[comp.sys.amiga] 68881 and the PA was:

billc@percival.UUCP (William J. Coldwell) (12/31/88)

In article <13012@cup.portal.com> Doug_B_Erdely@cup.portal.com writes:
>Ok guys.. I have a CMI PA. If I get the math chip and use 1.3.. Will ANY AND
>ALL programs math routines be speeded up, or does a program have to be written
>just for a math chip??? In other words... HOW transparent is this to the
>System??
>
>          - Doug -
>
> Doug_B_Erdely@Portal.Cup.Com

Doug (and everyone else interested):

  The math chip will only speed up IEEE math functions, not FFP, or regular
  integer/transcendental math functions.

   What it will speed up: Programs that are compiled to use IEEE by the IEEE
   libraries (provided on the WB1.3 enhancer or the PA disk).

 What this means to you is that just about all of the current software out
 does not support using the math chip.  Software developers will start using
 the math chip (if they feel like it - you know how "THEY" are) now that 1.3
 has been released.  The simple part of it, is that all that needs to be
 done is to recompile the software to use IEEE (the +fi option in Manx - 
 sorry , don't use Lettuce so I don't know what is needed there).

  It iss a good bet that anything that uses math intensive functions will
  eventually be recompiled for IEEE.  Don't give up - I have one (but I
  developed the software, so I guess that's not fair), and I know of a few
  projects being worked on that support the math chip.


   Need more info? Email me or CMI Tech Support at:

    ...tektronix!sequent!blowpig!lillian (CMI Tech Support)
    or me (at the address below).

     Bill

-- 
               William J. Coldwell - Amiga Attitude Adjuster
  CRYOGENIC SOFTWARE - 3647 W 97th #21B - Denver, CO 80030 - (303) 465-1330
 "We the unwilling, led by the unknowing,   ...tektronix!reed!percival!billc
  are doing the impossible, for the ungrateful."       \sequent!blowpig/

Doug_B_Erdely@cup.portal.com (01/03/89)

Hay Bill....
  I am glad you responded to that message. I have a problem that I just
narrowed down (for sure) to the Processor Accelerator. If I get a Guru, the
system goes down (the power light flashes) but it NEVER comes back up!! The
power LED just keeps blinking!! If I hit Control-Amiga-Amiga it will usually
come up! Is there anything we can do??? I tried to E-mail you, but the mailer
said that your site was unknown. So what do we do?? HELP!!!!!!!

          - Doug -

 Doug_B_Erdely@Portal.Cup.Com

DMasterson@cup.portal.com (David Scott Masterson) (01/03/89)

In article <1441@percival.UUCP>, billc@percival.UUCP writes:
>In article <13012@cup.portal.com> Doug_B_Erdely@cup.portal.com writes:
>>Ok guys.. I have a CMI PA. If I get the math chip and use 1.3.. Will ANY AND
>>ALL programs math routines be speeded up, or does a program have to be written
>>just for a math chip??? In other words... HOW transparent is this to the
>>System??
>>
>> Doug_B_Erdely@Portal.Cup.Com
>
>Doug (and everyone else interested):
>
>  The math chip will only speed up IEEE math functions, not FFP, or regular
>  integer/transcendental math functions.
>
>   What it will speed up: Programs that are compiled to use IEEE by the IEEE
>   libraries (provided on the WB1.3 enhancer or the PA disk).
>
>               William J. Coldwell - Amiga Attitude Adjuster
>  CRYOGENIC SOFTWARE - 3647 W 97th #21B - Denver, CO 80030 - (303) 465-1330
> "We the unwilling, led by the unknowing,   ...tektronix!reed!percival!billc
>  are doing the impossible, for the ungrateful."       \sequent!blowpig/

Curiousity question:

Can software be compiled in such a way as to leave it up to the consumer which
math library (or any libraries with competing types) is used in a generic
sort of fashion?  For instance, could it be set up such that an ASSIGN could
specify which library to use?  Would the program necessarily become more
memory expensive or could the expense be hidden (well...) in another library
that is loaded at run-time?

Why would software developers want to do this?  Some programs might run best
usinginteger arithmetic in the average system, but a math coprocessor might
significantly improve the average if it was available.  The user would
probably best know the set up in some of these cases.

David Masterson
DMasterson@cup.portal.com

billc@percival.UUCP (William J. Coldwell) (01/04/89)

In article <13145@cup.portal.com> DMasterson@cup.portal.com (David Scott Masterson) writes:
[Stuff deleted concerning PA and MathChip]
>Can software be compiled in such a way as to leave it up to the consumer which
>math library (or any libraries with competing types) is used in a generic
>sort of fashion?  For instance, could it be set up such that an ASSIGN could
>specify which library to use?  Would the program necessarily become more
>memory expensive or could the expense be hidden (well...) in another library
>that is loaded at run-time?
No.
>Why would software developers want to do this?  Some programs might run best
>usinginteger arithmetic in the average system, but a math coprocessor might
>significantly improve the average if it was available.  The user would
>probably best know the set up in some of these cases.
It has nothing to do with the software developers, but with compilers.
Most software developers will figure out the best way that the software
will run, and that usually means a standard machine.  If they wish, they
can put a version of the program on the disk that will run on an MC system.
Since most people do not have mathchips, most developers will only put a
(n) FFP version on the disk - even if a 68881 will speed things up.  On
the other hand, people won't buy MCs if there aren't any programs out
there that use it. Catch 22.

Hopefully the more PAs that are sold, and the more MCs bought will make
developers come out with IEEE versions.  As of now, the only 68881 versions
of programs are for a 68020/68881 combination which won't work with the
peripheral MC on the PA (because on the inline F-Code functions for the
68020).

Bill

>David Masterson
>DMasterson@cup.portal.com


-- 
               William J. Coldwell - Amiga Attitude Adjuster
  CRYOGENIC SOFTWARE - 3647 W 97th #21B - Denver, CO 80030 - (303) 465-1330
 "We the unwilling, led by the unknowing,   ...tektronix!reed!percival!billc
  are doing the impossible, for the ungrateful."       \sequent!blowpig/

rg20+@andrew.cmu.edu (Rick Francis Golembiewski) (01/05/89)

>Curiousity question:

>Can software be compiled in such a way as to leave it up to the consumer which
>math library (or any libraries with competing types) is used in a generic
>sort of fashion?  For instance, could it be set up such that an ASSIGN could
>specify which library to use?  Would the program necessarily become more
>memory expensive or could the expense be hidden (well...) in another library
>that is loaded at run-time?

Yes, they could, in fact I've already done this myself: Using the
function definitions that came with the StarBoard Multi function
module and Lattice C (NOT using the Lattice MathIeee.lib), I was able
to open the mathIEE.library, if it existed, or use the standard Lattice
Math (FFP I believe) or open the old (FFP) mathlib.library.  However,
I had to have a small bit of redundancy in the math portions (ie I
had a statement if (mathieeedoubbas != NULL) IEEEMATH(); else
FFPMATH(); ).   Kind of kludgy, but I must admit that the speedup on
Trans. Functions was worth the effort (plus by using preprossor stuff
[ ie #ifdef MATHLIB... ] I could easily compile a version with out the Math 
co-processor code).  So, it is possible, although I really doubt that
any of the software developers will bother to put it in their code,
just because there isn't a whole lot of people with '881s and a lot
of the programs that are math intensive use lots of basic (+,-,/,*)
operations where the speed of IEEE is about the same with 881 as FFP
is in software.  Still it would be nice to make use of that
co-processor...

+----------------------------------------------------------------------------+
| Disclaimer: Me?  Post That, impossible I never post anything...            |
| TypetoYouLater(Everyone); --> "functional Good bye"....                    |
| Rick Golembiewski [ Pronunciation is half the Battle, spelling the other]  |
+----------------------------------------------------------------------------+

murphy@pur-phy (William J. Murphy) (01/05/89)

This may be a stupid question, but here goes.  I looked at AmigaWorld yesterday
and saw their comments about the PA.  It is a speedup board which has a 16 MHz
68000 and a slot for the 68881.  I think they had it listed for ~$199 ? to 
buy the 68881 is what another 150 bucks or so? (I don't have a price list in 
front of me.  So supposing I spend 350 for the PA with the 881, will that same 
68881 be usable when the A2620 board arrives?

Will the 16MHz 68881 be able to be removed from the PA socket and inserted in 
the A2620?  What will the clock speed be on the A2620?  What is the projected
price tag on the 2620 (with 0Meg)? What is the Mail Order price tag for the PA?
(Note I didn't buy AmigaWorld or I wouldn't ask the last question.)

dale@boing.UUCP (Dale Luck) (01/05/89)

In article <1443@percival.UUCP> billc@percival.UUCP (William J. Coldwell) writes:
>In article <13145@cup.portal.com> DMasterson@cup.portal.com (David Scott Masterson) writes:
>[Stuff deleted concerning PA and MathChip]
>>Can software be compiled in such a way as to leave it up to the consumer which
>>math library (or any libraries with competing types) is used in a generic
>>sort of fashion?
>No.
More information is helpful in this case. The format of the floating point
numbers is different. Compilers will generate the floating point constants
in the format that it needs. This makes it difficult to make the decision
at runtime unless you have the constants compiled once for each format
as well as stash them away in a structure so you can easily pass pointers
around.
   The other way to make it retargetable in terms of floating point is
to isolate all your floating point in a couple of files that are compiled
multiple times and you use a table of functions to pick the appropriate
precision and speed.

>Since most people do not have mathchips, most developers will only put a
>(n) FFP version on the disk - even if a 68881 will speed things up.  On
>the other hand, people won't buy MCs if there aren't any programs out
>there that use it. Catch 22.

The new single precision IEEE libraries that CA is working on should help
solve this dilemma. They are currently in alpha test, support both style
access to the 68881(like the 1.3 dbl precision library) as well as the
goal of the software emulation is to be as fast as ffp. The single precision
ieee format also has an exponent range twice as big as ffp.

So maybe we can have our cake and eat it too.

>>David Masterson
>-- 
>               William J. Coldwell - Amiga Attitude Adjuster

-- 
Dale Luck     GfxBase/Boing, Inc.
{uunet!cbmvax|pyramid}!amiga!boing!dale

UH2@PSUVM.BITNET (Lee Sailer) (01/06/89)

In article <1441@percival.UUCP>, billc@percival.UUCP (William J. Coldwell) says:
>
>
>  The math chip will only speed up IEEE math functions, not FFP, or regular
>  integer/transcendental math functions.
>

Another posting suggested that for elementary operations, +-*/, FFP and
IEEE/68881 are about the same, but that for trig functions, the
68881 is about 30 times faster.

Would it be possible to write a NEW mathffp.library that was really a sort
of ieeemath.library (or whatever its name is) in disguise.  That way, if
you have a program that opens the ffp library, uses a lot of trig, and
is slow, you could substitute the new ffp library, and be fast.

Usual disclaimer.  I dunno nuthin'

billsey@agora.UUCP (Bill Seymour) (01/07/89)

From article <13139@cup.portal.com:, by Doug_B_Erdely@cup.portal.com:
: Hay Bill....
:   I am glad you responded to that message. I have a problem that I just
: narrowed down (for sure) to the Processor Accelerator. If I get a Guru, the
: system goes down (the power light flashes) but it NEVER comes back up!! The
: power LED just keeps blinking!! If I hit Control-Amiga-Amiga it will usually
: come up! Is there anything we can do??? 

	Hmmm, first of all I'm going to need some more info. Which Amiga do
you have? Do you have any expansion devices plugged in? What kinds are they?
How old is your Accelerator? I've seen this symptom on a couple of older
1000s, but it's been so sporadic that I haven't really tried to pin it down.
(ie, mine does it once a week or so... I have an early 1000 with *four* 
expansion devices plugged into the side...) If all else fails and you want
to get a solution in more of a hurry, give us a call at (503)684-9300.
Lilliane will probably be the one you'll talk to, she's doing the technical 
support. (Are you listening, Lilliane?)

: I tried to E-mail you, but the mailer
: said that your site was unknown. So what do we do?? HELP!!!!!!!

	You might try stopping the path at sequent. I'm not that much of
a UNIX buff, but I think you can get away with something like;

blowpig!billsey@sequent
	or
...tektronix!sequent!blowpig!billsey
 should get you there.

: 
:           - Doug -
: 
:  Doug_B_Erdely@Portal.Cup.Com
-- 
     -Bill Seymour             ...tektronix!reed!percival!agora!billsey
                               ...tektronix!sequent!blowpig!billsey
     Creative Microsystems   Northwest Amiga Group    At Home Sometimes
     (503) 684-9300          (503) 656-7393 BBS       (503) 640-0842

billsey@agora.UUCP (Bill Seymour) (01/07/89)

From article <1443@percival.UUCP:, by billc@percival.UUCP (William J. Coldwell):
: 
: Hopefully the more PAs that are sold, and the more MCs bought will make
: developers come out with IEEE versions.  As of now, the only 68881 versions
: of programs are for a 68020/68881 combination which won't work with the
: peripheral MC on the PA (because on the inline F-Code functions for the
: 68020).

	Not quite true, there are a couple of spreadsheets out there that use
the math chip via the IEEE libraries. Maxi-Plan and it's newest incarnation,
Plan/IT are two that I know of. Not to mention that great shareware game,
Moria 2.0!

: 
: Bill
: 
: -- 
:                William J. Coldwell - Amiga Attitude Adjuster
:   CRYOGENIC SOFTWARE - 3647 W 97th #21B - Denver, CO 80030 - (303) 465-1330
:  "We the unwilling, led by the unknowing,   ...tektronix!reed!percival!billc
:   are doing the impossible, for the ungrateful."       \sequent!blowpig/
-- 
     -Bill Seymour             ...tektronix!reed!percival!agora!billsey
                               ...tektronix!sequent!blowpig!billsey
     Creative Microsystems   Northwest Amiga Group    At Home Sometimes
     (503) 684-9300          (503) 656-7393 BBS       (503) 640-0842

billsey@agora.UUCP (Bill Seymour) (01/07/89)

From article <1765@pur-phy:, by murphy@pur-phy (William J. Murphy):
: 
: This may be a stupid question, but here goes.  I looked at AmigaWorld yesterday
: and saw their comments about the PA.  It is a speedup board which has a 16 MHz
: 68000 and a slot for the 68881.  I think they had it listed for ~$199 ? to 
: buy the 68881 is what another 150 bucks or so? (I don't have a price list in 
: front of me.  So supposing I spend 350 for the PA with the 881, will that same 
: 68881 be usable when the A2620 board arrives?

	There's no reason you shouldn't... But doesn't the 2620 come with a
68881 already installed? That makes it a bit of a moot point...

-- 
     -Bill Seymour             ...tektronix!reed!percival!agora!billsey
                               ...tektronix!sequent!blowpig!billsey
     Creative Microsystems   Northwest Amiga Group    At Home Sometimes
     (503) 684-9300          (503) 656-7393 BBS       (503) 640-0842