|
0561884e
|
2023-04-26T17:26:42
|
|
Vulkan: Dirty VertexArray binding bit if buffer storage change
In crrev.com/c/3669603, we did optimization for black_desert_mobile that
when vertex array is unbound, we remove vertex array from buffer's
observer list to reduce overhead of observer notifications when buffer
is been modified. To compensate for the lost notification, when vertex
array is bound, we always assume every buffer that is bound to vertex
array has been dirtied, for the simplicity at that time. This CL further
the optimization of that CL. In this CL, I moved the dirty bit set into
backend and improves vulkan backend by checking buffer's serial number
and only dirty the binding if the serial has changed. Given this, now we
can also remove all the non-current vertex array from buffer's observer
list (previously it is heuristic based with a hard coded observer count
limit). This and the previous CL improves asphalt_9 by ~1%.
Bug: b/277644512
Change-Id: Ibc3f8e3df9fe70c6879e0b2bca86d8487a9dba73
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4481241
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuxin Hu <yuxinhu@google.com>
Commit-Queue: Charlie Lao <cclao@google.com>
|
|
44bb4d7a
|
2020-07-22T13:58:50
|
|
D3D11: Fix bug with static vertex attributes.
In some specific cases after binding a zero size buffer we could end
up trying to use a buffer storage that was no longer valid. Fix this
by ensuring we don't flush dirty bits when we have an early exit due
to a zero size buffer.
Also adds a regression test.
Bug: chromium:1107433
Change-Id: I9db560e8dd3699abed2bb7fe6d91060148ba1817
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/2314216
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
9349e262
|
2020-06-25T14:28:17
|
|
Fix multiDraw* with DYNAMIC_DRAW buffer usage
Specialized Context11 impl to prevent promote Dynamic during MultiDraw.
Add different buffer usage to MultiDraw and BaseVertexBaseInstance
tests.
Bug: angleproject:3402, angleproject:4754
Change-Id: I50e1a6fd6c8b6fd48f130c43545b895335d2e55b
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/2268581
Commit-Queue: Shrek Shao <shrekshao@google.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
e6124500
|
2020-06-02T12:02:55
|
|
Fix baseVertex and baseInstance with streaming attributes
baseInstance: Fixed by adding the intial offset to each copy for
streaming attributes
baseVertex: make sure mShaderConstants.onFirstVertexChange takes
in correct firstVertex value for dynamic attribs (where firstVertex
passed to StateManager11::updateState already include baseVertex)
Bug: chromium:1078330, angleproject:3402
Change-Id: I289c4e3733fdf6f78af8c3adee84112c05a5abce
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/2227022
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Shrek Shao <shrekshao@google.com>
|
|
|
57d95828
|
2020-04-30T17:35:50
|
|
Revert "Add type for attribute locations."
This reverts commit 9349c14344b2d1fd6bc357063b602bc2626c140f
and commit d43b057435e6c9e3194dd20627681ffca0c0808e.
It's no longer needed after we bind attribute locations before link.
Original CL message:
This will allow the capture/replay tool to easily intercept and label
attribute locations for remapping.
There's some inconsistency in implementation in the GL desktop front-
end. This is a quick fix and the full implementation is left for when
we implement the full desktop GL API set.
Bug: angleproject:4598
Change-Id: Ic510159d4d1982eff41560503cabf983a1be0381
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/2174076
Reviewed-by: Cody Northrop <cnorthrop@google.com>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
9349c143
|
2020-04-29T16:36:17
|
|
Add type for attribute locations.
This will allow the capture/replay tool to easily intercept and label
attribute locations for remapping.
There's some inconsistency in implementation in the GL desktop front-
end. This is a quick fix and the full implementation is left for when
we implement the full desktop GL API set.
Bug: angleproject:4598
Change-Id: Ibf11bcb8669d27265ea376494a2e3124825cf3be
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/2171933
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Cody Northrop <cnorthrop@google.com>
|
|
7e0699a2
|
2020-02-05T17:04:06
|
|
Create the ProgramExecutable Class
The ProgramExecutable class is being created to collect data structures
that are common to both Programs and ProgramPipelines, as well as any
shared functions. This allows callers to request the current
ProgramExecutable from the State and make
Program-/ProgramPipeline-specific queries without needing to know
exactly which responded. This will also allow the necessary data
structures to only be populated and stored within the ProgramExecutable
when necessary and reused as often as necessary.
Bug: angleproject:3570
Change-Id: I101f08ab03421894667b4a426a04d2147489f0e1
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/2040512
Commit-Queue: Tim Van Patten <timvp@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
ebab670c
|
2019-04-19T14:36:45
|
|
Adding new extension GL_OVR_multiview
GL_OVR_multiview functions exactly the same as GL_OVR_multiview2.
All GL_OVR_multiview2 tests now also repeat the same test using GL_OVR_multiview
Bug: angleproject:3341
Change-Id: I7e5294fb6bbf7692535174a15da6a42e1b5fc4e2
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/1575904
Commit-Queue: Rafael Cintron <rafael.cintron@microsoft.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
c09ae15c
|
2019-02-01T14:16:32
|
|
Enable -Wextra-semi and -Wextra-semi-stmt.
This will prevent users from accidentally making semicolon errors in
the future.
Bug: chromium:926235
Change-Id: I79a6fa376fb1ad8f0fcf1b65b1f572a035d1f4e9
Reviewed-on: https://chromium-review.googlesource.com/c/1446493
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Nico Weber <thakis@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
c759b8b4
|
2019-01-03T15:16:50
|
|
Vulkan: More Vertex Array optimizations.
Inlines a number of Vulkan vertex array methods.
Also changes the way vertex buffers are bound. Note that Vulkan doesn't
support NULL buffer bindings. Thus we create an emulated NULL buffer
to work around the problem of having gaps in the bound vertex buffers.
This allows us to use a single bind call for ranges of vertex buffers
even when there are gaps.
Also changes how vertex array dirty bits are reset. Instead of calling
memset to clear the affected buffers we pass a mutable pointer to the
Vertex Array sync state. This allows us to only reset the dirty bits
that we sync. This saves on the memory clearing time.
Improves perf by about 10% in the Vulkan VBO state change test.
Bug: angleproject:3014
Change-Id: Ib7b742dff7897fc891606a652ea0b64255a24c86
Reviewed-on: https://chromium-review.googlesource.com/c/1390360
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
c3dc5d48
|
2018-12-30T12:12:04
|
|
Merge gl::Context and gl::ContextState.
This reduces the number of indrections when accessing the Extensions
or Caps structures. It will provide a small speed-up to some methods.
It also cleans up the code.
Bug: angleproject:2966
Change-Id: Idddac70758c42c1c2b75c885d0cacc8a5c458685
Reviewed-on: https://chromium-review.googlesource.com/c/1392391
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Markus Tavenrath <matavenrath@nvidia.com>
|
|
|
7c985f5c
|
2018-11-29T18:16:17
|
|
Make angle::Result an enum.
This moves away from a class type to a value type. This should improve
performance when using angle::Result as a return value. Previously the
generated code would return a pointer instead of a value.
Improves performance in the most targeted microbenchmark by 10%. In
more realistic scanarios it will have a smaller improvement. Also
simplifies the class implementation and usage.
Includes some unrelated code generation changes.
Bug: angleproject:2491
Change-Id: Ifcf86870bf1c00a2f73c39ea6e4f05ca705050aa
Reviewed-on: https://chromium-review.googlesource.com/c/1356139
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
8dc27f99
|
2018-11-29T11:45:44
|
|
Use packed enum for DrawElementsType.
The packing and unpacking take a few extra instructions. But it
completely obviates the need for any switches in the validation code.
Speed is slightly faster or the similar depending on the back-end.
Also add gl_angle_ext.xml to GL entry point generator inputs. This was
missing and would cause the code generation to miss certain changes.
Bug: angleproject:2985
Change-Id: I1ea41a71db71135000166ead8305ec42d22ff7b3
Reviewed-on: https://chromium-review.googlesource.com/c/1351729
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
b980c563
|
2018-11-27T11:34:27
|
|
Reformat all cpp and h files.
This applies git cl format --full to all ANGLE sources.
Bug: angleproject:2986
Change-Id: Ib504e618c1589332a37e97696cdc3515d739308f
Reviewed-on: https://chromium-review.googlesource.com/c/1351367
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
bfe31c42
|
2018-10-25T17:03:47
|
|
Remove uses of DrawCallParams.
Packing and referencing this structure was causing unnecessary draw
call overhead. This improves performance on all the back-ends. Impacts
the GL back-end the most.
In total this patch series reduces overhead by up to 5%.
Bug: angleproject:2933
Change-Id: Ief416ab874e481baf960d02965978a311214a146
Reviewed-on: https://chromium-review.googlesource.com/c/1299477
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
c1fd7376
|
2018-10-26T22:48:39
|
|
Move index range calculations into VertexArray.
This is in preparation for removing the entire DrawCallParams struct.
This struct was big enough to cause a performance hit on draw call perf
tests just by virtue of initializing the fields. Also dereferencing the
struct members is slower than reading function parameters since it adds
an indirection.
Also includes some error refactoring to enable moving code to a shared
location.
In total this patch series reduces overhead by up to 5%.
Bug: angleproject:2933
Change-Id: Ib663f2538c14ac30d4c31fd10d6350be469626e2
Reviewed-on: https://chromium-review.googlesource.com/c/1298380
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
cd0a0a3c
|
2018-10-18T18:41:57
|
|
Introduce SubjectBindingPointer.
We can share the same pointer for the subject binding and the binding
pointer. This further allows us to optimize buffer re-binding. The
shared memory increases cache coherency and reduces the number of
instructions needed.
Bug: angleproject:2891
Change-Id: Id3162fa79de203f75989e7289ea02cb2ea1bec73
Reviewed-on: https://chromium-review.googlesource.com/c/1270217
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Frank Henigman <fjhenigman@chromium.org>
|
|
|
6f755b21
|
2018-10-09T12:48:54
|
|
Use angle::Result in front-end. (Part 1)
This covers most of the hot paths used in draw calls. Gives in the
order of a 5% reduction in draw call overhead.
Bug: angleproject:2491
Change-Id: I2d53afb1163eaceed61fb9cd9ce6c1267c85c0fa
Reviewed-on: https://chromium-review.googlesource.com/c/1258149
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
ec1fe5b7
|
2018-08-10T10:05:52
|
|
D3D: Use angle::Result error pattern.
This completes the refactor for the D3D9/D3D11 back-ends.
Bug: angleproject:2752
Change-Id: Ie35a925f75c902d8f9fdab6bc33b3bb9c937f85b
Reviewed-on: https://chromium-review.googlesource.com/1167209
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
ad63728b
|
2018-07-31T11:22:14
|
|
VertexArray: Store bound attributes mask in binding.
Making the bound attributes mask part of the struct keeps the style a
bit more consistent. It will also make updating the cache variable a
bit easier.
Bug: angleproject:1391
Change-Id: I2b0ba1161d2579d95ddfbae7a05e490c9644a126
Reviewed-on: https://chromium-review.googlesource.com/1150513
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
306b6c16
|
2018-07-27T08:12:49
|
|
D3D11: Use angle::Result error pattern. 1/3
This CL improves performance on the draw call microbenchmark by 10%
when no-oping driver calls.
Bug: angleproject:2738
Change-Id: I4f5c11db90d9056ce4557b2a4432bc55b42b5bba
Reviewed-on: https://chromium-review.googlesource.com/1150093
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
b1565903
|
2018-07-27T08:12:48
|
|
D3D: Pass gl::Context to many more functions.
This makes the Context more available for logging errors.
Also includes more refactoring to VertexDataManager to ensure we can
access the gl::Context.
Bug: angleproject:2738
Change-Id: Iae3d22a1403078d236bfe63a3e2d203c13678dc4
Reviewed-on: https://chromium-review.googlesource.com/1151449
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
6a5d98c4
|
2018-05-04T15:42:20
|
|
ES31: Implement Vertex Attrib Binding on D3D11
This patch implements Vertex Attrib Binding on D3D11 and enables all
the test cases related to Vertex Attrib Binding on D3D11 back-ends.
On D3D11 back-ends the information in both GL vertex attributes and
bindings should be updated together. When a binding is dirty, we need
to find out and update all the attributes that are using this binding.
To speed up this process, this patch adds a map from each binding to
all the attrib indexes that are using this binding. This map may be
updated when VertexAttribBinding is implicitly or explicitly called.
With this map we can easily get all the attributes that should be
updated with the current dirty binding.
This patch also removes some unused variables in VertexArray11.cpp.
BUG=angleproject:2700
TEST=dEQP-GLES31.functional.vertex_attribute_binding.*
angle_end2end_tests
Change-Id: I9a28ec357fd3aba835812cecc410cfa4e3734f0c
Reviewed-on: https://chromium-review.googlesource.com/1048980
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
a56467e0
|
2018-04-11T16:19:41
|
|
VertexArray: Use switch macro for faster iteration.
Has a small but noticeable impact on performance on a microbenchmark.
Seems to improve a synthetic score by about 1%. Should have a very
small improvement in real-world performance.
Note that the odd formatting is an idiosyncrasy of clang-format.
Bug: angleproject:2389
Change-Id: I888bf101c6d8b80a0fbafdb9c5a84205c9c8fee6
Reviewed-on: https://chromium-review.googlesource.com/962963
Reviewed-by: Luc Ferron <lucferron@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
2dadd1d0
|
2018-04-11T09:25:18
|
|
D3D11: Fix inactive attrib VAO perf regression.
In the case there are any dirty inactive attributes, we were
continually invalidating the Vertex Buffer and Input Layout
bindings. This fixes the invalidation to only happen when
there are any dirty active attributes.
This regressed in "Move Buffer Subject/Observer to front end."
Bug: chromium:829906
Bug: angleproject:2389
Change-Id: I8ed616bb696e0be548344192037ad6cc6f9c595c
Reviewed-on: https://chromium-review.googlesource.com/1006998
Reviewed-by: Luc Ferron <lucferron@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
0946393d
|
2018-04-04T05:26:59
|
|
Move Buffer Subject/Observer to front end.
This makes BufferImpl into an Observer Subject. It also refactors
the Vertex Array updates for the D3D11 backend use more of a dirty
bit coding style.
This change makes it so Buffer contents changes trigger front-end
dirty bits from the back-end, which may be undesirable.
Bug: angleproject:2389
Change-Id: Iac8ce1171284a86851c18cd1373ddf24fcefe40b
Reviewed-on: https://chromium-review.googlesource.com/979812
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
0af5b86a
|
2018-03-27T20:19:33
|
|
Return gl::Error from VertexArray::syncState().
No functional change.
When we add vertex data format conversion to Vulkan we will need to be
able to return an error from VertexArray::syncState().
BUG=angleproject:2405
Change-Id: I4b537946ecbb6593280b6510c5cd8d8e3c65e8dd
Reviewed-on: https://chromium-review.googlesource.com/982897
Commit-Queue: Frank Henigman <fjhenigman@chromium.org>
Reviewed-by: Luc Ferron <lucferron@chromium.org>
|
|
|
e858cb1d
|
2018-03-27T09:44:32
|
|
Split VAO dirty bits to speed iteration.
Using > 64 bits (we had over 90) would use a much slower dirty bit
iteration. Speed this up by splitting the dirty bits into two levels.
The first top level only has a single dirty bit per attrib, per
binding, and one bit for the element array buffer. The next level has
separate dirty bits for attribs and bindings.
The D3D11 back-end doesn't actually care about individual dirty bits
of attribs or bindings, since it resets entire attributes at a time,
but the GL back-end only refreshes the necessary info.
Improves the score of a simple state change microbenchmark by 15% on
the D3D11 and GL back-ends with a no-op driver. Real-world impact will
be smaller.
Also includes a test suppression for an NVIDIA bug that surfaced when
we changed the order of that GL commands were sent to the driver.
BUG=angleproject:2389
Change-Id: If8d5e5eb0b27e2a77e20535e33626183d372d311
Reviewed-on: https://chromium-review.googlesource.com/556799
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
d7518622
|
2018-03-27T09:44:31
|
|
Buffer11: Refactor Subject/Observer pattern.
Instead of having a direct/static observer distinction, add two
messages for 'Contents Changed' and 'Storage Changed'. This makes
Buffer11 itself the subject with two different message handling
cases in the onSubjectStateChange methods.
Bug: angleproject:2389
Change-Id: I645cd4b7cc7ce51cb7f48a01c7fc72939cbe89fe
Reviewed-on: https://chromium-review.googlesource.com/957940
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
d779f6a9
|
2018-03-23T01:44:33
|
|
D3D11: Refactor draw call functions.
This allow for better code reuse for the Indirect draw calls.
It also cleans up the InputLayoutCache to be only responsible
for caching input layouts, and moves the logic for maintaining
state into StateManager11.
Bug: angleproject:2389
Change-Id: I84aae164bf1b94a394743cf58650adfdcfc2c17a
Reviewed-on: https://chromium-review.googlesource.com/948796
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
6f5444da
|
2018-03-14T10:08:11
|
|
Remove HasIndexRange.
This is superseded by the DrawCallParams class. Instead of
storing the context, we also return an error from the index
range resolution.
Bug: angleproject:2389
Change-Id: I9e7d58f006f51872eb3b52cbb9efbee16fff7ef6
Reviewed-on: https://chromium-review.googlesource.com/960570
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
44a73fcf
|
2018-03-06T09:32:17
|
|
Always Invalidate Vertex Attribute Translation
Fixes bug where mVertexAttribsNeedsTranslation was not being set when
vertex attribute changes internally from DIRECT to DYNAMIC. This changes
mVertexAttribsNeedTranslation to be set during any change in vertex
attribute type.
BUG=angleproject:2381
Change-Id: I91ae86624f0cb68bd8c41c3bcbdeb31fa1ccd9ed
Reviewed-on: https://chromium-review.googlesource.com/951894
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
97ab26fc
|
2018-03-14T10:08:10
|
|
Move DrawCallParams to the gl front-end.
Previously called DrawCallVertexParams. This params helper will be
passed into the VertexArray state sync method so we can handle
vertex attribute updates directly in the VertexArray implementation
instead of deferring the updates to the draw call, where the draw
call implementation method would have access to the draw call params.
Also includes the full range of draw parameters to DrawCallParams so
we can use it in more places.
Refactoring change only, subsequent work will contain more refactoring
and lead to code cleanups and easier dependent state updates for the
VertexArray implementation classes.
Bug: angleproject:2389
Change-Id: Ia84f8af54ae51eca94d8fb6f5b8adc88b8d981a7
Reviewed-on: https://chromium-review.googlesource.com/948787
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
87db7c52
|
2018-03-14T10:08:10
|
|
D3D11: Minor cleanup to DrawCallVertexParams.
The 'firstVertexDefinitelyZero' argument didn't actually have any
behavioural change to the code, so remove it. Also add an explicit
call to resolve the index range, instead of doing it implicitly when
certain values were queried.
Bug: angleproject:2389
Change-Id: I377c79006bfa541183fe94d0ed199fbaa1132afc
Reviewed-on: https://chromium-review.googlesource.com/948786
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
d444255a
|
2018-02-27T22:03:47
|
|
Refactor signal utils into Observer pattern.
These types were over-generalized. All use cases featured
arrays of resources attached to single parent resources. The
channel ID is sufficient to identify the child resource in the
parent, and having variadic template arguments wasn't necessary.
Futhermore we can rename these types to use the common Observer
pattern. This should make them more readable to new developers.
Also update some classes to inherit from Subject instead of
having a member Subject. This cleans up the code in a few places.
This should lead to a simpler refactor to allow dependent dirty
bits notifications in the Vulkan back-end.
In the following patch the signal_utils files will be renamed. They
are not renamed in this patch to ensure git history is preserved.
Bug: angleproject:2372
Change-Id: I17a3f2c8d92afd4bb3cba2d378c3a2e8a6d7fb11
Reviewed-on: https://chromium-review.googlesource.com/936690
Reviewed-by: Luc Ferron <lucferron@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
8e4bb4b1
|
2017-11-22T18:59:51
|
|
D3D11: Cache element array buffer updates.
This attempts to reduce the amount of redundant validation done
between indexed draw calls. It keeps the cached info in the
VertexArray11 class.
It also includes a fix to a missing direct buffer invalidation in
CopyBufferSubData which was turning up with the new caching.
Reduces overhead in the D3D11 indexed rendering perf test such that it
leads to an increased score of about 20%.
BUG=angleproject:2229
Change-Id: I63121bea19a9c8198e1925ed6a1460838e8f8955
Reviewed-on: https://chromium-review.googlesource.com/765262
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
acf2f3ad
|
2017-11-21T19:22:44
|
|
Apply Chromium style fixes.
This addresses several minor code quality issues that are validated
in Chromium, but not yet applied to ANGLE:
* constructors and destructors must be defined out-of-line
* auto is not allowed for simple pointer types
* use override everywhere instead of virtual
* virtual functions must also be defined out-of-line
Slightly reduces binary size for me (~2k on Win, 150k on Linux).
Bug: angleproject:1569
Change-Id: I073ca3365188caf5f29fb28d9eb207903c1843e6
Reviewed-on: https://chromium-review.googlesource.com/779959
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
378734c5
|
2017-11-20T16:32:56
|
|
D3D11: Lazy evaluation for draw call vertex params.
The vertex params are the 'start' vertex index and the number of
vertices in the draw call. For indexed draws, they sometimes mean we
need to compute the index range. In other cases, we don't. Defer
computing the index range as long as possible so we don't end up
computing the index range when we don't have to.
BUG=angleproject:2229
Change-Id: I8e125416ef6767787a14509b322efa8dcd8e4a34
Reviewed-on: https://chromium-review.googlesource.com/764676
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
|
|
|
492f58ec
|
2017-10-09T19:41:33
|
|
Rename VertexArrayImpl::mData to mState.
Refactoring change only.
BUG=angleproject:1898
Change-Id: I9f55651f923ff930c395a9bb575b4f86ad5d9cbd
Reviewed-on: https://chromium-review.googlesource.com/707689
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
e416e527
|
2017-10-10T11:01:20
|
|
Remove some unused codes
This patch will remove unused functions syncVertexAttributes and
hasDirtyOrDynamicAttrib, and private member mAppliedIBChanged.
Meanwhile, it changes hasDynamicAttrib to hasActiveDynamicAttrib
to avoid unnecessary update in some cases.
BUG=angleproject:1155
Change-Id: I29a0aa7fbc13874b7b070cf1ac5ec4134728f519
Reviewed-on: https://chromium-review.googlesource.com/707014
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
|
|
|
2bc94733
|
2017-09-20T21:57:29
|
|
D3D11: Minor optimizations to vertex attribute application.
Introduce a dirty bit for current value attribs, and try to speed up
the check for dirty vertex attribs.
This series of small optimizations gives about a 15% improvement on
the draw call benchmark for the D3D11 backend with the null driver.
BUG=angleproject:1155
Change-Id: Idf560efa3af62776e8bdbdf693f8b06db8792c21
Reviewed-on: https://chromium-review.googlesource.com/666048
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
c8bee335
|
2017-09-20T15:42:09
|
|
D3D11: Implement a dirty bit for Shaders.
This allows us to skip calling the dynamic shader generation and
program shader application when there haven't been any state
changes. It builds on the previous work that immediately update
state caches in the VertexArray11 and Framebuffer11. It should
improve performance in draw-call limited applications by a small
margin.
For reference, here are the conditions under which the shaders
are refreshed:
1. Directly changing the program executable
2. The vertex attribute layout
3. The fragment shader's rendertargets
4. Enabling/disabling rasterizer discard
5. Enabling/disabling transform feedback
6. An internal shader was used
7. Drawing with/without point sprites
Improves the score of the draw call stress test for the D3D11
back-end (with null driver) by about 40% on my test machine.
The 9_3 back-end seems to have an issue where the getSRV call
to a texture storage can change the "use level zero workaround"
status of the storage, which in turn will invalidate the state.
Since this is localized to 9_3 only, put in a hack to disable
an assert check for now.
BUG=angleproject:2151
Change-Id: Idbd0a31376691b33972e735d5833a9b02a8a4aa9
Reviewed-on: https://chromium-review.googlesource.com/666278
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Frank Henigman <fjhenigman@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
33510107
|
2017-09-20T10:39:18
|
|
Pass gl::Context to more Buffer methods.
This will allow us to pull out the Renderer from the Context in more
places in Buffer11, for state update. Impacts a few method calls in
a few places.
BUG=angleproject:2151
Change-Id: I1360caea65a94d3de4cd9f52d1b74b10439b02b3
Reviewed-on: https://chromium-review.googlesource.com/673136
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Frank Henigman <fjhenigman@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
ced5c86c
|
2017-08-17T16:05:29
|
|
D3D11: Handle multi-view Draw* calls
Because the ANGLE_multiview extension uses instancing to multiply
geometry for each view, Draw* calls with an active multiview program
have to be handled in the follwing way:
1) Convert non-instanced Draw calls to their instanced versions.
2) Multiply the number of instances in an instanced Draw call by the
number of views.
The patch also applies the viewport offsets to the viewport and scissor
rectangle and propagates the computed viewports and scissors to the
D3D11 runtime.
BUG=angleproject:2062
TEST=angle_end2end_tests
Change-Id: I8b4295c95c2cc0c1046c67e1fb1a782a46703292
Reviewed-on: https://chromium-review.googlesource.com/618331
Commit-Queue: Martin Radev <mradev@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
abf66fb3
|
2017-08-23T11:58:27
|
|
D3D11: Use dirty bits for applyVertexBuffers.
If the vertex array has any dirty or dynamic attribs, we must do an
update. Similarly, if a prior state change has left the vertex
state invalidated, we do an update. For instance, a program change
means we need a new input layout. If there was no such invalidation
or dirtyness we can skip the call to InputLayoutCache.
This improves the performance of the draw call benchmark (with no
state changes) by about 50% on the D3D11 null driver. Increases the
frames per second count of the aquarium demo with the passthrough
command buffer by about 25% on a test machine.
BUG=angleproject:1156
Change-Id: I8381999029f5b1912030a3342e96285a58f95e82
Reviewed-on: https://chromium-review.googlesource.com/616784
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
401345e4
|
2017-08-21T10:52:40
|
|
D3D11: Move more state into StateManager11.
This moves the input layout cache and vertex and index data managers
and related info into the state manager. This makes it easier to
manage the state application with regards to dirty bits.
Also updates the dirty current value handling in StateManager11.
BUG=angleproject:1156
BUG=angleproject:2052
Change-Id: I8de968a1f8416363aa1c49d9e9da129942d21275
Reviewed-on: https://chromium-review.googlesource.com/616783
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
a779b610
|
2017-07-24T11:46:05
|
|
Smart caching of VAO input layout.
Don't recompute the cached attribute layout for a program if the
vertex array info hasn't changed. We can use the Serial class to
know when a vertex array has identical state.
BUG=angleproject:1156
Change-Id: Ia11f6ac268f63c3299f6d6d80c2866009cb8429c
Reviewed-on: https://chromium-review.googlesource.com/529768
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
dde78e8c
|
2017-05-22T14:13:27
|
|
ES31: Implement Vertex Attrib Binding on OpenGL
This patch intends to implement Vertex Attrib Binding on OpenGL
back-ends:
1. Add supports for updating vertex attributes by Vertex Attrib
Binding APIs.
2. Refactor the process of updating vertex attribtues in class
VertexArray to make it easier to implement this feature.
BUG=angleproject:1593
TEST=dEQP-GLES31.functional.vertex_attribute_binding.*
Change-Id: I800e61518c552b94b84c415895ad31668b0a84b2
Reviewed-on: https://chromium-review.googlesource.com/510251
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
4928b7ca
|
2017-06-20T12:57:39
|
|
Proliferate gl::Context everywhere.
This gives the D3D back-end access to the GL state almost anywhere.
This uses the onDestroy hook for Textures to push errors up from
destructors, although they still don't quite make it to the Context.
There are places, such as in EGL object (Context/Surface) destruction,
where we end up calling through to GL implementation internals without
having access to a gl::Context. We handle this via a proxy Context
to a Display, basically a null context, that has access to impl-side
state like the Renderer pointer if necessary. It does not have access
to the normal GL state.
Also Pass gl::Context to RefCountObject::release(). Since we're using
destroy() methods now, we should not ever call the destructor directly.
BUG=angleproject:1156
Change-Id: Ie4c32ad6bf6caaff0289901f30b5c6bafa2ce259
Reviewed-on: https://chromium-review.googlesource.com/529707
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
fb05bcba
|
2017-06-07T15:43:18
|
|
Move the vk Serial class to renderer_utils.
This can be useful for other back-ends, for various types of state
management. Also redesign the class to use an opaque factory instead
of an increment operator. The class maintains the property of being
ordered. Also assume we don't overflow with 64-bit serials. We could
maybe redesign this to use 32-bit serials for memory constrained
situations, and handle overflow more gracefully.
I plan to use the serials to track state revisions for the vertex
array class, to avoid doing redundant work.
BUG=angleproject:1156
Change-Id: I02c78b228bc6e2fb3ee786fe67a4e607baaca18e
Reviewed-on: https://chromium-review.googlesource.com/529704
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
dd5f27ee
|
2017-06-07T10:17:09
|
|
Make VertexBinding's member variables private
The patch decorates all members in VertexBinding as private and limits
access to them only through getters and setters. This makes it easier to
debug and keep track of any assignments to the class members.
BUG=angleproject:2062
TEST=angle_end2end_tests
Change-Id: Iddd49063d060f136bc9cf11c313a5af0931d433c
Reviewed-on: https://chromium-review.googlesource.com/530786
Commit-Queue: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
c564c070
|
2017-06-01T12:45:42
|
|
Pass gl::Context to impl methods instead of ContextImpl.
In some cases we might have to call back into the GL layer, passing
the Context, and if we just have a ContextImpl pointer this isn't
possible. It also removes the need for SafeGetImpl.
BUG=angleproject:2044
Change-Id: I6363e84b25648c992c25779d4c43f795aa2866d6
Reviewed-on: https://chromium-review.googlesource.com/516835
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
25e297e9
|
2017-04-20T17:01:20
|
|
D3D11: Alias dirty buffer channel bindings.
The types for this were a bit disorganized. Clean this up.
BUG=angleproject:1635
Change-Id: I669d347697e0c7c5b06ed7165fa0997c375e876e
Reviewed-on: https://chromium-review.googlesource.com/483315
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
6de51858
|
2017-04-12T09:53:01
|
|
Optimize angle::BitSetIterator.
Adds a new custom bitset template to handle packing as many bits as
possible into a single variable. Intelligently select the right class
depending on platform features and bit sizes.
For now, always use a packed 64-bit set on 64-bit, instead of using
a 32-bit set for smaller bitsets.
BUG=angleproject:1814
Change-Id: I3ffef815c15515555833f6fc9302d8a4eee5423b
Reviewed-on: https://chromium-review.googlesource.com/471827
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
1e5499db
|
2017-04-05T11:22:16
|
|
Refactor Signal utils into template classes.
This will allow us to pass on extra information to the receiving end,
such as the specific texture levels that are dirty.
BUG=angleproject:1635
Change-Id: Idb7ca1d625499e50e7712c458b694f6e9bfc0595
Reviewed-on: https://chromium-review.googlesource.com/453382
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
20e005b2
|
2017-04-07T14:19:22
|
|
Rename BitSetIterator.h to bitset_utils.h
BUG=angleproject:1814
Change-Id: I152ae13b6b7cf0ba72259967f0f124e199b20e07
Reviewed-on: https://chromium-review.googlesource.com/471826
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
dd43e6cd
|
2017-03-24T14:18:49
|
|
Pass Context to VertexArray and Framebuffer syncstate.
This will enable more Vulkan-friendly idioms like clearing the
vulkan pipeline caches correctly on GL state changes immediately
because we have access to the ContextVk.
BUG=angleproject:1898
Change-Id: I16c848d8abdde8e26a38d384e565cec8548a66d0
Reviewed-on: https://chromium-review.googlesource.com/459079
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
fbe49a8f
|
2016-12-19T14:19:10
|
|
ES31: Implement DrawArraysIndirect D3D part
There are four buffer types for vertex attribute storage in D3D11:
DIRECT
STATIC
DYNAMIC
CURRENT_VALUE
When drawing, it will call applyVertexBuffer to bind the right type
buffers in D3D11.
DIRECT uses the gl buffer directly without any translation.
CURRENT_VALUE uses a single value for the attribute.
STATIC translates the whole vertex buffer once. So it doesn't need the
first, count and instance informations since it always translates the
whole buffer.
DYNAMIC translates the data every frame. To improve the performance,
in implementation, it only translates 'count' vertexes from 'first'
location in vertex buffer with one drawing for non-instanced vertices.
'first' and 'count' are got from draw parameter list. And for the
translated vertex buffer, when drawing, the first vertex location is 0.
From above analysis, we can see that if all attribute storages are
non-dynamic, we can directly use the indirect buffer to draw. But for
dynamic storages, we have to calculate the first, count, and instances
from indirect buffer and apply them to translate the dynamic type
buffers. Meanwhile, we have to set the first to 0 (see above
description)when drawing.
DrawArrysIndirect implementation is like below:
1. Check whether all vertex attributes are non-dynamic
2. If yes, applyVertexBuffer and DrawInstancedIndirect
3. If no, 1) calculate first, count, and instances from indirect buffer.
2) applyVertexBuffer with these parameters.
4) Use DrawInstanced instead of DrawInstancedIndirect.
BUG=angleproject:1595
TEST=dEQP-GLES31.functional.draw_indirect.draw_arrays_indirect*
Change-Id: I36431f416443279d51de523b07ce60727914cbbf
Reviewed-on: https://chromium-review.googlesource.com/446690
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
cde4f024
|
2017-02-21T14:06:02
|
|
Remove unused parameters of StoreStaticAttrib
The parameters count and instances in VertexDataManager::StoreStaticAttrib
are never used in the function body. So this patch will remove them.
BUG=none
Change-Id: Ic3fe21ae50e5016145bfd0c78542540c8ec76fd2
Reviewed-on: https://chromium-review.googlesource.com/445337
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yunchao He <yunchao.he@intel.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
2597fb64
|
2016-12-09T16:38:02
|
|
ES31: Refactor VertexArray for Vertex Attrib Binding
OpenGL ES3.1 feature Vertex Attrib Binding requires vertex arrays should
be split into two arrays:
1. an array of vertex buffer binding points, each of which specifies:
- a bound buffer object,
- a starting offset for vertex attribute data in that buffer object,
- a stride used by all attributes using that binding point,
- a frequency divisor used by all attributes using that binding point.
2. an array of generic vertex attribute format information records, each
of which specifies:
- a reference to one of the new buffer binding points above,
- a component count and format, and a normalization flag for the
attribute data,
- the offset of the attribute data relative to the base offset of each
vertex found at the associated binding point.
Current ANGLE implementation simply uses a struct to represent a vertex
attribute object, which does not meet the requirements above.
This patch aims to be the the basis of the implementation of all ES3.1
Vertex Attrib Binding APIs by refactoring the struct VertexAttribute and
the class VertexArray to fit the new data layout and ensuring all current
functionality is retained.
BUG=angleproject:1593
TEST=angle_unittests, angle_end2end_tests, gpu_unittests
Change-Id: Ieb41f1bf503f815fd0476d2ea045dcb863465254
Reviewed-on: https://chromium-review.googlesource.com/418880
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
acd1898e
|
2017-01-04T10:46:42
|
|
Replace Error(GL_NO_ERROR) with NoError().
In order to make the errors be consistent throughout ANGLE.
BUG=angleproject:1686
Change-Id: I0a2d86091d640aedeac94beae345c1fb6971b00d
Reviewed-on: https://chromium-review.googlesource.com/424835
Commit-Queue: Yunchao He <yunchao.he@intel.com>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
8234e7b3
|
2016-10-07T15:12:10
|
|
Refactor some errors into ANGLE_TRY macros.
BUG=None
Change-Id: I00ff2523995cb49d1af60cae62c2bba0d020eed4
Reviewed-on: https://chromium-review.googlesource.com/395569
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
e02a6834
|
2016-06-09T06:52:41
|
|
Refactor how we signal dependent state changes.
Dependent state changes happen when the user calls TexImage
on a Texture attached to a Framebuffer. The Framebuffer should be
told 'hey, you should know about this'. Other objects also have
dependent relationships, like VertexArrays and Buffers.
This refactoring uses a binding pointer design, similar to the type
'RefCountObject'. This design fixes the need for manual decoupling
when one or the other is destroyed. The pointers are cleaned up in
destructors, and do no-ops when either the source or dest is missing.
Also move these new classes to a location where they are accessible to
the GL layer; they will be important for framebuffer completeness.
BUG=angleproject:1388
Change-Id: I92610acb85dae6f9c009b8f071e121fde53782ae
Reviewed-on: https://chromium-review.googlesource.com/348953
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
3f572680
|
2016-04-26T13:41:36
|
|
Rename gl::VertexArray::Data to gl::VertexArrayState.
BUG=angleproject:1363
Change-Id: I5acf670bd88988941676cc9bc75606d55cca224e
Reviewed-on: https://chromium-review.googlesource.com/340744
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
52b09c2f
|
2016-04-11T14:12:31
|
|
Re-re-land "D3D11: Implement dirty bits for VertexArray11.""
Translated attributes are now stored in the VertexArray11 in a cache,
and only updated when dirty bits change. Currently dynamic attributes
must be re-translated every call, so these are stored in a list and
processed repeatedly.
This skips doing a lot of the VertexDataManager work for vertex
attributes that don't change between draw calls.
Current value attributes, which correspond to disabled attributes that
the program will pulls vertex data from, are owned by the Context, so
these need to be handled outside of the VertexArray11.
Further changes will be necessary to reduce the redundant work we do in
the InputLayoutCache. We shouldn't need to re-check the cache if
nothing relevant changed.
This give about a 23% performance improvement on the draw call
benchmark on my machine.
Re-land with a fix for the start vertex offset.
Re-re-land with a fix for using XFB with deleted buffers.
BUG=angleproject:1327
Change-Id: I0fba49515375c149bbf54d933f8d1f747fbb8158
Reviewed-on: https://chromium-review.googlesource.com/338003
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
53a36004
|
2016-04-08T19:03:18
|
|
Revert "Re-land "D3D11: Implement dirty bits for VertexArray11."""
Seems to make the following dEQP test flaky:
dEQP-GLES3.functional.lifetime.attach.deleted_output.buffer_transform_feedback
doesn't show up on every bot test, but run it a few times and it'll flake. Reverting while I investigate.
BUG=angleproject:1327
This reverts commit 3477f3a62dc139a253a0b361ee138116e9fa881f.
Change-Id: Ic23a392526f5f6e107cf0aa06448389804d6b208
Reviewed-on: https://chromium-review.googlesource.com/337961
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
|
3477f3a6
|
2016-03-29T17:15:29
|
|
Re-land "D3D11: Implement dirty bits for VertexArray11.""
Translated attributes are now stored in the VertexArray11 in a cache,
and only updated when dirty bits change. Currently dynamic attributes
must be re-translated every call, so these are stored in a list and
processed repeatedly.
This skips doing a lot of the VertexDataManager work for vertex
attributes that don't change between draw calls.
Current value attributes, which correspond to disabled attributes that
the program will pulls vertex data from, are owned by the Context, so
these need to be handled outside of the VertexArray11.
Further changes will be necessary to reduce the redundant work we do in
the InputLayoutCache. We shouldn't need to re-check the cache if
nothing relevant changed.
This give about a 23% performance improvement on the draw call
benchmark on my machine.
Re-land with a fix for the start vertex offset.
BUG=angleproject:1327
Change-Id: Ic23e48fb18ed7f29c1999914a2f799ac04aa03e9
Reviewed-on: https://chromium-review.googlesource.com/334225
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|
|
00f394ec
|
2016-03-16T12:09:11
|
|
Revert Dirty bits for VertexArray11
This is a combination of two reverts:
Revert "D3D11: Remove unused mRenderer from VertexArray11."
Revert "D3D11: Implement dirty bits for VertexArray11."
Reverting only the first commit would trigger warnings on the Windows
clang bot.
BUG=594509
BUG=angleproject:1327
This reverts commit fc4712b5ed270436f2993bfda9e916d4f92684a4.
This reverts commit 7d8585b802b7eb741b380bd0d05769281d9507c9.
Change-Id: I612dbba0816d6144f71ce815701c13a798585bc7
Reviewed-on: https://chromium-review.googlesource.com/332989
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
|
|
|
fc4712b5
|
2016-03-10T11:31:25
|
|
D3D11: Remove unused mRenderer from VertexArray11.
This was uncovered by the Clang-win bot, which uses a different set of
warnings from MSVS.
BUG=angleproject:1327
Change-Id: Id298fee2df92752b77bf6e2b93307fa5311cf1fc
Reviewed-on: https://chromium-review.googlesource.com/332035
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
|
|
|
7d8585b8
|
2016-03-09T15:53:12
|
|
D3D11: Implement dirty bits for VertexArray11.
Translated attributes are now stored in the VertexArray11 in a cache,
and only updated when dirty bits change. Currently dynamic attributes
must be re-translated every call, so these are stored in a list and
processed repeatedly.
This skips doing a lot of the VertexDataManager work for vertex
attributes that don't change between draw calls.
Current value attributes, which correspond to disabled attributes that
the program will pulls vertex data from, are owned by the Context, so
these need to be handled outside of the VertexArray11.
Further changes will be necessary to reduce the redundant work we do in
the InputLayoutCache. We shouldn't need to re-check the cache if
nothing relevant changed.
This give about a 23% performance improvement on the draw call
benchmark on my machine.
BUG=angleproject:1327
Change-Id: I7fb944d32ea7e6c78b9e478406bdb7e10a7fc05b
Reviewed-on: https://chromium-review.googlesource.com/330173
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
|