers, as well as giving you
some "extra"
> registers
> | to play with.
> |
> | Does anyone out there use FP registers in this manner?

> Close.  I used to use the floating point (FP) registers all the time
for
> tokenization under CMS  (8 characters per token) ---- I used the FP
> registers in pairs.  It really speeded up my parsing code (for various
> options/parameters).

> Gerard S.

Sent via Deja.com
http://www.deja.com/


Wed, 25 Jun 2003 07:16:45 GMT  
 Using FP registers as additional GP registers
Yep, been doing it since the early 70s.  In some appls, I also put info in
the fp regs indicating addrs and info when a dump is taken.  That way,
you don't have to search thru a dump to find important pointers, etc.
Makes it easy debugging over the phone with a client.

Doug Hofbauer
RJM Systems, Inc.

Quote:

> I was coding a load/store loop of double words and thought,
> instead of using:

>          LM    R00,R01,source
>          STM   R00,R01,target

> I could instead use:

>          LD    R00,source
>          STD   R00,target

> Since there is no validation of the contents of these registers for loads and
> stores, there can be used for transient storage. It also has an advantage over
> LM/STM of allowing index registers, as well as giving you some "extra" registers
> to play with.

> Does anyone out there use FP registers in this manner?

> --
> Robert Ngan
> CSC Financial Services Group



Sun, 29 Jun 2003 08:02:50 GMT  
 Using FP registers as additional GP registers
Great idea! I hadn't thought of this, I'll start using it immediately.


says...

Quote:
> Yep, been doing it since the early 70s.  In some appls, I also put info in
> the fp regs indicating addrs and info when a dump is taken.  That way,
> you don't have to search thru a dump to find important pointers, etc.
> Makes it easy debugging over the phone with a client.

> Doug Hofbauer
> RJM Systems, Inc.


> > I was coding a load/store loop of double words and thought,
> > instead of using:

> >          LM    R00,R01,source
> >          STM   R00,R01,target

> > I could instead use:

> >          LD    R00,source
> >          STD   R00,target

> > Since there is no validation of the contents of these registers for loads and
> > stores, there can be used for transient storage. It also has an advantage over
> > LM/STM of allowing index registers, as well as giving you some "extra" registers
> > to play with.

> > Does anyone out there use FP registers in this manner?

> > --
> > Robert Ngan
> > CSC Financial Services Group

--
Robert Ngan
CSC Financial Services Group


Mon, 30 Jun 2003 10:34:02 GMT  
 
 [ 8 post ] 

 Relevant Pages 

1. Error: Clock skew plus hold time of destination register exceeds register-to-register delay

2. Error: Clock skew plus hold time of destination register exceeds register-to-register delay

3. Status Register/Control Register

4. minimum delay for Register-to-Register path in DC

5. Pairing 2 32Bit registers to make a 64Bit register

6. Are 64 Int or FP registers useful?

7. Why are interger and FP registers separated?

8. Pipelined FP add (Separate Register Sets)

9. Using Regsvr32 to register a inproc dll

10. Capturing Windows messages registered using RegisterWindowsMessage

11. Clearing chart buffer using shift registers

12. Using a shift register to detect input...

 

 
Powered by phpBB® Forum Software