atari email archive

a collection of messages sent at Atari from 1983 to 1992.

RECAP: 68010 AMS source-level Debugger

(1 / 1)


Here is the collected opinion on the AMS/Intermetrics Source-level Debugger:

	1) The "general" Tool would be neat to have.
	2) The Intermetrics Implementation (according to those who know)
	   is'nt worth the paper its printed on (need to update that
	   old saying to: Isn't worth the ferric oxide its magnetized on)

First of all, Intermetrics does not have a good reputation here at Atari.
We have used an Intermetrics C cross-compiler here and have
adandoned it due to: 1) MANY bugs which Intermetrics refused to fix; and
2) It produces code that is 20-40% longer and slower than the Greenhill
compiler we currently use. 

Second, Jim Morris has seen and used the Intermetrics Debugger at a show and 
reports that:
	- NOT realtime
	- It patches breakpoints into the code, excutes one instruction, checks
  	  if its the breakpoint you want and so on...it will not exucute
	  code realtime which makes it just about useless in a video game.
	- Doesnt use any of the sophisticated breakpoint features of the AMS.
	- The VAX C debugger would do the same job only better.

Third, Dave Shepperd notes that a "cross-debugger" would indicate that the
software runs on the VAX and sends commands down to the AMS to debug...this
would most certainly bring KIM and CHARM to their knees.

Forth, Mike Albaugh points out that we have a large base of software that
is based on the Greenhill compiler and converting to the Intermetrics would
be painful. Morris adds that Greenhill keeps 32-bit integers while
Intermetrics keeps 16-bit integers.......

In Summary, the Validate Software is not what we need - at any price. Jim
also reports that his contacts at AMS say they are working on an internally
developed source level debugger that truly uses the AMS capabilities which
may suit our needs. Unfortunately, that software is 1-2 years away.
Message 1 of 1

May 21, 1986