[comp.unix.xenix] '287 will speed up Xenix?

tony@killer.UUCP (Tony Holden) (10/14/87)

I have just heard a statement that if true, I can't believe.

The statement is:  if you have a math co-processor on board then Xenix itself
will speed up by 20-30%.

Is this true?  How come I've never heard this before?

Dumbfoundedly yours

Tony Holden
ihnp4!killer!tony

brian@ncrcan.Toronto.NCR.COM (Brian Onn) (10/16/87)

In article <1819@killer.UUCP> tony@killer.UUCP (Tony Holden) writes:
>I have just heard a statement that if true, I can't believe.
>
>The statement is:  if you have a math co-processor on board then Xenix itself
>will speed up by 20-30%.
>
>Is this true?  How come I've never heard this before?
>
>Dumbfoundedly yours
>
>Tony Holden
>ihnp4!killer!tony

I doubt it.  The addition of a math coprocessor will only speed up those
programs that take advantage of it.

I have never seen a floating point operation being performed in any unix kernel
I have worked on, and I suspect the same holds true for the Xenix kernel.  Thus
expect no speed-up for the kernel.

Very few utilities use floating point, and only one comes to mind on Xenix,
spline.  Thus expect no speed up from the utilities.

The only advantage would be if you were running an application on Xenix that
did extensive floating point operations, and it actually *uses* the
co-processor. 

-- 

 +-------------------+--------------------------------------------------------+
 | Brian Onn         | UUCP:..!{uunet!mnetor, watmath!utai}!lsuc!ncrcan!brian |
 | NCR Canada Ltd.   | INTERNET: brian@ncrcan.Toronto.NCR.COM                 |
 +-------------------+--------------------------------------------------------+

ericg@sco.COM (Mwa ha ha) (10/19/87)

tony@killer.UUCP (Tony Holden) wrote in article <1819@killer.UUCP>:
>
>The statement is:  if you have a math co-processor on board then Xenix itself
>will speed up by 20-30%.
>
>Is this true?  How come I've never heard this before?
>
>Tony Holden
>ihnp4!killer!tony


SCO Xenix has no floating point instructions in 
the kernel.  Therefore, a math coprocessor
has no effect on the Xenix kernel.   Applications
which use floating point arithmetic, will of course
be sped up significantly.


-- 
Eric Griswold  (ericg@sco.COM)
"When 80286s are outlawed, only outlaws will have 80286s"
I do not speak for SCO, I barely have enough room for my own opinions.

davidsen@steinmetz.steinmetz.UUCP (William E. Davidsen Jr) (10/20/87)

In article <332@ncrcan.Toronto.NCR.COM> brian@ncrcan.Toronto.NCR.COM (Brian Onn) writes:
|In article <1819@killer.UUCP> tony@killer.UUCP (Tony Holden) writes:
|>I have just heard a statement that if true, I can't believe.
|>
|>The statement is:  if you have a math co-processor on board then Xenix itself
|>will speed up by 20-30%.

|Very few utilities use floating point, and only one comes to mind on Xenix,
|spline.  Thus expect no speed up from the utilities.
|
|The only advantage would be if you were running an application on Xenix that
|did extensive floating point operations, and it actually *uses* the
|co-processor. 

If you're using nroff/troff much you might want it. We tested nroff on a
VAX and found that it was cost effective to add an FPA. I suspect the
same is (even more) true on a [23]86.
-- 
	bill davidsen		(wedu@ge-crd.arpa)
  {uunet | philabs | seismo}!steinmetz!crdos1!davidsen
"Stupidity, like virtue, is its own reward" -me