Path: utzoo!utgpu!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!cwjcc!hal!nic.MR.NET!shamash!nis!ems!amdahl!oliveb!amiga!boing!dale From: dale@boing.UUCP (Dale Luck) Newsgroups: comp.sys.amiga Subject: Re: 1.3's 68881 Problem... Message-ID: <497@boing.UUCP> Date: 13 Nov 88 05:09:12 GMT References: Reply-To: dale@boing.UUCP (Dale Luck) Organization: Boing, Milpitas, Ca. Lines: 62 In article rg20+@andrew.cmu.edu (Rick Francis Golembiewski) writes: < bunch of stuff about released ieee lib not compatible with > < old beta > Are you using a 68020/68881 combination? or a peripheral68881? We need to know what version of the ieee library you are using. There was a version of the ieee libraries that was released that addressed the peripheral style 68881 in only one address. This was before the autoconfig stuff was put in. Maybe you are using a really ancient ieee library. You may need to update your multitasking/autoconfig software from the vendor. If you are using a peripheral 68881 make sure there is somthing in the 'expansion' drawer, or some other suitable startup that creates the public resource that ieee library searches for when it starts up. The ieee library as released works just fine as far as I know. > >Also, along the same lines, does anyone know if it would be possible to have >a replacement math.library that could just pass the calls to the IEE.libraries? >The reason I ask is that (currently) most programs that would benifit the most >from really fast math use the regular Math.library (ie Sculpt 3D, Photon Paint, >etc.). and it would be nice to be able to just have a dummy math.library that >opened up the IEE.library and just passed the calls and returned the value to >a program. Not a trivial task as well as it does not always improve anything. The programs you mention do make use of floating point, however they use the basic functions (+,-,/,*,cmp) much more that anything else. These functions are already quick in the math library 80-200usec. Except for an occasional sqrt of course. To make them use the 881 would require: 1. convert arguments from MFFP (Motorola Fast Floating Point) to IEEE. 2. Call the ieee library routines unless you want to duplicate all the special hardware code. 3. Convert result back to MFFP. So I think that in most cases the code would actually be slower. > >As an aside, does anyone know how the Lucas board works with the 1.3 Math >Libraries? (I'de certainly like to have the '20 using the '881 as a >co-processor, instead of a perif. devce...). Looks like you just answered my question from above. I suggest you contact the vendor of your board. > >Enough Ramblings... >+----------------------------------------------------------------------------+ >| Disclaimer: Me? Post That, impossible I never post anything... | >| TypetoYouLater(Everyone); --> "functional Good bye".... | >| Rick Golembiewski [ Pronunciation is half the Battle, spelling the other] | >+----------------------------------------------------------------------------+ -- Dale Luck Boing, Inc. {uunet!cbmvax|pyramid}!amiga!boing!dale Although I do contract work for Amiga-LosGatos, my opinions probably don't represent those of Commodore or its management or its engineers, but I think the world would be a better place if they did.