It's redundant to add an UNDEFINED transition here for committed
resources. We need it for sparse and placed resources to handle aliasing
rules, but that's it.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
On some implementations, it doesn't matter for performance what we use,
and we can avoid a lot of ugly barriers this way.
Opt-in to use this extensions on GPUs we know handles it well,
otherwise, keep using the tracking paths.
With VK_KHR_dynamic_rendering, this is now feasible to do since we no longer
have to deal with shenanigans related to VkRenderPass layouts and
complicated compatibility rules.
To make this work with the existing framework, just need to consider
that GENERAL can be a common layout alongside DEPTH_STENCIL_OPTIMAL,
which are both common layouts that do not need to be tracked at all.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
When performing a decay of a DSV resource, make sure to transition all
subresources, not just the particular aspect being transitioned.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
We require separate DS layouts.
Fixes validation errors where we transition from read-only, but our
neighbor aspect might have been optimal.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
These only existed for VRS attachment, which is no longer
necessary with VK_KHR_dynamic_rendering.
Signed-off-by: Philip Rebohle <philip.rebohle@tu-dortmund.de>
Proves out the viability of this style of implementation. Ideally we'd
have a more officially sanctioned way of doing similar things later :)
Unfortunately, the overhead removal is too great to ignore on target
platform. Makes use of a private (reserved) extension for now ...
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Elden Ring in particular spam frees and allocates command pools despite
this being a very bad idea.
Add a simple 8-entry cache which seems to take care of it.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Tune memory layout so that we can deduce various information without
making a single pointer dereference:
- d3d12_descriptor_heap*
- heap offset
- Pointer to various side data structures we need to keep around.
Instead of having one big 64 byte data structure with tons of padding,
tune it down to 32 + 8 bytes per descriptor of extra dummy data.
To make all of this work, use a somewhat clever encoding scheme for CPU
VA where lower bits store number of active bits used to encode
descriptor offset. From there, we can mask away bits to recover
d3d12_descriptor_heap. Metadata is stored inline in one big allocation,
and we can just offset from there based on extracted log2i_ceil(descriptor count).
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Forgot to offset buffer offset. Fun!
Found when bumping VA allocation limit to 2 MiB instead of 1 MiB.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
This became basically a rewrite in the end, and it got too awkward to
split these commits in any meaningful way.
The goals here were primarily to:
- Support serializing SPIR-V and load SPIR-V.
To do this robustly requires a lot more validation and checks to make
sure end up compiling the same SPIR-V that we load from cache.
This is critical for performance when games have primed their pipeline
libraries and expect that loading a PSO should be fast. Without this,
we will hit vkd3d-shader for every PSO, causing very long load times.
- Implement the required validation for mismatched PSO descriptions.
- Rewrite the binary layout of the pipeline library for flexibility
concerns and performance.
If the pipeline library is mmap-ed from disk - which appears to be
the intended use - we only need to scan through the TOC to fully parse
the library contents.
From a flexibility concern, a blob needs to support inlined data,
but a library can use referential links. We introduce separate
hashmaps which store deduplicated SPIR-V and pipeline cache blobs,
which significantly drop memory and storage requirements.
For future improvements, it should be fairly easy to add information
which lets us avoid SPIR-V or pipeline cache data altogether if
relevant changes to Vulkan/drivers are made.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
We did not test the scenario where we first render with depth enabled,
and then bind a NULL DSV with the same pipeline.
Also fix issues if we bind NULL RTVs with same pipeline bound.
Fixes crash in Guardians of the Galaxy.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
This is benign and easily gets spammed a TON.
We will warn if an indexed draw is actually made like this.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
UBSAN found a bug here since we store RTV descriptors inline, the
compiler can assume the pointer is 64 byte aligned.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
D3D12 expects drivers to implicitly synchronize transfer operations,
since there is no TRANSFER barrier ala UAV barriers.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
In DEATHLOOP, there is a render pass which renders out a simple image,
which is then directly followed by a compute dispatch, reading that
image. The image is still in RENDER_TARGET state, and color buffers are
*not* flushed properly on at least RADV, manifesting as a very
distracting glitch pattern. This is a game bug, but for the time being,
we have to workaround it, *sigh*.
For a simple workaround, we can detect patterns where we see these
events in succession:
- Color RT is started
- StateBefore == RENDER_TARGET is not observed
- Dispatch()
In particular, when entering the options menu, highly distracting
glitches are observed in the background.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
HZD issues an aliasing barrier for an alias of a resource that it
still needs.
Because D3D12 requires you to call DiscardResource or a full resource
clear/copy, we can just rely on those to do the actual image layout
transition and treat the aliasing barrier as a pure sync + flush.
This behavior is also observed in a test case where D3D12 drivers
do not seem to discard / fast-clear anything in an aliasing barrier.
Signed-off-by: Robin Kertels <robin.kertels@gmail.com>
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Co-authored-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Mesa RADV translates these legacy entrypoints to the 2 variants. Using
them directly will cost a bit less CPU cycles.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Otherwise, we accidentally merge ranges from different pools if
the indices happen to align.
Signed-off-by: Philip Rebohle <philip.rebohle@tu-dortmund.de>
Guardians of the Galaxy hits this case. Fallback is to disable depth
attachment entirely in a fallback pipeline.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
This function fails if the counter overflows.
CP77 hits this case a lot and we should just warn the specific failure
instead of a random error.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Potentially reduces the size of the query map, and makes each entry
versioned so that we no longer have to clear the entire map for multiple
dispatches even if it is sparsely populated.
Signed-off-by: Philip Rebohle <philip.rebohle@tu-dortmund.de>
Need to use fallback pipeline system here.
Keep track of active masks for PSO and current render target.
The intersection of those sets are the attachments which should be
active in the render pass.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
We will need to consider some form of budgeting, so make sure that all
allocation and freeing is done in a central place.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Emitting render pass clears while we're in the process of starting
a render pass overrides dsv layout tracking info.
Signed-off-by: Philip Rebohle <philip.rebohle@tu-dortmund.de>
D3D12 validation layer errors out, so unless we can prove that specific
behavior is relied upon, we should be okay to just ignore.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Goal here is to avoid unnecessary image layout transitions when render
passes toggle depth-stencil PSO states. Since we cannot know which
states a resource is in, we have to be conservative, and assume that
shader reads *could* happen.
The best effort we can do is to detect when writes happen to a DSV
resource. In this scenario, we can deduce that the aspect cannot be
read, since DEPTH_WRITE | RESOURCE state is not allowed.
To make the tracking somewhat sane, we only promote to OPTIMAL if an
entire image's worth of subresources for a given aspect is transitioned.
The common case for depth-stencil images is 1 mip / 1 layer anyways.
Some other changes are required here:
- Instead of common_layout for the depth image, we need to consult the
command list, which might promote the layout to optimal.
- We make use of render pass compatibility rules which state that we can
change attachment reference layouts as well as initial/finalLayout.
To make this change, a pipeline will fill in a
vkd3d_render_pass_compat struct.
- A command list has a dsv_plane_optimal_mask which keeps track
of the plane aspects we have promoted to OPTIMAL, and we know cannot
be read by shaders.
The desired optimal mask is (existing optimal | PSO write).
The initial existing optimal is inherited from the command list's
tracker.
- RTV/DSV/views no longer keep track of VkImageLayout. This is
unnecessary since we always deduce image layout based on context.
Overall, this shows a massive gain in HZD benchmark (RADV, 1440p ultimate, ~16% FPS on RX 6800).
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Idea is to keep track of scenarios where we know a resource's aspect is
known to be in a OPTIMAL state. Based on this, we can override the image
layout from the common_layout in order to avoid unnecessary full
barriers.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
For copies, we can always use the intended aspects, since we have
separate DS layouts now.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
When clearing a DSV, we must get aliasing guarantees, so we must
transition away from UNDEFINED. This is only possible when using
separate_ds_layouts and for render pass clears we need to use
renderpass2 mechanisms to do this.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
GravityBench ends up using ClearView with too large dimensions.
This is a validation error in Vulkan, so just clamp the extents.
To make full rect detection a bit more robust, do a range check instead
of memcmp().
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
If we're doing a layout transition of depth-stencil aspects, we need to ensure all potential
accesses are made visible.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
- Honor resource barriers for resource states which cannot automatically
decay or promote. This includes COLOR_ATTACHMENT, UNORDERED_ACCESS and
VRS image. If SIMULTANEOUS_ACCESS is used, we can still promote, and
we handle that by setting common layout to GENERAL for these resources.
- Avoid redundant barriers in render passes since normal resource
barriers will always make sure we are already in
COLOR_ATTACHMENT_OPTIMAL.
- Do not force GENERAL layout if resource has UNORDERED_ACCESS flag set.
As this is not a promotable state, we have to explicitly transition
into it. I tested this on validation layers, where even COMMON state
refuses to promote to UAV state. The exception here of course is
SIMULTANOUS_ACCESS, but we handle that properly now.
- Verify that UAV or SIMULTANEOUS access is not used together with DSV
state. This is explicitly banned in the API docs.
- Actually emit image barriers. Batch the image transitions as that's
what D3D12 docs encourage app developers to do, and it also expects
that drivers can optimize this. Ensure that we respect the in-order
resource barrier rules by splitting batches if there are overlaps in
the transitions.
- Ensure that correct image layout is used when clearing a suspended
render pass attachment.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Avoid using the separate layouts if we're only using formats with one
aspects. This makes it more likely to match layouts with common layout,
and we can avoid awkward transition barriers.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Fixes test TODOs. Apparently Vulkan drivers can saturate here, which
caused the TODO to appear, at least on AMD Windows.
Signed-off-by: Hans-Kristian Arntzen <post@arntzen-software.no>
Hans-Kristian Arntzen
Philip Rebohle
Robin Kertels
Samuel Pitoiset
ifedorov
Joshua Ashton
Georg Lehmann
David McCloskey
rochaudhari