The compute renderer is more expensive to run than the old renderer on
low-end GPUs, and even more so on CPUs. To ensure good performance
regardless of the end-user device, this change implements automatic
re-use of content rendered in the frame before the current.
The basic idea is that every drawing operation (PaintOp), along with its
transform and clipping, can be hashed and efficiently looked up. A naïve
caching approach is then to rasterize every operation to separate
sections of several large texture atlases, turning a cache hit into a
very cheap texture copy.
However, for scenes with lots of overlapping operations, the resulting
texture memory from separating the operations would be much larger than
the memory for just the window framebuffer.
So instead of caching individual operations, this change caches layers,
which are sequences of drawing operations. It starts by putting all
operations into a single layer. Then, if the subsequent frame re-uses a
sub-sequence of that larger layer, it is split.
For example, consider a UI similar to the kitchen sample:
Hello, Gio
<Editor>
<Line Editor>
<Button> <Button> <Button>
<ProgressBar>
<Checkbox> <Toggle>
In the first frame, all of the drawing operations comprising the UI will
be stored and cached in a single layer. In the second frame the
progress bar will have moved and the renderer splits the UI into three
layers: layer A for everything up to (but not including) the progress
bar, layer B with just the progress bar, and layer C for the rest. Note
that nothing has been re-used yet. In the third frame, the progress bar
moves again, and this time layer A and C can be copied from the cache
only the progress bar needs redrawing through the compute programs.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
The performance difference is negligible, but is useful when the compute
pipeline can skip rendering to empty tiles.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Until now, the two renderers have shared structures and code for
decoding drawing ops and convert them to GPU-friendly structures.
However, the decoder is tailored to the old renderer and use
structures that poorly fits the new compute renderer.
This change copies the decoder and specializes the copy for the compute
renderer, avoiding a round-trip through the old renderer decoder.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
The compute renderer doesn't run on Windows yet, but the d3d11 backend needs
the method to satisfy the driver interface.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Otherwise, making a context current on another thread may result in
an EGL_BAD_ACCESS error.
Fixes gio#248
Signed-off-by: Elias Naur <mail@eliasnaur.com>
The CPU fallback renderer relies on posix_memalign, which was exposed in
Android 4.2. Support for Android 4.1 could be restored by writing
posix_memalign outselves, but this change is easier.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
The transformation information in ops.Key is a layer violation.
Introduce a key type specific to package gpu and use that instead.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Releasing the renderer is fine, but releasing the underlying context
introduces flicker when restoring a Gio window on macOS.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
This commit ensures that the dimensions returned by material.List
include the size of the scrollbar when the scrollbar is set to
the Occupy AnchorStrategy.
Signed-off-by: Chris Waldon <christopher.waldon.dev@gmail.com>
This change ensures that the scrollbar anchors to the proper edge of the
content even when the list was drawn with a zero minimum constraint.
Without this, the layout.Direction used to anchor the scrollbar will choose
to use the minimum size and anchor it to the opposite edge of the content.
Signed-off-by: Chris Waldon <christopher.waldon.dev@gmail.com>
For some reason, commit d331f63d20 didn't
update the generated code for material.vert properly. The outdated
version is equivalent to the new, so I only discovered this discrepancy
while changing some other shader.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
The scrollbar implementation prior to this change only adjusted
list.Position.Offset. This works in all circumstances except when
list.Position.BeforeEnd=false. If the position indicates that the
scroll position is at the end of the list, the offset is ignored.
This change ensures that manually dragging the scrollbar always
causes BeforeEnd to be set to true. If the drag ends with the
scrollbar at the end of the list, BeforeEnd will be set
automatically by the next list.Layout call, so this doesn't
prevent the list from optimizing for that case in general.
Signed-off-by: Chris Waldon <christopher.waldon.dev@gmail.com>
To use these lists instead of layout.List, callers simply need to
change declarations of layout.List to widget.List, and to change
calls to layout.List.Layout to material.List(th,&list).Layout.
So this:
var list layout.List
list.Layout(gtx, 10, func(gtx C, index int) D {
return material.Body1(th, fmt.Sprintf("%d", index)).Layout(gtx)
})
Becomes:
var list widget.List
material.List(th, &list).Layout(gtx, 10, func(gtx C, index int) D {
return material.Body1(th, fmt.Sprintf("%d", index)).Layout(gtx)
})
Naturally, the material.ListStyle type supports tweaking the scrollbar's
appearance and behavior.
Signed-off-by: Chris Waldon <christopher.waldon.dev@gmail.com>
This commit adds a Length field to the
layout.Position. This field contains an approximation
of the overall length of the list's contents.
Signed-off-by: Chris Waldon <christopher.waldon.dev@gmail.com>
This is a breaking change as Icon.Layout no longer requests a size.
Before:
sz := unit.Dp(20)
ic.Layout(gtx, sz)
After:
sz := gtx.Metric.Px(unit.Dp(20))
gtx.Constraints.Min = image.Pt(sz, 0)
ic.Layout(gtx)
Fixes gio#240
Signed-off-by: pierre <pierre.curto@gmail.com>
This creates a floating-point analog to layout.Axis.Convert
for converting from (x,y) coordinate space to (main,cross)
coordinate space.
Signed-off-by: Chris Waldon <christopher.waldon.dev@gmail.com>
The platform GPU context must be Refreshed on the window event thread,
but our rendering loop must not, because it may also want access to the
window event thread.
Fixes gio#236
Signed-off-by: Elias Naur <mail@eliasnaur.com>
The app.Window owner may run SetAnimating just before window close,
which in turn rely on an active display link. This change makes sure
the link is stopped after window close where no more driver calls
can occur.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
The macOS app menu would respond to clicks, only to shotcuts (Cmd-Q,
Cmd-H). Moving setActivationPolicy to applicationDidFinishLaunching
seems to fix that, although I can't explain why.
Move a SetDriver call after initialization while here.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
The NSViewGlobalFrameDidChangeNotification notification is documented to
be fired every time [NSOpenGLContext update] needs to be called.
However, the notification fails to fire on my setup when a window is
moved to a display with a different pixel scale, which leads to
incorrectly sized output.
This change gets rid of the notification and updates the context before
every frame.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
To ease the integration with foreign OpenGL contexts, carefully save the
context state before rendering a frame and restore it afterwards. Gio
rendering can then be mixed with OpenGL code that expects exclusive
control over context state.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Now, it's possible to use `app.Fullscreen` on Android devices. It uses
the "Fullscreen Sticky Immersive" mode.
Signed-off-by: Inkeliz <inkeliz@inkeliz.com>
Previously, the on-screen keyboard always displays the text keyboard,
(QWERTY or equivalent).
For optimal user experience, it's possible to specify the keyboard type
using `InputHint`. The on-screen keyboard will provide shortcuts or
restrict what the user can input.
Due to some limitations (gio#116), only numeric and text keyboards are
supported on Android.
Signed-off-by: Inkeliz <inkeliz@inkeliz.com>
Elias Naur