This tests submitting commands, signals, and waits to graphics and
multiple compute queues where the entire frame's submission for each
queue depends on the other queues and the submission order is not the
order in which the GPU (and replay) need to be processed. This is to
ensure any capture or replaying deserialises adequately.
Added 'D3D12BufferCreator::InitialState(D3D12_RESOURCE_STATES)' to allow
user code to avoid adding unnecessary transition barriers after resource
creation.
Added 'D3D12TextureCreator::ClearVal(const D3D12_CLEAR_VALUE &)' to avoid
a PIX performance warning/tip about clearing resources that haven't had
their clear value set on creation.
New Binding 3
Initial size is UINT32_MAX
Updated to be sized DESC_ARRAY3_SIZE
Used in the shader to ensure the binding is referenced
Stress tests the FetchShaderFeedback code when it computes the maximum size of feedbackStorageSize.
Compiles the demo APK with -Wno-deprecated-declarations when the NDK is v26 or over, otherwise it fails to compile the fmt core header. In the LLVM version used in the NDK, using std::char_traits with non-std types is marked as deprecated.
* Fixed issue where a previously built demo APK would be added into the APK archive upon next build
* Correct the functional test's --in-process help output
Change package name to "renderdoc.org.demos" instead of "org.renderdoc.demos" (RenderDoc filters out packages containing "org.renderdoc.")
Added an <intent-filter> to the Android manifest.
This registers the activity with the android package system which means RenderDoc Launch Application knows the activity name to launch.
Used category CATEGORY_ALTERNATIVE to prevent the activity from appearing on the launcher UI.
This is a basic test that mirrors the Vulkan pixel history test,
but repeats draws multiple times to verify SM5.1 and SM6.0.
SM6.6 should be tested as well for writes from direct heap access
but is not yet implemented differently than SM6.0.
* We need to check for map writes even if the mapped buffer isn't the one that's
referenced by a submit but still overlaps it. We do this by moving to the heap
for all placed resources so any map to a buffer on a heap is checked as long
as any buffer bound to that heap is referenced.
Jake Turner