* PipelineLibrary (PSO Caching) support for Metal
- API changes to handle Metal drivers implictly doing save/load of PipelineLibrary data
- Fixed up code related to Metal device selection
- PipelineLibrary support for Mac and ios
Signed-off-by: moudgils <moudgils@amazon.com>
* Fix compile errors for Dx12, Vulkan backend + Unit tests
Signed-off-by: moudgils <47460854+moudgils@users.noreply.github.com>
* Fixed errors related to M1 GPU
Signed-off-by: moudgils <moudgils@amazon.com>
* Fix a minor 'tab' validation issue
Signed-off-by: moudgils <moudgils@amazon.com>
* Addressed feedback
Signed-off-by: moudgils <moudgils@amazon.com>
* Minor feedback
Signed-off-by: moudgils <moudgils@amazon.com>
* Added a few asserts
Signed-off-by: moudgils <47460854+moudgils@users.noreply.github.com>
* Fix a typo
Signed-off-by: moudgils <moudgils@amazon.com>
* Updated all array_view uses with the C++20 span.
The updates were done in the following order
1. `AZStd::array_view<([^>].+)\* ?>` -> `AZStd::span<\1 const>`
2. `AZStd::array_view<(?:const )(.+)>` -> `AZStd::span<const \1>`
3. `AZStd::array_view` -> `AZStd::span`
Removed the implementation of array_view.
Signed-off-by: lumberyard-employee-dm <56135373+lumberyard-employee-dm@users.noreply.github.com>
* Added missing whitespace between `const` and the typename for spans.
Updated the ShaderTest comparison of the ShaderResourceGroupLayout span
to compare the sizes as well
Updated comments on some of the methods that stated that they return "an
array" to mention they return "a span".
Signed-off-by: lumberyard-employee-dm <56135373+lumberyard-employee-dm@users.noreply.github.com>
* Support to add a different PSO cache per vendor/driver version. Also added support to have a differnt cache for Warp
Signed-off-by: moudgils <47460854+moudgils@users.noreply.github.com>
* Added a way to reset PSO cache for everyone
Signed-off-by: moudgils <47460854+moudgils@users.noreply.github.com>
* Fix tabbing for one line which is failing validation
Signed-off-by: moudgils <47460854+moudgils@users.noreply.github.com>
* Removed chatty profile markers.
These 4 markers removed tens of thousands of events, which reduced an example capture from 1685 ms to 1614 ms and made it far more readable.
Signed-off-by: Mike Balfour <82224783+mbalfour-amzn@users.noreply.github.com>
* Restored one marker
Signed-off-by: Mike Balfour <82224783+mbalfour-amzn@users.noreply.github.com>
* The Build Time Stamp of ShaderAsset And
ShaderVariantAsset Should Be Based On
GetTimeUTCMilliSecond()
GetTimeNowMicroseconds() is useful to measure time stamp differences.
GetTimeUTCMilliSecond() is for time stamps based on absolute clock/wall time.
* Updated DiffuseGlobalIllumination precompiled shaders
Co-authored-by: dmcdiar <dmcdiar@amazon.com>
Signed-off-by: galibzon <66021303+galibzon@users.noreply.github.com>
During asset processing, no `RPISystem` component is created, so nothing
so nothing calls `ShaderSystem::Init()`, so nothing calls
`ShaderReloadDebugTracker::Init()`. Consequently, the
AZ::EnvironmentVariables that are used during
`ShaderReloadDebugTracker::IsEnabled()` never got created, causing a read
from a nullptr at runtime. This fixes that issue by making `IsEnabled()`
call `CreateVariable()` on the variables it needs if they are not valid.
In addition, it changes the call to `CreateVariable()` to initialize the
variable's values directly, to ensure they are only initialized once. It
also switches to use `AZ::Crc32` so that the variable's id is computed at
compile time.
Signed-off-by: Chris Burel <burelc@amazon.com>
* Shaders changes require two or more change cycles before updating
This fixes the problem described in the title.
Consolidated the responsibility to update the root shader variant
asset into the Shader() class. It was unnecessarily spread across
Shader(), ShaderVariant() and ShaderAsset().
In particular OnAssetReloaded now makes a temporary copy of the root
ShaderVariantAsset and updates the ShaderAsset with such reference
only when OnAssetReloaded() is called on behalf of the ShaderAsset.
Signed-off-by: galibzon <66021303+galibzon@users.noreply.github.com>
Made ShaderReloadDebugTracker store its static data in Environment system variables, so they are shared across dlls. This fixes issues with inconsistent indenting when debug operations are performed in different libraries.
New ShaderReloadDebugTracker operations in FullscreenTrianglePass.
Added a ShaderReloadDebugTracker message to Shader::GetVariant that includes asset built timestamp infromation, which I think will be really helpful in sorting out reload issues.
Renamed some functions and variables to remove a redundant "ShaderAsset" term.
Signed-off-by: santorac <55155825+santorac@users.noreply.github.com>
- Each resource type is tracked and updated separately
- Added caching ability for Raytracing srg to save ~2ms for a scene containing 100 x 50 vegetation patch
Signed-off-by: moudgils <moudgils@amazon.com>
* ATOM-16249 Adding draw srg caching to DynamicDrawContext since creating SRG can be expensive for some backend
Signed-off-by: Qing Tao <qingtao@amazon.com>
* Fix issues with seedlist for AutomatedTesting
Fixes error reporting so it will show the file hint in the tool.
Removes any missing assets from the .seed file.
Remove an unnecessary dependency from AutomatedTesting dependencies
file.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Make ArchiveComponent use AZ::IO::IArchive
Initial changes that will get the sychronous calls in ArchiveComponent
to use IArchive interface rather than external zip/7z tools.
Some of the asynchronous api are still in place, anything that wasn't
being used has been removed for now. This may change later if we move
towards all the api being asynchronous. Until then, we can't remove the
reliance upon the external archive tools completely.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Updates AZStd::thread constructors
Adds a variadic constructor which forwards args to the functor.
Because of our thread_desc extension, there was confusion on the
arugments, so the args were reordered to take the thread_desc first,
before the functor and args.
Also the thread_desc is taken as reference rather than by pointer.
Update callsites to account for this change.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Async operation of ArchiveComponent api
This sets up the ArchiveComponent to operate asynchronously. It uses
promise/future to transfer results to caller.
This is still broken, there's a few things that need to get fixed up,
but this is a good checkpoint for the work as it solidifies the api,
cleans up a bunch of unused code, and compiles.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Removes the platform-specific ArchiveComponen
These are no longer needed, as they control the direct interaction with
host OS tools like 7za.exe or /bin/zip.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Removes the platform-specific files from cmake
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Removes the 7za.exe (and legal notice)
This tool is no longer needed in the repo.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Fixes usage of IArchive::GetFullPath()
This changed to return a PathView, updated to reflect that.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Fix promises and threads
Make sure promises are only set exactly once. This meant reworking some
of the initial error checking.
Detach threads when created. Adds [[nodiscard]] to the functions that
return a future. Since threads are detached, the future is the main way
to get communication from the thread.
Clean up interface, add comments.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* More edits to thread creation
Changes to thread construction to account for parameter change.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Fix some remaining issues with ArchiveComponent
Put created threads inside a container, then join them at Deactivate.
Fix asset bundler case when injecting a file with no working directory.
Fix thread constructor that applies args to a function.
Fix lambdas to take string args by value rather than reference.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Fixes some remaining bugs in ArchiveComponent
Open archive as read-only during extract & list operations.
Fixes paths issues.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Fix initialize of opaque thread handle in thread_UnixLike.h
Signed-off-by: lumberyard-employee-dm <56135373+lumberyard-employee-dm@users.noreply.github.com>
* Removed unused variable in AssetBundleComponent.cpp to fix compiler warning
Signed-off-by: lumberyard-employee-dm <56135373+lumberyard-employee-dm@users.noreply.github.com>
* Fix some issues with archives
File paths in the CDR and the local headers need to match, but there
were issues with path separators and case that made it possible to get
invalid headers errors in some archives.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Adds some new ArchiveComponent unit tests
Adds new tests for extraction of archive and adding files from a file
list to an archive.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Fix file data offset issues when opening archives
When opening an INestedArchive it would run through the CDR headers to
create file entries in the zip cache. The offsets to the compressed
data were being calculated incorrectly because they were using the CDR
headers rather than jumping to the local file headers and getting
offsets from those sizes.
Removed and refactored some archive validation flow and zip cache
factory init methods to either init default or init w/ additional
validation checks.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Addresses PR feedback
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Address more points of feedback in PR
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Address additional PR feedback
Fixes up some error checks and uses of strings vs paths.
Enable archive component tests on Linux so we can see if they will
pass.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Address PR feedback
Change the INestedArchive interface to list files as AZ::IO::Path.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Disabling the ArchiveComponent tests on Linux
They failed so we will revisit them to attempt a fix.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Rename a member variable to be more accurate
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Address feedback on PR
Bump version of Archive Components for serialize context.
Improve error messages during archive open and validation.
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
* Revert recent changes
Signed-off-by: amzn-phist <52085794+amzn-phist@users.noreply.github.com>
Co-authored-by: lumberyard-employee-dm <56135373+lumberyard-employee-dm@users.noreply.github.com>
- Added new interface type AZ::Debug::Profiler to externally register profiler systems
- Modified the Atom CPU profiler to register as an AzCore profiler
-- This allows full engine markers to be visualized in the associated ImGui tool
- Converted all AZ_ATOM_PROFILE_* macros to use AZ_PROFILE_* macros instead
Signed-off-by: AMZN-ScottR 24445312+AMZN-ScottR@users.noreply.github.com
Many of these are just extra noise in the profile, but the one in Archive.cpp could also cause PIX to crash.
Signed-off-by: Mike Balfour <82224783+mbalfour-amzn@users.noreply.github.com>
* Fix loading of PipelineLibraries from disk for DX12 backend.
Signed-off-by: moudgils <moudgils@amazon.com>
* Disabled Saving out PipelineLibraries (for DX12) if pix or Renderdoc is enabled.
Addressed some feedback
Signed-off-by: moudgils <moudgils@amazon.com>
* Fixed an issue withe loading PipelineLibraries and added a cleaner abstraction to not save empty libraries for dx12
Signed-off-by: moudgils <moudgils@amazon.com>
Merge pull request #3574 from aws-lumberyard-dev/Atom/santorac/MoreShaderHotReloadFixes
Fixed a shader hot reload issue where the new root shader variant asset was not getting saved in the Shader object during OnAssetReloaded, it was only saved during OnAssetReady.
Fixed shader variant hot reload which was failing due to mismatched timestamps. The ShaderAsset was using microseconds and the ShaderVariantAsset was using system ticks. Since ticks will always be higher than microseconds, stale variants were not prevented from being used.
AtomSampleViewer's MaterialHotReloadTest script is passing again.
NOTE: The memory driller is still intact for now to avoid needing to
modify allocators, but the frame/cpu portions of driller and the
standalone executable are now gone.
Signed-off-by: Jeremy Ong <jcong@amazon.com>
* Metal sync interval support
Add shader name to SRGPool for debugging purposes
Set default value of rpi_vsync_interval to 1
Signed-off-by: moudgils <moudgils@amazon.com>
* Final update copyright headers to reference license files at the repo root
Signed-off-by: spham <spham@amazon.com>
* Fix copyright validator unit tests to support the stale O3DE header scenario
Signed-off-by: spham <spham@amazon.com>
ATOM-15837 Remove Unused Shader System Related *2 Classes From Stabilization Branch.
Testing: ASV full test suite on dx12 and vulkan. Only saw known issues.
Added a ShaderSystem supervariant to provide a system-wide supervariant name.
Changed ShaderAsset to append the system-wide supervariant name when searching for supervariants.
Added the NoMSAA supervariant to several shaders.
moudgils
lumberyard-employee-dm
Chris Galvan
dmcdiarmid-ly
nemerle
Mike Balfour
Esteban Papp
Tommy Walton
galibzon
Chris Burel
santorac
Qing Tao
amzn-phist
bosnichd
Scott Romero
Jeremy Ong
chcurran
Steve Pham
Gene Walters
Chris Santora
Doug McDiarmid
nvsickle