Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!magnus.acs.ohio-state.edu!csn!seri!wind55!marshall From: marshall@wind55.seri.gov (Marshall L. Buhl Jr.) Newsgroups: comp.sys.ibm.pc.hardware Subject: Re: Qn: 80287 with a 386 Message-ID: Date: 31 May 91 21:03:52 GMT References: Sender: news@seri.gov (news) Distribution: na Organization: Solar Energy Research Institute Lines: 23 ft0d+@andrew.cmu.edu (Feichin Tschang) writes: >Has anyone out there got any comments on using a 80287 math coprocessor with >a 386-25 or similar? I believe my system board accommodates a 287 and they're >about 1/4 the price of 387's, so my question is whether or not the 287 improves >performance within a factor of using a 387 on a 386 system. Back years ago, I bought several of the original Compaq 386's. That was before the invention of the 387. They came with a socket for an 8MHz 287. We later got a Compaq 386 with a 387. We put 12MHz 287 daughter cards from Microway into the 287 sockets (they had their own clock). We found that the 16 MHz 387 was 2.5 times faster than the 12 MHz 287 when running floating point intensive FORTRAN programs. If you're doing a lot of math, you'd be better off spending a few hundred extra on the 387. Marshall -- Marshall L. Buhl, Jr. EMAIL: marshall@seri.gov Senior Computer Engineer VOICE: (303)231-1014 Wind Program 1617 Cole Blvd., Golden, CO 80401-3393 Solar Energy Research Institute Solar - safe energy for a healthy future