Prior to this change, it's possible for m_wake_me_up_again to be used
while it's in an uninitialized state from the exposed API.
e.g.
- Using SetEnable after construction would perform an uninitialized read.
- Using PushEvent would perform an uninitialized read by way of operator |=.
internally, an uninitialized read can happen if PullEventsInternal() is
executed before other functions.
Just to avoid the whole possibility of performing uninitialized reads,
we just give the class member a default value of false.
This way, if we load a UID cache where the data was incomplete (e.g.
Dolphin crashed), we don't lose the existing UIDs which were previously
at the beginning.
This will create a merge conflict if two PRs try to increment the
cache version at the same time, which makes it noticeable that the
PR that gets merged last needs to increment the cache version again.
We already use this for savestates and the game list cache.
Minimizes repetition.
std::minmax_element can be used for the 256 * 2 case, as it's only performing byte comparisons
and thus, there will always be an element smaller than 0xffff, so it doesn't need to be included
in the set of compared values.
Skip ubershader mode works the same as hybrid ubershaders in that the
shaders are compiled asynchronously. However, instead of using the
ubershader to draw the object, it skips it entirely until the
specialized shader is made available.
This mode will likely result in broken effects where a game creates an
EFB copy, and does not redraw it every frame. Therefore, it is not a
recommended option, however, it may result in better performance on
low-end systems.
This wouldn't be much of a data reader if it can't access the
read-only data pointer in read-only contexts. Especially if it
can get a writable equivalent in contexts that aren't read-only.
It's questionable to not return a reference to the instance being
assigned to. It's also quite misleading in terms of expected behavior
relative to everything else. This fixes it to make it consistent with
other classes.
Allows the default constructor to be defaulted and ensures the default
values are associated with the member variables directly.
Also corrects a prefixed underscore in the two parameter constructor.
Given that this only contains functions from the VideoBackendBase class,
it makes more sense to move these to the relevant cpp file to keep them
all together.
This is only ever memset to zero and never used again.
This also gets rid of an instance of undefined behavior considering the
draft standard for C++17 (N4659) states at [dcl.type.cv] paragraph 5:
"
The semantics of an access through a volatile glvalue are implementation-defined.
If an attempt is made to access an object defined with a volatile-qualified type
through the use of a non-volatile glvalue, the behavior is undefined.
"
Currently, when immediately compile shaders is not enabled, the
ubershaders will be placed before any specialized shaders in the compile
queue in hybrid ubershaders mode. This means that Dolphin could
potentially use the ubershaders for a longer time than it would have if
we blocked startup until all shaders were compiled, leading to a drop in
performance.
- In D3D, shaders could be compiled on the main thread, blocking
startup.
- Reduced the latency between a pipeline being requested and used in all
backends in hybrid ubershader mode, when no shader stages were present.
- Fixed a case where async compilation could cause the same UID to be
appended multiple times to the UID cache.
- Fix incorrect number of threads being used when immediately compile
shaders was enabled.
Lowest hanging fruit I could find with a profiler.
Not sure this stuff actually needs to be done, but assuming it is, why
not do it quickly? 10x faster, goes from 1% CPU to 0.09%.
This enables shaders to be compiled while the game is starting, instead
of blocking startup. If a shader is needed before it is compiled,
emulation will block.
As these are stored in a map, operator< will become a hot function when
doing lookups, which happen every frame. std::tie generated a rather
large function here with quite a few branches.
tl;dr: This PR speedups dolphin on mobiles with the Mali GPU and ES 3.2
drivers by a factor of 10 by using the method with the biggest overhead.
Please keep care not to buy this shit!
The ARM driver team seems to care very well about their customers. But
bad luck, users and open source developers are *not* their customers. So
even device-independent feature requests are just ignored for *years*:
https://community.arm.com/graphics/f/discussions/4645/gl_ext_buffer_storage-support
The bad point, they neither implement any of the other common ways to
stream dynamic content in unextented GL:
- They just ignore the GL_MAP_UNSYNCHRONIZED_BIT flag
- They don't support on-device buffer updates and just stall with
glBufferSubData
It seems like no benchmark is using any dynamic content - and like no
customer cares about anything but benchmarks, or users...
We have a flag to disable the glBufferSubData way, this PR adds the flag
to also disable the unsychronized mapping way. The second one is
available since their ES 3.2 update, but slow as hell.
So how to continue? The last remaining technical way to stream dynamic
content at all is to alloc a new buffer per draw call with glBufferData.
This is very gross, but still a factor 10 speedup compared to stalling
the GPU. Small tests shows that you can expect another 3-5 times speedup
with EXT_buffer_data, so Mali would be on pair with Adreno here. So if
you have bought such a device unfortunately, please try to make noise on
your vendor forums/support and ask for this extension. If you are going
to buy a new mobile, I'd recormend to avoid *any* mobile with a Mali GPU
in it.
We now differentiate between a resize event and surface change/destroyed
event, reducing the overhead for resizes in the Vulkan backend. It is
also now now safe to change the surface multiple times if the video thread
is lagging behind.
The option is named DisableCopyToVRAM under the Hacks section in
GFX.ini. It is intentionally not exposed to the GUI, as users should not
need to use it under normal circumstances. The main use is debugging
issues in the EFB-to-RAM shaders.
The console appears to behave against standard IEEE754 specification
here, in particular around how NaNs are handled. NaNs appear to have no
effect on the result, and are treated the same as positive or negative
infinity, based on the sign bit.
However, when the result would be NaN (inf - inf, or (-inf) - (-inf)),
this results in a completely fogged color, or unfogged color
respectively. We handle this by returning a constant zero for the A
varaible, and positive or negative infinity for C depending on the sign
bits of the A and C registers. This ensures that no NaN value is passed
to the GPU in the first place, and that the result of the fog
calculation cannot be NaN.
Otherwise we might get UB if the value we cast is larger than the
max value of the underlying type that the compiled picked for the enum.
I haven't done any extensive check through Dolphin to find cases
of this, I'm just fixing the cases I already know of.
Tested on a linux Intel Skylake integrated graphics with
blend_func_extended force-disabled, as it's the only platform I have
that doesn't crash with ubershaders and supports fb_fetch
It seems it doesn't like modifying inout variables in place - so instead
use a temporary for ocol0/ocol1 and only write them once at the end of
the shader
GLES doesn't support C-style array initialisers, so stuff like:
Type var[2] = {
VALUE_A,
VALUE_B
};
isn't supported in GLES (it was added in
GL_ARB_shading_language_420pack).
The texture conversion shader used this, so would fail to compile on
GLES.
HLSL does not define roundEven(), only round(). This means that the
output may differ slightly for OpenGL vs Direct3D. However, it ensures
consistency across OpenGL drivers, as round() in GLSL can go either way.
This fixes the rendering of the scan visor in Metroid Prime 2: Echoes,
as seen in https://fifoci.dolphin-emu.org/dff/mp2-scanner/
The alpha channel was off-by-one on Ivy Bridge due to the rounding
after multiplication with colmat. This commit removes this matrix
altogether in most cases, making them simple GLSL swizzles.
This will generate one shader per copy format. For now, it is the same
shader with the colmat hard coded. So it should already improve the GPU
performance a bit, but a rewrite of the shader generator is suggested.
Half of the patch is done by linkmauve1:
VideoCommon: Reorganise the shader writes.
Also skips swapping the window system buffers in headless mode, as there
may not be a surface which can be swapped in the first place. Instead,
we call glFlush() at the end of a frame in this case.
Cel-damage uses the color from the lighting stage of the vertex pipeline
as texture coordinates, but sets numColorChans to zero.
We now calculate the colors in all cases, but override the color before
writing it from the vertex shader if numColorChans is set to a lower value.
This is already initialized in the class definition. This would
previously cause a -Wreorder warning on macOS, as m_config is
defined after m_currently_mapped.
We need this because VS currently doesn't consider
std::is_trivially_copyable<typename
std::remove_volatile<SCPFifoStruct>::type>::value
to be true and because no compiler should consider it
to be true if we replace the volatiles with atomics.
Some lines of code in Dolphin just plainly grabbed the value of
g_ActiveConfig.iEFBScale, which resulted in Auto being treated as
0x rather than the actual automatically selected scale.
There was a race condition between the video thread and the host thread,
if corrections need to be made by VerifyValidity(). Briefly, the config
will contain invalid values. Instead, pause emulation first, which will
flush the video thread, update the config and correct it, then resume
emulation, after which the video thread will detect the config has
changed and act accordingly.
Calling vkCmdClearAttachments with a partial rect, or specifying a
render area in a render pass with the load op set to clear can cause the
GPU to lock up, or raise a bounds violation. This only occurs on MSAA
framebuffers, and it seems when there are multiple clears in a single
command buffer. Worked around by back to the slow path (drawing quads)
when MSAA is enabled.
Currently, this is only the logic op bit, but this will be extended to
the framebuffer fetch/blend modes. In the future, when/if we move to
VideoCommon pipelines, this state will be part of the pipeline UID
anyway, and we can mask it out in the backend by using a two-level map,
so the shaders/programs are shared.
This optimisation doesn't work on PowerVR's Vulkan implementation. We
(incorrectly) disallow Framebuffer objects to be used with a different
load or store op than that which they were created with, despite the
spec allowing such.
This fixes the windwaker intro "smearing"
Modernizes the arrays and makes future simplifications possible (e.g. usages within the software renderer).
It also makes cases where we use array->pointer decay explicit.
The class NonCopyable is, like the name says, supposed to disallow
copying. But should it allow moving?
For a long time, NonCopyable used to not allow moving. (It declared
a deleted copy constructor and assigment operator without declaring
a move constructor and assignment operator, making the compiler
implicitly delete the move constructor and assignment operator.)
That's fine if the classes that inherit from NonCopyable don't need
to be movable or if writing the move constructor and assignment
operator by hand is fine, but that's not the case for all classes,
as I discovered when I was working on the DirectoryBlob PR.
Because of that, I decided to make NonCopyable movable in c7602cc,
allowing me to use NonCopyable in DirectoryBlob.h. That was however
an unfortunate decision, because some of the classes that inherit
from NonCopyable have incorrect behavior when moved by default-
generated move constructors and assignment operators, and do not
explicitly delete the move constructors and assignment operators,
relying on NonCopyable being non-movable.
So what can we do about this? There are four solutions that I can
think of:
1. Make NonCopyable non-movable and tell DirectoryBlob to suck it.
2. Keep allowing moving NonCopyable, and expect that classes that
don't support moving will delete the move constructor and
assignment operator manually. Not only is this inconsistent
(having classes disallow copying one way and disallow moving
another way), but deleting the move constructor and assignment
operator manually is too easy to forget compared to how tricky
the resulting problems are.
3. Have one "MovableNonCopyable" and one "NonMovableNonCopyable".
It works, but it feels rather silly...
4. Don't have a NonCopyable class at all. Considering that deleting
the copy constructor and assignment operator only takes two lines
of code, I don't see much of a reason to keep NonCopyable. I
suppose that there was more of a point in having NonCopyable back
in the pre-C++11 days, when it wasn't possible to use "= delete".
I decided to go with the fourth one (like the commit title says).
The implementation of the commit is fairly straight-forward, though
I would like to point out that I skipped adding "= delete" lines
for classes whose only reason for being uncopyable is that they
contain uncopyable classes like File::IOFile and std::unique_ptr,
because the compiler makes such classes uncopyable automatically.
The switch statements in these functions appear to get transformed into
an if..else chain on NVIDIA's OpenGL/Vulkan drivers, resulting in lower
performance than the D3D counterparts. Transforming the switch into a
binary tree of ifs can increase performance by up to 20%.
Settings that come from the SYSCONF are now included in Dolphin's
config system as part of the base layer. They are handled in a
special way compared to other settings to make sure they are only
loaded from and saved to the SYSCONF (to avoid different, possibly
contradicting sources of truth).
This was breaking Metroid Prime 2: Echoes’s scanner in some rooms, such
as the one in https://fifoci.dolphin-emu.org/dff/mp2-scanner/
It was found on Ivy Bridge on Mesa, the alpha value read back from the
EFB was off-by-4 in multiple objects, which was a conversion error
because int4() is equivalent to floor() and the value wasn’t always
higher.
This is supposed to get efb2tex to the same texture as efb2ram, by applying the related efb copies as updates after each other, in the order of their creation.
Improve bookkeeping around formats. Hopefully make code less confusing.
- Rename TlutFormat -> TLUTFormat to follow conventions.
- Use enum classes to prevent using a Texture format where an EFB Copy format
is expected or vice-versa.
- Use common EFBCopyFormat names regardless of depth and YUV configurations.
This was mainly included for debugging, but could end up being confusing
for users, as well as polluting the GL program cache with a mix of uber
and specialized shaders if the option was changed.
These vertex formats enable all attributes. Inactive attributes are set
to offset=0, and the smallest type possible. This "optimization" stops
the NV compiler from generating variants of vertex shaders.
Markus Wick