|
Authored by: Anonymous on Friday, March 29 2013 @ 09:40 AM EDT |
BCD? Seriously? What system works with BCD (besides some older
mainframes that might work with BCD via COBOL)? No modern
microcomputer does that. You'd need a software library to deal
with such number encoding - and I'm quite sure nothing in the
Linux kernel does so.
You might, as part of input/output convert from/to characters
(ASCII/ANSI, UNICODE, EBCDIC, etc.), but that has nothing to
do with floating point rounding at all.[ Reply to This | Parent | # ]
|
|
Authored by: Anonymous on Friday, March 29 2013 @ 09:43 AM EDT |
Forgot to mention....
BCD is BINARY coded decimal. So changing it to BINARY might
just be a NOP or at the very least is purely a format change.
Also, as best I can remember, BCD is integer (or fixed point),
not floating point.[ Reply to This | Parent | # ]
|
|
Authored by: JamesK on Friday, March 29 2013 @ 01:08 PM EDT |
In my experience, "registers" hold binary data. It is how that data
is implemented that determines whether it's BCD, ASCII or whatever. I have only
worked on one system, that Teleregister at the Toronto Stock Exchange that I
mentioned in another thread, where data was stored in BCD excess 6, which is BCD
with 6 added to it, to make some of the logic simpler. Beyond that, registers
hold binary data and only binary data. How that binary data is interpreted is
up to the programmer.
BTW, in my experience,"registers" are a temporary scratch pad area,
used in the various calculations etc. and separate from the main memory. There
are, however, some systems (TI 9900?) where specific memory locations are
reserved for use as a "register". The Data General computers also
contained a block of auto increment and decrement memory locations, where if you
made an indirect access through them, the contents would be incremented (or
decremented), readying them to point to the next memory access.
---
The following program contains immature subject matter.
Viewer discretion is advised.[ Reply to This | Parent | # ]
|
|
|
|
|