gnupic: gpsim Scope window
Subject:
gpsim Scope window
From:
"Scott Dattalo" ####@####.####
Date:
4 Sep 2006 04:52:46 +0100
Message-Id: <60275.71.139.23.118.1157341949.squirrel@ruckus.brouhaha.com>
Scope Window
I've added a scope window to gpsim. Here are a couple of screen
shots:
http://www.dattalo.com/gnupic/scope1.png
http://www.dattalo.com/gnupic/scope2.png
It's fairly functional at the moment, but I expect to add a few more
features. Here's currently what's supported:
- Up to 8 signals can be monitored (there's no reason for a limit on
the number)
- Only I/O pins are currently scopable.
- Time is shown in simulation cycles
- Signal names can be entered in one of two ways:
a) You can click in the signal area on the left side of the window
to edit or add a new signal
b) You can specify the signal name at the command line using an
attribute:
gpsim> scope.ch0 = "porta0"
(note that the quotes are necessary)
The command line version is useful for setting up your simulation
project from within a script.
- Lower case 'z' and upper case 'Z' zoom in and zoom out
- Lower case 'l' (ell) and lower case 'r' pan left and right
- Traces are red with a black background
- High density events (i.e. more than one edge per pixel) are shown in
white.
- Glitches are shown in yellow. (A glitch is a signal that changes
states more than once in one simulation cycle).
- A light gray grid aligns with the major tick marks.
- The colors are currently hardcoded.
Here's what's not supported but will be added later:
- Support for an arbitrary number of traces
- Support for multi-bit objects (e.g. registers).
- Support for user-define colors
- Support for signal overlays
- Support for analog signals
- Support for infinite horizontal scrolling (where infinite means
from the start to the stop of simulation).
- Support for markers.
- Provide a mechanism for writing the trace information to an LXT
file.
Here's a little background on the implementation: The scope window is
integrated into gpsim (as opposed to being an external module that is
loaded). Once a signal has been selected to be viewed, every event
will be captured (in other words, there is no concept of "sample rate"
like in a real scope). However, the gui only shows a small portion of
the entire waveform. The scroll bar at the bottom of the window
scrolls through this small portion of time. The panning functions get
around this scrolling limitation.
The event buffer that stores the event states is currently a circular
buffer. I haven't tested what happens when it wraps around, but it
probably ain't pretty. However, I intend to make the buffer a linear
one that can grow indefinitely. A binary algorithm searches for event
times while plotting. The execution time for this is dictated by the
number of pixels as opposed to the number events. In other words, if
you're zoomed all the way out and viewing a bazillion event states,
the time to render the screen will be nearly as fast as the case when
there are only a few events.
----
If you have any questions or comments, please feel free to ask or
suggest (or complain!).
Scott
