[net.micro] Intel 80186 rumors

kds@omsvax.UUCP (08/17/83)

This note is to clear up (hopefully) some nasty rumors that have
been posted to this newsgroup about the 80186.

We (Intel) have NO intention of recalling the 80186.

The 80186 is fully supported by Intel and addresses totally
different customer needs than the 80286.  Hence, there is no
truth to the rumor that Intel is pushing the 80286 to 80186
customers.

The real truth on bugs is that there are a few in the existing
parts.  They have caused no slippage in shipping the part.  They
are very minor, and most users can easily design around them.
For example, probably the most serious bug, the integer divide bug,
causes an incorrect result only when one of the operands is in
memory.  An easy workaround is to just move this operand into
a register before performing the operation.  This bug will be
fixed in a new stepping of the part which should start shipping
in about 3 months.

The integrated DMA controller DOES work.  The only problem is that
the user cannot read the upper 4 bits of the pointer registers.
These registers can be written, and after they are written, any
DMA transfers that occur will access the proper 20 bit address.  This
bug will also be fixed in a future stepping of the part.

Intel has a field support team of application engineers scattered
throughout the world.  They have the lowdown on all the known
existing bugs in the 80186.  If you need additional information
concerning what the part does (or doesn't do), please contact
them directly.
				Ken Shoemaker
				Microprocessor Applications
				Intel Corp.
				Santa Clara, California
				...pur-ee!icalqa!omsvax!kds
				...hplabs!icalqa!omsvax!kds
				...teklabs!ogcvax!omsvax!kds

thomson@utcsrgv.UUCP (Brian Thomson) (08/17/83)

A little gem from omsvax ...

    "For example, probably the most serious bug, the integer divide bug,
    causes an incorrect result only when one of the operands is in
    memory.  An easy workaround is to just move this operand into
    a register before performing the operation."

Hey, of course.  All my compilers have options to do just that.
Don't yours? (hee-yuk)
-- 
			Brian Thomson,	    CSRG Univ. of Toronto
			{linus,ihnp4,uw-beaver,floyd,utzoo}!utcsrgv!thomson

reg@vaxine.UUCP (Rick Genter) (08/18/83)

Sure.  Every compiler I ever worked with had a switch like

	-PUTMEMORYOPERANDSINREGISTERSWHENDOINGINTEGERDIVIDE

P.S. Real programers don't divide; they manipulate their data so they can
     use shifts!

					R. Genter
					Automatix Inc.
					...!allegra!linus!vaxine!reg