Age | Commit message (Collapse) | Author | Files | Lines |
|
I have a hard time typing anything else at this point.
Signed-off-by: Eric Anholt <eric@anholt.net>
|
|
Trying to make all the callchain related to these const is too hard, and
it's not like marking things const helps either the developer or the
compiler.
Signed-off-by: Eric Anholt <eric@anholt.net>
Acked-by: Keith Packard <keithp@keithp.com>
|
|
Removed redundant declarations. Removed an unused variable.
Fixed the sign of some loop variables.
Signed-off-by: Matt Craighead <mcraighead@nvidia.com>
Signed-off-by: Robert Morell <rmorell@nvidia.com>
Reviewed-by: Aaron Plattner <aplattner@nvidia.com>
Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
|
|
RootWindow(dpy, 0) always uses screen 0, even if a different screen is
specified in the $DISPLAY environment variable. Use
DefaultRootWindow(dpy) instead to use the right one.
Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
Reviewed-by: Jeremy Huddleston <jeremyhu@apple.com>
|
|
Sample the entire result with a single call to GetImage in order to
batch up the transfers when checking the result.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
|
|
Separate the evalation of two colors given a format and the printing of
the result. This allows eval_diff() to be used elsewhere and also means
that we do not need to waste cycles generating unused strings for
eval_diff() [this accounts for as much time as composite_op!]. As part
of this process pass in the expected accuracy of the operation to
eval_diff().
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
|
|
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
|
|
This significantly improves test performance (~60% in blend).
|
|
rendering results to the window, which reduces the runtime of the blend tests by
10.2% +/- 6.2% on my system. The copy remains on by default because I think it
can be useful.
|
|
render acceleration. Thanks to frederikh for the testcase that suggested the
necessity here.
|