1
0
mirror of https://github.com/doitsujin/dxvk.git synced 2024-12-12 04:08:52 +01:00
Commit Graph

2856 Commits

Author SHA1 Message Date
Philip Rebohle
3063d7fc7c [dxvk] Improve DxvkImageView::handle()
Should fix a silly compiler warning and improves code gen, which
is important since this is *the* most frequently called function
in the backend.
2019-12-11 03:18:39 +01:00
Joshua Ashton
d5d6ae4fe1 [dxvk] Add support for implicit samplers 2019-12-11 03:18:39 +01:00
Joshua Ashton
3fa8691033 [util] Implement simplest ratio helper 2019-12-11 03:18:39 +01:00
Joshua Ashton
16dd1249b7 [util] Add alignDown helper 2019-12-11 03:18:39 +01:00
Joshua Ashton
57b2c02528 [util] Implement lzcnt 2019-12-11 03:18:39 +01:00
Joshua Ashton
aef12f7ee3 [util] Add countof helper 2019-12-11 03:18:39 +01:00
Joshua Ashton
d4cad9055c [util] Implement a clamped version of ComObject, for D3D9
Satisfies a quirk in D3D9, solving an issue in SWTFU.
2019-12-11 03:18:39 +01:00
Joshua Ashton
511ed27733 [util] Add GetPrivateRefCount helper 2019-12-11 03:18:39 +01:00
Joshua Ashton
9280818a57 [util] Implement fclamp for fp special cases 2019-12-11 03:18:39 +01:00
Joshua Ashton
ec197b49f9 [util] Implement some basic vector and matrix math utils 2019-12-11 03:18:39 +01:00
Joshua Ashton
d44707e349 [util] Mark clamp & align as constexpr 2019-12-11 03:18:39 +01:00
Joshua Ashton
a3f74b5eda [util] Implement bit cast. 2019-12-11 03:18:39 +01:00
Philip Rebohle
fc91fe1d34
[dxgi] Add fake implementation of DxgiFactory::EnumWarpAdapter
Apparently FIFA 19 needs this.
2019-12-09 23:27:56 +01:00
Philip Rebohle
945a64252d
[d3d11] Determine exclusive fullscreen mode based on swap chain flags 2019-12-05 13:11:06 +01:00
Philip Rebohle
575a267f07
[vulkan] Support exclusive fullscreen control in presenter 2019-12-05 13:10:11 +01:00
Philip Rebohle
9c26fad40e
[vulkan] Add feature bit for exclusive fullscreen control 2019-12-05 13:05:54 +01:00
Philip Rebohle
dc13f48318
[dxvk] Enable VK_EXT_full_screen_exclusive if available
Also pulls in VK_KHR_get_surface_capabilities2 as a dependency.
2019-12-05 13:05:51 +01:00
Philip Rebohle
07c2de62be
[d3d11] Silence some errors about invalid API usage on deferred contexts
Halo MCC apparently spams calls to GetData on deferred contexts, which
is obviously illegal.
2019-12-04 20:18:51 +01:00
Philip Rebohle
1ae7d4b302
Revert "[dxvk] Don't clear bind mask for unbound UBOs and samplers"
For some bizarre reason, this crashes Star Citizen. The reason might be
that we create multiple pipelines with identical pipelines when using a
state cache, which used to crash some drivers in the past.

Fixes #1266.

This reverts commit 00a064e32b.
2019-12-04 12:24:25 +01:00
Philip Rebohle
0e44bc3068
[meta] Release 1.4.6 2019-12-03 14:52:22 +01:00
Philip Rebohle
27ea176295
[d3d11] Remove d3d11.strictDivision option 2019-11-30 20:39:32 +01:00
Philip Rebohle
3230cec3f3
[util] Enable d3d11.enableRtOutputNanFixup for FFXIV
And disable strict division. The previous workaround broke radial blur.
2019-11-30 20:39:32 +01:00
Philip Rebohle
3b030d9569
[dxbc] Implement workaround to replace NaN render target outputs by zero 2019-11-30 20:39:32 +01:00
Philip Rebohle
2e51e28849
[spirv] Implement opIsNan 2019-11-30 20:39:28 +01:00
Luis Cáceres
abff2afeaf [dxvk] Fix multiple inclusion of dxvk_platform_exts.h
This fixes build failures when using `--unity on` meson parameter
2019-11-28 17:43:08 +01:00
Philip Rebohle
5fe8d823a0
[dxvk] Disable VK_EXT_conditional_rendering
We're currently not using it, and it prevents sharing RenderDoc captures
between different AMD drivers.
2019-11-28 15:41:02 +01:00
Philip Rebohle
4c0e4fba0c
[dxvk] Don't return a value from updateShaderResources
We're checking the bind point all the time anyway, so the previous
design was inconsistent.
2019-11-28 15:25:12 +01:00
Philip Rebohle
2aa1ff414c
[dxvk] Bind descriptor sets at descriptor update time
Gets rid of one 'if' per draw/dispatch and two functions.
2019-11-28 15:17:42 +01:00
Philip Rebohle
49e7df96ec
[dxbc] Emit a spec constant for uniform buffers
This partially reverts commit fd547b666e.

For some reason, not doing so breaks Overwatch (because of course it does).
2019-11-28 00:55:38 +01:00
Philip Rebohle
00a064e32b
[dxvk] Don't clear bind mask for unbound UBOs and samplers
These don't need special treatment, so we really don't need to
recompile the pipeline if they are not bound.
2019-11-28 00:55:36 +01:00
Philip Rebohle
32ac8a8d51
[util] Make Fence and Win32Fence final
Otherwise, Josh will keep complaining about this until the end of time.
2019-11-27 12:31:17 +01:00
Philip Rebohle
c9c6b1886b
[d3d11] Synchronize presentation when destroying swap chain
Otherwise, we might destroy the presenter before the CS thread
actually issues the present operation, and the waitForIdle has
not the desired effect.
2019-11-26 23:51:31 +01:00
Philip Rebohle
9ccad0d197
[dxvk] Use waitForIdle when destroying DXVK device
Otherwise, we might call vkDeviceWaitIdle before all command buffers
have been submitted to the Vulkan queue. Found by inspection.
2019-11-26 22:06:13 +01:00
Philip Rebohle
7446d3c58a
[dxgi] Don't allow changing the FRAME_LATENCY_WAITABLE_OBJECT flag 2019-11-26 16:54:52 +01:00
Philip Rebohle
9785fba66e
[dxgi] Implement IDXGISwapChain2::GetFrameLatencyWaitableObject 2019-11-26 16:54:52 +01:00
Philip Rebohle
65cc8c2b31
[d3d11] Create frame latency event for swap chain
Needed in order to support the DXGI 1.3 frame latency API.
2019-11-26 16:54:52 +01:00
Philip Rebohle
6ebf3e1656
[util] Implement fence capable of signaling win32 events 2019-11-26 16:11:46 +01:00
Philip Rebohle
ef37b5fed6
[dxgi] Implement IDXGISwapChain2::SetFrameLatency 2019-11-26 16:11:46 +01:00
Philip Rebohle
2d1fb52b2f
[util] Reimplement Signal
The new implementation is more akin to D3D12 fences or timeline
semaphores, and should make implementing similar concepts easier.
2019-11-26 16:11:46 +01:00
Philip Rebohle
69bad7bf8c
[d3d11] Move frame latency handling into D3D11SwapChain 2019-11-26 16:11:46 +01:00
Philip Rebohle
a0651392c4
[dxvk] Fix include awfulness 2019-11-26 16:10:58 +01:00
Philip Rebohle
e648d59a10
[util] Fix winelib build
Fixes #1256.
2019-11-26 16:09:35 +01:00
Joshua Ashton
bf14371f9e [util] Wide character conversion changes
Replaces tows with an easier helper that fits in nicer with fixed-size arrays in DXGI, etc.

Prefer UTF8 in tows/fromws.
2019-11-26 01:53:49 +01:00
Joshua Ashton
7e3142b2ed [d3d11] Hook up platform-specific clock
See 89dfa2bc22
2019-11-26 01:52:58 +01:00
Joshua Ashton
c39c3e8dcc [dxvk] Hook up platform-specific clock
See 89dfa2bc22
2019-11-26 01:52:58 +01:00
Joshua Ashton
38f945bf0b [util] Add platform-specific clock implementation
MinGW calls to GetSystemTimeAsFileTime via gettimeofday for std::chrono::high_resolution_clock::now,
This is horribly slow and inaccurate.

There are also issues if MinGW is not built with libstdc++ at the same time that can cause the precision to be bad enough to cause massive hangs.
This effectively works around that as well.

Relevant: ValveSoftware/Proton#3198
2019-11-26 01:52:58 +01:00
Philip Rebohle
25a1e0d355
[d3d11] Actually fix subresources discarded by DiscardView1 2019-11-24 00:34:15 +01:00
Philip Rebohle
53fca5143f
[d3d11] Ignore D3D11_COPY_DISCARD
Various truck simulations are broken and set this on every
CopySubresourceRegion call, which, if we were to implement
DiscardBuffer for non-mappable resources again, would break
them. This flag seemingly has no effect on native D3D11.
2019-11-24 00:08:06 +01:00
Philip Rebohle
a7c21a617c
[d3d11] Overhaul DiscardResource and DiscardView implementations
For host-visible resources, these behave like MAP_DISCARD.
Euro Truck Simulator 2 and friends relies on this (see #1250).
2019-11-23 23:57:07 +01:00
Philip Rebohle
1211bb5e5f
[dxvk] Sort buffer slices before returning them to the buffer
Buffer slices often get shuffled over time due to timing and thread
synchronization, which makes it less and less likely for the dynamic
uniform buffer binding optimization to be effective. Sorting the
slices beforehand addresses the issue and may help CPU performance.
2019-11-23 00:42:54 +01:00
Philip Rebohle
596001c37e
[dxvk] Enable shader-based depth-stencil copies for AMDVLK
Turns out to be slightly faster in practice.
2019-11-22 02:12:02 +01:00
Philip Rebohle
9e965546fc
[meta] Release 1.4.5 2019-11-19 23:36:01 +01:00
Philip Rebohle
6b3d60ab25
[dxvk] Enable asynchronous presentation on all hardware
...and remove the dxvk.asyncPresent option.
2019-11-19 23:34:24 +01:00
Philip Rebohle
ceddbaf7c4
[dxvk] Fix synchronization around vkDeviceWaitIdle
From the spec:
	"Host access to all VkQueue objects created
	from device must be externally synchronized"

We were not doing that, which could cause hangs and
all sorts of issues when recreating the swap chain
on Nvidia GPUs.
2019-11-19 23:34:14 +01:00
Philip Rebohle
4fff2343c1
[dxvk] Fix handling of undefined shader inputs
If the previous stage or the input layout does not define an input,
D3D11 will read zeroes wheras the result is undefined in Vulkan.

Fixes performance degradation and rendering issue in Final Fantasy XV
with the "Geomapping" (terrain tessellation) option enabled.
2019-11-19 15:34:29 +01:00
Philip Rebohle
a1f55330ee
[dxvk] Implement pass to eliminate undefined shader input variables 2019-11-19 15:26:32 +01:00
Philip Rebohle
5a2fd7c71b
[spirv] Add method to retrieve literal string from instruction 2019-11-19 12:49:07 +01:00
Philip Rebohle
8252d1ccd5
[spirv] Add method to erase data from code buffer 2019-11-19 12:17:11 +01:00
Philip Rebohle
9f88249b91
[spirv] Add method to allocate new ID from code buffer. 2019-11-19 12:12:05 +01:00
Philip Rebohle
014798161c
[dxvk] Avoid some unnecessary barriers around render target clears 2019-11-18 19:36:19 +01:00
Philip Rebohle
c7718e5952
[d3d11] Zero-initialize UAV counters
Fixes a hang in Dirt Rally on RADV.
2019-11-18 18:40:39 +01:00
Philip Rebohle
cee7db1c57
[dxvk] Always align texel buffers to at least 16 bytes
Since we don't know the view format in advance, and some
drivers require single-texel alignment.
2019-11-18 13:45:35 +01:00
Philip Rebohle
73a5b33375
[d3d11] Also fix reset counter memory order for deferred context queries 2019-11-15 19:48:41 +01:00
Philip Rebohle
e787077554
[util] Remove traces of the allowMapFlagNoWait option 2019-11-15 19:45:40 +01:00
Philip Rebohle
c24dad75dc
[d3d11] Remove allowMapFlagNoWait option 2019-11-15 11:09:11 +01:00
Philip Rebohle
dbc14fe65c
[d3d11] Fix memory order for query reset counter
Probably doesn't change anything, but let's fix it anyway.
2019-11-14 23:43:58 +01:00
Philip Rebohle
a6483e02fe
[d3d11] Submit stalling event queries before flushing
Otherwise, instead of preventing syncs, this would actually
have the opposite effect.
2019-11-14 22:57:30 +01:00
Philip Rebohle
3457e312b0
[dxvk] Always align index buffer slices to 256 Bytes
To be on the safe side.
2019-11-14 14:32:01 +01:00
Philip Rebohle
420d95e396
[dxvk] Store shader capability information in separate set of flags 2019-11-12 18:05:03 +01:00
Philip Rebohle
1ca235d186
[util] Disable strict DC mode for MGSV again
Not necessary since the resubmitted contexts don't contain mapped buffers.
2019-11-12 00:59:33 +01:00
Philip Rebohle
4f7e7979e3
[util] Add app profile for Metal Gear Solid 5 2019-11-11 23:31:14 +01:00
Philip Rebohle
cc18730967
[dxvk] Enable option to disable OpenVR integration 2019-11-11 23:30:35 +01:00
Philip Rebohle
9f66351b82
[dxvk] Move extension provider list to DxvkInstance 2019-11-11 23:30:07 +01:00
Philip Rebohle
6948d18f5f
[dxvk] Always align vertex buffer slices to 256 bytes
Works around an unknown geometry rendering isssue in Warhammer Chaosbane.
Should probably be investigated at some point.
2019-11-10 22:37:11 +01:00
Philip Rebohle
31baf3529a
[d3d11] Fix uninitialized DSV flags
Fixes #1242.
2019-11-10 15:02:21 +01:00
Philip Rebohle
4b199ef60d
[dxvk] Remove option to disable transfer queue 2019-11-08 11:29:22 +01:00
Philip Rebohle
a74449c367
[dxvk] Remove ability to query instance from adapter
Adapters don't hold a reference to the instance.
2019-11-08 11:28:08 +01:00
Philip Rebohle
77574d9970
[d3d11] Don't query DXVK instance from adapter 2019-11-08 11:25:58 +01:00
Philip Rebohle
b2317cad4d
[dxvk] Pass DXVK instance to DXVK device directly, not through the adapter 2019-11-08 11:17:02 +01:00
Philip Rebohle
212f5ba1f3
[dxgi] Query DXVK instance from DXGI factory, not the adapter 2019-11-08 11:06:15 +01:00
Philip Rebohle
ed8af3ccc4
[dxvk] Document wonky DxvkAdapter ref count behaviour 2019-11-07 21:08:20 +01:00
Philip Rebohle
a0dba6bbf9
[d3d11] Hold reference to DxvkInstance
Hack to keep the instance alive which owns the adapters. Should
fix #1240 for now, but we should fix this properly later on.
2019-11-07 20:21:27 +01:00
Philip Rebohle
b18c50d5ab
[dxvk] Use static variables for extension provider instances
Also, remove redundant interface documentation.
2019-11-07 01:35:10 +01:00
Joshua Ashton
dbf2407fd3 [dxvk] Implement extension provider system
This change introduces a new system for providing extra instance/device extensions.

This consolidates platform-specific and vr-related to its' own thing for potential future cross-platform/native Linux work.
2019-11-07 01:24:52 +01:00
Joshua Ashton
27898ebbfc [build] Simplify and correct errors in options handling
Fix some typos...

D3D10 is dependent on D3D11 therefore DXGI is always dependent on D3D11
2019-11-07 01:24:52 +01:00
Joshua Ashton
4ad99feb73 [build] Demote no frontends/tests to a warning 2019-11-07 01:24:52 +01:00
Philip Rebohle
1d961b7dc3
[d3d11] 'a' is not a very good name for a variable 2019-11-04 13:46:18 +01:00
Philip Rebohle
77a0cb2b19
[d3d11] Don't synchronize with CS thread on present
Instead, submit from the CS thread in order to prevent out-of-order
submissions. Improves minimum FPS in Final Fantasy XIV by 5-10%.
2019-11-02 17:53:17 +01:00
Philip Rebohle
e6c3f0479d
[dxvk] Let device know that async presentation is enabled 2019-11-02 17:53:17 +01:00
Philip Rebohle
d96c22be05
[d3d11] Do not synchronize with CS thread in GetData
Instead, use a counter to determine whether there are any
pending operation for the query on the CS thread.
2019-11-02 17:53:17 +01:00
Philip Rebohle
0924bb469c
[d3d11] Move query state tracking to immediate context implementation 2019-11-02 17:53:17 +01:00
Philip Rebohle
be5dc234c1
[d3d11] Move common Begin/End implementation to immediate context 2019-11-02 13:48:03 +01:00
Philip Rebohle
63dbca82e7
[d3d11] Track used queries in deferred contexts and command lists 2019-11-02 13:48:03 +01:00
Philip Rebohle
0671007437
[d3d11] Only execute Begin for scoped queries 2019-11-02 13:48:03 +01:00
Philip Rebohle
1459f0e852
[util] Fix != and == operator for private/public COM pointers 2019-11-02 13:48:00 +01:00
Philip Rebohle
1780c549e5
[d3d11] Only add storage buffer usage flag to constant buffers if needed
Otherwise, the backend optimization for dynamic uniform buffers will not
kick in.
2019-10-30 10:58:05 +01:00
Philip Rebohle
e95bc3256f
[d3d11] Move handling of constantBufferRangeCheck option to D3D11Options 2019-10-30 10:57:36 +01:00
Philip Rebohle
9e084e63ca
[d3d11] Remove useless members from D3D11DeferredContextMapEntry 2019-10-30 00:44:17 +01:00
Philip Rebohle
9e69a610cb
[dxgi] Use 64-bit integers for refresh rate matching math
Otherwise, there may be integer overflows for certain parameter values.
2019-10-29 08:42:53 +01:00
Philip Rebohle
c4e5323e0e
[dxbc] Remove old constant buffer range check 2019-10-28 17:52:50 +01:00
Philip Rebohle
fd547b666e
[dxbc] Don't emit a spec constant for uniform buffers 2019-10-28 15:44:02 +01:00
Philip Rebohle
3d213efe53
[dxbc] Use SSBOs for dynamically indexed constant buffers if needed
SSBOs are tightly bound-checked on all Nvidia GPUs, so this allows for
a more accurate workaround for games relying on OOB access behaviour.
2019-10-28 15:42:46 +01:00
Philip Rebohle
68760f5b20
[dxbc] Parse dynamically indexed flag for constant buffers 2019-10-28 14:15:44 +01:00
Philip Rebohle
7db98a1aa4
[meta] Release 1.4.4 2019-10-27 17:56:31 +01:00
Alexandr
dacf1ab4d6 [util] Enable constant buffer range check for Saints Row games
Fixes character flickering on Nvidia.
2019-10-27 11:55:48 +01:00
Philip Rebohle
a2b629415e
Revert "[d3d11] Fix ref counting for D3D11CommandList"
This reverts commit 55bae45915.
2019-10-27 11:10:49 +01:00
Philip Rebohle
fc0ede0657
Revert "[d3d11] Recycle command lists from deferred contexts"
This reverts commit 4e3da45fde.

For some reason this caused crashes.
2019-10-27 11:00:59 +01:00
Philip Rebohle
8c34f4ff8a
[dxvk] Validate image layouts in render pass formats read from cache
This will discard invalid cache entries generated by DXVK 1.4.3.
2019-10-27 00:04:57 +02:00
Philip Rebohle
03dc59ef39
[dxvk] Write correct depth-stencil attachment layout to state cache
Fixes a dumb issue where we'd compress layouts with high enum values
to a single byte and incorrectly interpret the result as PREINITIALIZED.
2019-10-26 23:59:11 +02:00
Philip Rebohle
9361f19da6
[dxvk] Fix uninitialized depth attachment reference
Also don't redundantly null color attachment refs for cosmetic reasons.
2019-10-26 23:37:46 +02:00
Philip Rebohle
6888a98c89
[dxbc] Always set geometry shader invocation count
Fixes validation errors when the DXBC shader does not specify
the invocation count itself.
2019-10-26 23:19:59 +02:00
Philip Rebohle
144f4badef
[dxbc] Always allocate at least one input array element
Turns out that shaders with no inputs can still have an input signature,
in which case we were generating a zero-length array which is illegal.
2019-10-26 23:13:59 +02:00
Philip Rebohle
20f79a754b
[d3d11] Adjust buffer memory flags if apitrace is attached
Using cached memory speeds up apitrace significantly as it
reads back mapped memory regions.
2019-10-26 22:56:47 +02:00
Philip Rebohle
44c0f96fc1
[dxvk] Pause transform feedback on buffer updates
Otherwise, we might override a currently bound transform
feedback buffer or counter buffer. Fixes Unity Engine.
2019-10-26 19:57:39 +02:00
Philip Rebohle
3c4a57acc6
[dxvk] Fix bogus xfb buffer update check 2019-10-26 19:40:32 +02:00
Philip Rebohle
0683f4f2c0
[d3d11] Remove D3D11 counter buffer class 2019-10-26 17:44:29 +02:00
Philip Rebohle
e967c52ff7
[d3d11] Allocate predicate buffer per query
Allows us to get rid of the D3D11 counter buffer class.
2019-10-26 17:44:29 +02:00
Philip Rebohle
09f507c284
[d3d11] Allocate xfb counter buffer per buffer
Same idea as with UAV counters, allows for more efficient updates.
2019-10-26 17:44:29 +02:00
Philip Rebohle
191c9644af
[d3d11] Allocate counter buffer per UAV
Might slightly increase memory overhead, however this allows
the backend to execute UAV counter updates more efficiently.
2019-10-26 17:44:29 +02:00
Philip Rebohle
320e0de4a0
[dxvk] Handle xfb barriers in commitGraphicsBarriers
Avoids spilling the render pass when switching xfb buffers.
2019-10-26 17:44:29 +02:00
Philip Rebohle
5b66f1ec0b
[dxvk] Replace buffer in updateBuffer even outside of render passes
Needed to avoid barriers arount D3D11 UAV counter buffer updates.
2019-10-26 17:44:29 +02:00
Philip Rebohle
d1e9e1392d
[dxvk] Lazily allocate slices of small buffers
Reduces memory overhead in case games create small static
buffers. This is somewhat common for index buffers used
to render fullscreen quads or triangles.
2019-10-26 15:24:55 +02:00
Philip Rebohle
7751541662
[dxvk] Don't lock swap lock if not necessary
The swap lock only protects the 'next' free list, which
is not accessed at all when creating a new buffer.
2019-10-26 15:24:54 +02:00
Philip Rebohle
e868f829b5
[dxvk] Pad buffers more tightly
This will allow for smaller buffers to be laid out more efficiently.
2019-10-26 15:24:54 +02:00
Philip Rebohle
83fc15c594
[dxvk] Force-update draw buffer for graphics pipeline barrier checks 2019-10-26 12:41:40 +02:00
Philip Rebohle
c44d30d78b
[dxvk] Consider indirect draw buffer in graphics pipeline barriers 2019-10-26 04:13:05 +02:00
Philip Rebohle
f60d1db1f1
[dxvk] Consider vertex/index buffers in graphics pipeline barriers
Leaving these out was an oversight.
2019-10-26 04:12:32 +02:00
Philip Rebohle
e918104ef3
[dxvk] Introduce checkGfxImageBarrier 2019-10-26 03:56:57 +02:00
Philip Rebohle
7a78852776
[dxvk] Introduce checkGfxBufferBarrier 2019-10-26 03:56:52 +02:00
Philip Rebohle
2b812f07c1
[dxvk] Move finalizeDraw into commitGraphicsState
There is no readon to perform this after the draw anymore.
2019-10-26 03:55:03 +02:00
Philip Rebohle
4e3da45fde
[d3d11] Recycle command lists from deferred contexts
Can save a few memory allocations and deallocations at runtime.
2019-10-25 23:09:21 +02:00
Philip Rebohle
7f66373a69
[util] Add virtual destructor to NoWrapper 2019-10-25 22:08:00 +02:00
Philip Rebohle
55bae45915
[d3d11] Fix ref counting for D3D11CommandList 2019-10-25 21:37:18 +02:00
Philip Rebohle
41f04ffb61
[d3d11] Fix incorrect AddRef return value 2019-10-25 21:36:21 +02:00
Philip Rebohle
f9c2e43ffc
[d3d11] Don't set HOST pipeline stage and access flags for resources
The backend does not and will not use that information at all.
2019-10-24 16:57:06 +02:00
Philip Rebohle
2d6a3396dd
[dxbc] Declare length of shader input array late
Otherwise, we generate out-of-bounds array reads and writes
in broken ENB shaders.
2019-10-23 16:36:52 +02:00
Philip Rebohle
df61a479a2
[spirv] Add 'late' constants
Late constants can be used to reserve a placeholder ID for a constant
before the constant's value is known. The value can be changed later.
Only scalar 32-bit integer and floating point types are supported
as of right now.
2019-10-23 16:09:28 +02:00
Philip Rebohle
59d4556641
[dxbc] Extend scalar constants to vectors if required
Needed for ENB. Fixes #865.
2019-10-23 15:26:22 +02:00
Philip Rebohle
ec8ca8a403
[dxbc] Assume vec4 type for i/o vars not declared in signature
Works around an ENB bug.
2019-10-23 15:26:22 +02:00
Philip Rebohle
3e12a116bb
[dxbc] Declare temp registers on demand
And ignore dcl_temps. Needed to successfully compile some broken
ENB shaders (this does work on Windows too). See #865.
2019-10-23 13:55:04 +02:00
Philip Rebohle
ee77afb6af
[d3d11] Validate shader module capabilities
Only create a shader module if the device actually supports
the required features. Apparently this is needed for some
Unity Engine games.
2019-10-21 12:10:05 +02:00
Philip Rebohle
ddf010479d
[d3d11] Refactor shader module creation 2019-10-21 12:09:53 +02:00
Philip Rebohle
9444162ca6
[util] Enable constant buffer range check for Titan Quest
Reportedly flickers on Nvidia GPUs otherwise.
2019-10-20 13:27:12 +02:00
Philip Rebohle
7a0360d7c2
[dxgi] Remove useless d3d10.enable option 2019-10-19 00:34:51 +02:00
Philip Rebohle
ad3542f4fc
[meta] Release 1.4.3 2019-10-18 17:50:38 +02:00
Philip Rebohle
8211225ad4
[dxbc] Swap OpSelect arguments for Ne/Dne, don't use OpLogicalNot
Saves one instruction.
2019-10-18 03:22:20 +02:00
Philip Rebohle
b756c229a9
[dxvk] Implement more compact state cache data format (v8)
Omits a lot of unnecessary data and considerably reduces
the state cache file size, often by over 80%.
2019-10-18 00:33:50 +02:00
Philip Rebohle
0ac89ccd9e
[util] Add missing != operator to Sha1Hash 2019-10-18 00:33:40 +02:00
Philip Rebohle
4b0a5dffb2
[dxvk] Support component mappings in blitter 2019-10-16 20:00:38 +02:00
Philip Rebohle
d998aaad12
[dxvk] Add some component mapping helpers 2019-10-16 20:00:38 +02:00
Philip Rebohle
0d7f658f96
[dxvk] Implement framebuffer-based blit 2019-10-16 20:00:38 +02:00
Philip Rebohle
6290cfdb48
[dxvk] Add class for framebuffer-based blits 2019-10-16 17:31:40 +02:00
Philip Rebohle
6234a1a6b0
[dxvk] Support arbitrary source texture coordinates in blitter 2019-10-16 02:45:44 +02:00
Philip Rebohle
859ac59e6c
[dxvk] Support multisampled destinations in blitter
Apparently this is required for some D3D9 content.
2019-10-16 02:17:55 +02:00
Philip Rebohle
e747315ba6
[dxvk] Support both linear and nearest samplers in blitter 2019-10-16 02:08:04 +02:00
Philip Rebohle
6ea21d57fe
[dxvk] Factor out mip gen classes and rename them to DxvkMetaBlit*
Since mip map generation is a blit operation, we can make a more
general-purpose blitter and implement mip map generation on top
of that.
2019-10-16 01:52:14 +02:00
Philip Rebohle
d899bd2d76
[d3d11] Remove obsolete DefaultDesc methods from state classes 2019-10-14 05:28:46 +02:00
Philip Rebohle
68c257fc0d
[d3d11] Fix ref counting for state objects bound to a context
Using raw pointers is safe here since the objects never get destroyed
during the lifetime of the context.
2019-10-14 02:06:33 +02:00
Philip Rebohle
1282c2b99e
[d3d11] Fix immediate context reference counting
Fixes various wine test failures.
2019-10-14 01:56:34 +02:00
Philip Rebohle
4d0cc3e24e
[d3d11] Don't subclass state objects from ComObject
These override reference counting, so we shouldn't use ComObject.
2019-10-14 01:44:37 +02:00
Philip Rebohle
c5c43fb2a4
[d3d11] Allow choosing type wrapper for D3D11DeviceChild base class
Allows subclasses to replace ComObject with something else.
2019-10-14 01:44:27 +02:00
Philip Rebohle
dd9a55ecc0
[util] Don't allow multiple inheritance for COM objects
This never made any sense whatsoever in the first place.
2019-10-14 01:42:24 +02:00
Philip Rebohle
c281e76bac
[d3d11] Don't get private references for state objects
Otherwise we will end up deleting the objects even though they
aren't explicitly heap-allocated. Whoops...
2019-10-14 01:39:54 +02:00
Philip Rebohle
9c6209fbf5
[d3d11] Fix reference counting for state objects
Fixes various wine test failures. Also, state objects are now
allocated in the hash map itself rather than a wrapped COM object.
2019-10-14 01:27:59 +02:00
Philip Rebohle
accbc8828c
[d3d11] Return error if no desc is provided for Create*State methods
Fixes wine test failures.
2019-10-14 01:08:31 +02:00
Philip Rebohle
bd58f1a913
[d3d11] Don't sync CS thread if resource to map is already in use
SynchronizeCsThread can only update the in-use state from available
to in-use, so doing this on a resource that is already in-use is not
necessary. May improve performance in combination with DO_NOT_WAIT.
2019-10-14 00:14:00 +02:00
Philip Rebohle
12f0f8b13f
[d3d11] Don't use a state object for default blend state 2019-10-13 23:15:23 +02:00
Philip Rebohle
c30fd8fb97
[d3d11] Don't use a state object for default depth-stencil state 2019-10-13 23:15:23 +02:00
Philip Rebohle
b0231403fe
[d3d11] Don't use a state object for default rasterizer state 2019-10-13 23:15:23 +02:00
Philip Rebohle
a89c662984
[d3d11] Introduce ResetState and use it for ClearState
ClearState gets used a lot in games that use deferred
contexts, so we should make sure it's fast. Since we
apply default state everywhere, there is no need to
perform any expensive RestoreState operations.

Reduces the amount of time spent on ClearState on the
CS thread by ~40%, and by ~90% on the calling thread.
2019-10-13 23:15:23 +02:00
Philip Rebohle
dbe8b09b05
[dxvk] Don't store sampler description in sampler objects
We aren't using this anywhere in the backend or client APIs.
2019-10-13 04:36:33 +02:00
Philip Rebohle
b67e5809ba
[dxvk] Fix buffer usage flags in commitGraphicsBarriers
Only checking for one single usage flag is incorrect since
buffers can have both the storage buffer and storage texel
buffer usage bits set.
2019-10-13 03:18:08 +02:00
Philip Rebohle
950ea21b83
[dxvk] Don't rely on binding mask in commitGraphicsBarriers
We can't actually use that here since we check barriers before
updating shader resources, unlike on the compute path. Check
all resources manually instead.
2019-10-13 02:19:48 +02:00
Philip Rebohle
0e578adcf5
[dxvk] Don't clear binding masks in update*Pipeline
Has no effect anymore since the entire mask gets
overridden in updateShaderResources anyway.
2019-10-13 02:18:09 +02:00
Philip Rebohle
b25ab4155b
[dxvk] Remove outdated comment 2019-10-13 02:01:14 +02:00
Philip Rebohle
0a5b427ded
[dxvk] Add function to insert framebuffer read-back barriers
Required for some D3D9 content.
2019-10-13 02:00:10 +02:00
Philip Rebohle
41cb5ab5f7
[dxvk] Add dependency flag parameter to emitMemoryBarrier
Will be needed for framebuffer-local barriers inside render passes.
2019-10-13 01:51:29 +02:00
Philip Rebohle
8446c28de1
[dxvk] Remove spill parameter from bindRenderTargets
No longer needed.
2019-10-13 01:44:31 +02:00
Philip Rebohle
102a18060e
[d3d11] Don't spill render pass when rebinding render targets
The backend handles this now, so it's no longer necessary.
2019-10-13 01:44:31 +02:00
Philip Rebohle
4360021539
[dxvk] Implement accurate barrier tracking for draws with side effects
Similar to how this is handled for compute shaders, with some caveats:
- The barriers are never actually emitted, only tracked. Spilling the
  render pass will reset the barriers as it acts as a full barrier.
- Doing this for all draws and all resources would be prohibitively
  expensive, so whenever switching between pipelines with side effects
  and pipelines without side effects, we'll spill the render pass.
2019-10-13 01:44:28 +02:00
Philip Rebohle
baf81473a6
[dxvk] Introduce HasStorageDescriptors to signify pipeline side effects 2019-10-13 01:43:18 +02:00
Philip Rebohle
c5676d3108
[dxvk] Remove incorrect framebuffer-space barriers
We're technically required to use VK_DEPENDENCY_BY_REGION_BIT,
but that isn't actually good enough.
2019-10-12 23:54:54 +02:00
Philip Rebohle
803ec3542d
[dxbc] Fix sample positions
From the D3D11.3 functional specification:
"The sample position is relative to the pixel's center"

Fixes wine test failures.
2019-10-11 18:41:10 +02:00
Philip Rebohle
3a39027987
[d3d11] Add more rigid validation for buffer creation
Fixes several wine test failures.
2019-10-11 17:32:46 +02:00
Philip Rebohle
762df0bedf
[d3d11] Change ValidateBufferProperties to NormalizeBufferProperties 2019-10-11 17:23:02 +02:00
Philip Rebohle
409991b9db
[d3d10] Fix reported bind flags for D3D10 resources
Turns out that the UAV bind flag remains set in the D3D10 description,
even though D3D10 does not support the feature. Fixes wine test failures.
2019-10-11 17:15:33 +02:00
Philip Rebohle
127e037627
[d3d11] Validate texture array size
Fixes a wine test failure and possibly prevents invalid Vulkan API usage.
2019-10-11 17:11:01 +02:00
Philip Rebohle
554b77b47a
[d3d11] Initialize feature level to zero in D3D11CreateDevice
Fixes a wine test failure.
2019-10-11 17:01:23 +02:00
Philip Rebohle
d8c3002b92
[dxvk] Don't use dynamic storage buffers
Doesn't really help in pracrice, but getting rid of them reduces
the number of dynamic offsets we have to update per draw/dispatch.
2019-10-11 14:33:45 +02:00
Philip Rebohle
0068740341
[dxvk] Optimize invalidateBuffer for use with uniform buffers
Since this is by far the most common use case for this
method, an early-out path helps save a bit of CPU time.

Ditch dynamic descriptor set offsets for storage buffers
at the same time since it does not seem to benefit any
real-world applications.
2019-10-11 14:16:47 +02:00
Philip Rebohle
bd5630439e
[dxvk] Inline freeBufferSlice implementation
May save a cycle or two in invalidateBuffer.
2019-10-11 13:06:55 +02:00
Philip Rebohle
594f04d4ed
[dxvk] Dirty framebuffer in beginRecording
This guarantees that the framebuffer is never null.
2019-10-11 12:19:45 +02:00
Philip Rebohle
48b3b3ee85
[dxvk] Use float format for sampled unbound image views
Fixes validation errors in many cases if a texture is not bound.
2019-10-11 02:58:39 +02:00
Philip Rebohle
e615416b31
[dxvk] Fix tessellation validation
Prevents crashes in case an app tries to use tessellation with
an incorrect primitive topology.
2019-10-11 00:04:06 +02:00
Philip Rebohle
2c974cbe1e
[dxvk] Validate graphics state in commitGraphicsState
Same idea as for the related compute work.
2019-10-10 23:56:54 +02:00
Philip Rebohle
649c8d10d6
[dxvk] Validate compute state in commitComputePipelines
This way we can implement early-out for invalid dispatch calls
and save a few validation checks later in the pipeline.
2019-10-10 23:56:54 +02:00
Philip Rebohle
63cc8cdd35
[dxvk] Move descriptor set updates to updateShaderResources
Saves two state flags and allows us to move the descriptor info
array into the function itself.
2019-10-10 22:48:06 +02:00
Philip Rebohle
7df3b409c3
[dxvk] Move dynamic descriptor offset array to stack
There's absolutely no reason to store it in the object.
2019-10-10 22:19:31 +02:00
Philip Rebohle
a743ba6531
[dxvk] Use memcmp replacement for pipeline state lookup
Measured to be over twice as fast as memcmp on Ryzen for the
512-byte graphics pipeline state struct, achieving two cycles
per iteration.
2019-10-07 22:01:48 +02:00
Philip Rebohle
5cb7be2454
[dxvk] Implement state cache v6 -> v7 conversion 2019-10-07 22:01:48 +02:00
Philip Rebohle
345e263a3b
[dxvk] Increase maximum spec constant count to 12
Fills some padding created by the 32-byte alignment.
2019-10-07 22:01:48 +02:00
Philip Rebohle
782b021690
[dxvk] Move compute pipeline state to dxvk_graphics_state.h 2019-10-07 22:01:48 +02:00
Philip Rebohle
e086db5ce5
[dxvk] Use new struct for specialization constant info
Stylistic reasons.
2019-10-07 22:01:47 +02:00
Philip Rebohle
b2087b2e7e
[dxvk] Use packed blend state 2019-10-07 22:01:47 +02:00
Philip Rebohle
079b480602
[dxvk] Use packed depth-stencil state 2019-10-07 22:01:47 +02:00
Philip Rebohle
a933bec72c
[dxvk] Use packed multisample state 2019-10-07 22:01:47 +02:00
Philip Rebohle
09c813c934
[dxvk] Use packed rasterizer state 2019-10-07 22:01:47 +02:00
Philip Rebohle
a5ab88d8f2
[dxvk] Use packed input layout state
Also move attribute and binding descriptions to the end
so that comparisons are more likely to fail early.
2019-10-07 22:01:47 +02:00
Philip Rebohle
c0ae4e38eb
[dxvk] Use packed input assembly state 2019-10-07 22:01:47 +02:00
Philip Rebohle
8d09aa12da
[dxvk] Move DxvkGraphicsPipelineStateInfo to new file
We're going to put a lot of code here, so try to keep things clean.
2019-10-07 22:01:47 +02:00
Philip Rebohle
9dad44a6b1
[dxvk] Prepare for pipeline state changes and bump cache format to v7
One of the changes includes hard-coding certain constants into the old
pipeline state structs, since changing the constants would invalidate
any old state cache and making the conversion pointless.
2019-10-07 22:01:47 +02:00
Philip Rebohle
32dff89b2d
[dxvk] Don't use alignas(16) for DxvkMetaClearArgs members
Instead, pad members explicitly. This is necessary because GCC
makes incorrect assumptions about stack alignment on 32-bit.
2019-10-07 22:01:47 +02:00
Joshua Ashton
ca634ec484 [dxvk] Add missing {} to border color list
Fixes a warning when compiling with Clang
2019-10-06 00:28:19 +02:00
Joshua Ashton
e8ee7a0790 [util] Fix Sha1 dword extraction
Previously we were getting incorrect values here.

The u suffix is no longer necessary, which was originally there to work around a MSVC compiler warning which now doesn't happen under the new code 🤷‍♀
2019-10-06 00:28:19 +02:00
Joshua Ashton
19fa1c405c [d3d10] Remove unused copies of device ptr in D3D10 wrappers
Silences a warning when building with Clang, and removes duplicate unused data.
2019-10-06 00:28:19 +02:00
Joshua Ashton
6d0757520b [dxgi] Initialize size of DEVMODEW structure
The documentation says we should do this.
2019-10-06 00:28:19 +02:00
Philip Rebohle
1117daec24 [dxvk] Don't redundantly clear unused vertex strides to zero
We're already doing this when setting the input layout, and we
don't touch the strides in other parts of the code. May save
a handful of CPU cycles.
2019-10-05 16:52:26 +02:00
Philip Rebohle
70ef456911
[meta] Release 1.4.2 2019-10-04 21:07:36 +02:00
Philip Rebohle
02b79fb738
[util] Report Nvidia GPUs for Modern Warfare Remastered 2019-10-04 18:35:36 +02:00
Philip Rebohle
50349d1bfe
[d3d11] Add missing break; in CreateSurface
Found by a frog. Initial testing did not check return values.
2019-10-02 12:06:51 +02:00
Philip Rebohle
85581bdace
[util] Move GetAdapterLUID out of DXGI internals 2019-10-02 08:42:22 +02:00
Philip Rebohle
8f00af556e
[d3d11] Implement CreateSurface for D3D11DXGIDevice
Documentation for this method is really poor so we'll just try to
make pretty much everything work. Fixes various wine tests.
2019-10-02 02:29:06 +02:00
Philip Rebohle
1366fce0f8
[dxgi] Generate adapter LUIDs if Vulkan does not provide them
Needed because winevulkan does not provide adapter LUIDs.
Fixes various wine test failures, and will be required to
get D3D12 (vkd3d) to work on top of our DXGI implementation.
2019-10-02 01:07:18 +02:00
Philip Rebohle
73b9846c4f
[dxgi] Fix some IDXGIAdapter return codes
Fixes various wine test failures.
2019-10-02 00:33:06 +02:00
Philip Rebohle
d90d450618
[dxgi] Handle invalid arguments passed to Present1
fixes a wine test failure.
2019-10-02 00:19:41 +02:00
Philip Rebohle
f58c1c2d1d
[dxgi] Fix SetFullscreenState behaviour with invalid usage
Fixes various wine test failures.
2019-10-02 00:01:26 +02:00
Philip Rebohle
60ff2b8977
[dxgi] Fix GetContainingOutput and GetFullscreenState output pointer
We're supposed to keep a reference to the output around and return
that for fullscreen swap chains. Fixes various wine test failures.
2019-10-02 00:00:37 +02:00
Philip Rebohle
8d8f4b9200
[d3d11] Add support for DXGI_SWAP_CHAIN_FLAG_GDI_COMPATIBLE
Trivial since we treat back buffers as regular D3D textures.
Fixes a wine test failure.
2019-10-01 23:37:54 +02:00
Philip Rebohle
aa35cdbb18
[d3d11] Fix return code on swap chain creation failure
Fixes a wine test failure.
2019-10-01 23:37:54 +02:00
Philip Rebohle
0f15c60192
[d3d11] Use private reference to D3D11 device in D3D11SwapChain
Fixes a wine test failure.
2019-10-01 22:26:07 +02:00
Philip Rebohle
0436e6ddce
[dxgi] Fix some return codes in DxgiSwapChain
Fixes some wine test failures.
2019-10-01 22:25:37 +02:00
Philip Rebohle
0d1048fa87
[d3d11] Fix error handling in SetMaximumFrameLatency
Fixes a wine test failure.
2019-10-01 22:12:23 +02:00
Philip Rebohle
2cb9d34f8f
[dxgi] Fix QueryVideoMemoryInfo return values
Fixes a wine test failure.
2019-10-01 22:12:19 +02:00
Philip Rebohle
1e00f8046c
[dxgi] Fix CheckInterfaceSupport UMD version writeback
Fixes a wine test failure.
2019-10-01 22:12:16 +02:00
Philip Rebohle
cd82669a85
[dxgi] Fix GetDisplayModeList1 output for DXGI_FORMAT_UNKNOWN
Fixes a wine test failure.
2019-10-01 22:12:09 +02:00
Philip Rebohle
f5515075f2
[dxgi] Fix GetDisplayModeList behaviour with a zero mode count
Allocate at least one array element so that we don't accidentally
treat the mode list like it was NULL. Fixes a wine test failure.
2019-10-01 22:12:08 +02:00
Philip Rebohle
2f4e4abcac
[dxgi] Fix CheckInterfaceSupport for IDXGIDevice
On Windows, this succeeds and reports the D3D9 driver version to the
application.
2019-10-01 15:06:50 +02:00
Philip Rebohle
ab51aac6d7 [dxvk] Implement context-local lookup cache for pipeline objects
On hit, looking up a pipeline becomes an extremly cheap O(1) operation,
avoiding a mutex lock and an expensive std::unordered_map lookup. This
works because we don't ever destroy pipeline objects.
2019-09-30 03:42:20 +02:00
Philip Rebohle
35a8cedbc2 [dxvk] Move shader set hash/eq functions into respective structs
Makes it easier to use them outside the pipeline manager.
2019-09-30 03:42:20 +02:00
Philip Rebohle
ba5f319809 [dxvk] Use new shader lookup hash in pipeline manager 2019-09-30 02:09:09 +02:00
Philip Rebohle
5115a42b08 [dxvk] Expose key-based lookup hash in DxvkShader 2019-09-30 02:08:09 +02:00
Philip Rebohle
54ff7bf769 [dxvk] Add methods to retrieve shaders from pipeline 2019-09-30 01:54:49 +02:00
Philip Rebohle
59d4e8a36a [dxbc] Fix NaN handling in not-equal comparison
Ne and Dne are unordered in DXBC. Avoid FUnordNotEqual to
avoid potential issues with drivers and debugging tools.
2019-09-30 00:19:47 +02:00
Philip Rebohle
d128d776ad
[dxvk] Re-introduce state check for updateFramebuffer
Turns out we broke batching render target clears earlier.
2019-09-28 03:29:52 +02:00
Philip Rebohle
6b5d01c934
[meta] Release 1.4.1 2019-09-27 23:03:23 +02:00
Philip Rebohle
2e6da26ff0
[dxvk] Spoof Nvidia GPU for HITMAN 2
This game requires a functioning AGS implementation when it detects
an AMD GPU and complains about missing D3D11 support otherwise.
2019-09-26 19:27:19 +02:00
Philip Rebohle
a920a7275b
[dxvk] Eliminate some redundant state checks 2019-09-26 14:25:56 +02:00
Philip Rebohle
7cdc402a58
[dxvk] Introduce finalizeDraw
May improve code gen by reducing function call overhead, in a
similar way as the duplicated checks in commitGraphicsState do.
2019-09-26 01:31:51 +02:00
Philip Rebohle
187748e4d3
[d3d11] Add a comment explaining ID3D11Predicate/ID3D11Query hack 2019-09-25 22:18:08 +02:00
Philip Rebohle
53fe3423e2
[d3d11] Fix GetPredication 2019-09-25 20:55:43 +02:00