gtkwave-1.3.71, fixes gcc -O2 floating point rounding issues

For those who package it for various OSs, the new source tarball can be
found here at http://home.nc.rr.com/gtkwave/ ...I didn't upload the
latest windows binary as I'm still out of country and nowhere near my
windows box that has mingw installed on it.

I recently had to use the viewer to debug some VCDs from a customer sim
job and noticed that for long sim runs with small timescales like
femtoseconds, there are problems with the primary and lettered cursors
not lining exactly up the way they should at very close zooms as
they're off by a pixel.

Apparently gcc when compiling with -O2 on x86 does some interesting
reordering of floating point ops--or the internal representation is
more precise than when spilling floating point regs to main memory and
bringing them back in--which is causing rounding differences when I
precalculate some useful scaling factors...I fixed this and some other
visual thrashing with the widgets in the top part of the display
wobbling all over the place when using GTK+2.x. If the viewer has been
giving you some issues with weird visuals, give it a shot.

-t

 

[Jason Zheng]

Tony,

The new flash screen is a nice art work, but seeing it everytime I start
gtkwave is getting a little annoying. Is there a way to prevent the logo
screen from showing?

-jz

bybell@rocketmail.com wrote:

For those who package it for various OSs, the new source tarball can be
found here at http://home.nc.rr.com/gtkwave/ ...I didn't upload the
latest windows binary as I'm still out of country and nowhere near my
windows box that has mingw installed on it.

I recently had to use the viewer to debug some VCDs from a customer sim
job and noticed that for long sim runs with small timescales like
femtoseconds, there are problems with the primary and lettered cursors
not lining exactly up the way they should at very close zooms as
they're off by a pixel.

Apparently gcc when compiling with -O2 on x86 does some interesting
reordering of floating point ops--or the internal representation is
more precise than when spilling floating point regs to main memory and
bringing them back in--which is causing rounding differences when I
precalculate some useful scaling factors...I fixed this and some other
visual thrashing with the widgets in the top part of the display
wobbling all over the place when using GTK+2.x. If the viewer has been
giving you some issues with weird visuals, give it a shot.

-t

 

[Jason Zheng]

Sorry, I should have looked the sample gtkwaverc file before asking. The
option is "splash_disable on"

-jz

Jason Zheng wrote:

Tony,

The new flash screen is a nice art work, but seeing it everytime I start
gtkwave is getting a little annoying. Is there a way to prevent the logo
screen from showing?

-jz

bybell@rocketmail.com wrote:

For those who package it for various OSs, the new source tarball can be
found here at http://home.nc.rr.com/gtkwave/ ...I didn't upload the
latest windows binary as I'm still out of country and nowhere near my
windows box that has mingw installed on it.

I recently had to use the viewer to debug some VCDs from a customer sim
job and noticed that for long sim runs with small timescales like
femtoseconds, there are problems with the primary and lettered cursors
not lining exactly up the way they should at very close zooms as
they're off by a pixel.

Apparently gcc when compiling with -O2 on x86 does some interesting
reordering of floating point ops--or the internal representation is
more precise than when spilling floating point regs to main memory and
bringing them back in--which is causing rounding differences when I
precalculate some useful scaling factors...I fixed this and some other
visual thrashing with the widgets in the top part of the display
wobbling all over the place when using GTK+2.x. If the viewer has been
giving you some issues with weird visuals, give it a shot.

-t