New subject: AMD 9511, 9512, Intel 8231, 8232 wanted, will tradeNational MM57109

27 Jan 1999

You're absolutely correct . . . I actually got out my old AMD MOS
Microprocessors and Peripherals data book from 1985, and found that the
9511 is a stack oriented  (like the HP calculators, I guess) arithmetic
processor capable of 16 and 32-bit floating point operations, while the
9512 is capable of 32 and 64-bit operations.

The 9511, it seems is capable of many more functions, being somewhat
reminiscent of a "scientific" calculator, while the 9512 is capable of none
of the transcendental functions which the 9511 supports.  Both are
stack-oriented, therefore ideal for RPN implementation.

My steel-trap mind moves closer and closer every day to being a sievel . .
 oh, well . . . it's HELL getting old.

Dick

----------
...
  From: Gary Oliver &lt;go(a)ao.com&gt;
 To: Discussion re-collecting of classic computers
&lt;classiccmp(a)u.washington.edu&gt;
...
  Subject: Re: AMD 9511, 9512, Intel 8231, 8232 wanted,
will tradeNational MM57109
...
  Date: Wednesday, January 27, 1999 12:36 PM

 From my copy of "Floating Point Program
Manual Am9511A/Am9512:"  
 Am9511A Arithmetic Processor

 Distinctive Characteristics
 ---------------------------
 2,3 and 4 MHz operation
 Fixed point 16 bit and 32 bit operations
 Floating point 32 bit operations
 Binary data formats
 Add, Subtract, Multiply and Divide
 Trigonometric and inverse trigonometric functions
 Square roots, logarithms, exponentiation
 Stack-oriented operand storage
 ...

 Am9512 Floating Point Processor

 Distinctive Characteristics
 ---------------------------
 Single (32-bit) and double (64-bit) precision capability
 Add, Subtract, Multiply and Divide
 Compatible with proposed IEEE format
 ...

 The 9511 is very much like a calculator with a 32 bit operand
 stack of up to four items.  Operations work on the top or top
 two items on the stack.  Some operands (such as PUPI "push PI")
 push a new item on the stack.

 At 4Mhz, a FADD takes (worst case) about 92 uSec and an FMUL
 (worst case) at 42 uSeconds.  Wasn't too shabby when software
 took 5 to 10 times as long on a Z-80.  Plus it was possible to
 get some other work done while the 9511 was off crunching a
 number or two.

 We still use the 9511 on an old (very old) legacy product, although
 they will eventually disappear - AMD is certainly not making them
 anymore and there are very few left on distributor shelves or even
 in "part surplus" houses.

 Gary

 At 10:18 AM 1/27/99 -0700, you wrote:
 >The 9511 was the floating point unit, and the 9512 was for 64-bit integer
...
  >arithmetic, I believe.  I've actually got one
of each of these from back in
...
  >the old days.  They were sort of a calculator for
your computer, in that
 >they could be operated faster than the cpu could perform the function .  .
...
  MUCH faster.

Dick

---------- 