When I added the software FMA path in 2c38d64 and made us use
it when determinism is enabled, I was assuming that either the
performance impact of software FMA wouldn't be too large or CPUs
that were too old to have FMA instructions were too slow to run
Dolphin well anyway. This was wrong. To give an example, the
netplay performance went from 60 FPS to 30 FPS in one case.
This change makes netplay clients negotiate whether FMA should
be used. If all clients use an x64 CPU that supports FMA, or
AArch64, then FMA is enabled, and otherwise FMA is disabled.
In other words, we sacrifice accuracy if needed to avoid massive
slowdown, but not otherwise. When not using netplay, whether to
enable FMA is simply based on whether the host CPU supports it.
The only remaining case where the software FMA path gets used
under normal circumstances is when an input recording is created
on a CPU with FMA support and then played back on a CPU without.
This is not an especially common scenario (though it can happen),
and TASers are generally less picky about performance and more
picky about accuracy than other users anyway.
With this change, FMA desyncs are avoided between AArch64 and
modern x64 CPUs (unlike before 2c38d64), but we do get FMA
desyncs between AArch64 and old x64 CPUs (like before 2c38d64).
This desync can be avoided by adding a non-FMA path to JitArm64 as
an option, which I will wait with for another pull request so that
we can get the performance regression fixed as quickly as possible.
https://bugs.dolphin-emu.org/issues/12542
The config version should always be incremented whenever config is
changed, regardless of callbacks being suppressed or not.
Otherwise, getters can return stale data until another config change
(with callbacks enabled) happens.
The way Config::Get works in master, it first calls
Config::GetActiveLayerForConfig which searches for the
setting in all layers, and then calls Config::Layer::Get
which searches for the same setting again within the given
layer. We can remove this second search by combining the
logic of Config::GetActiveLayerForConfig and
Config::Layer::Get into one function.
Noticed missing include as a build failure on gcc-11:
```
[ 15%] Building CXX object Source/Core/Common/CMakeFiles/common.dir/Config/Config.cpp.o
Source/Core/Common/Config/Config.cpp:23:24:
error: 'unique_lock' in namespace 'std' does not name a template type
23 | using WriteLock = std::unique_lock<std::shared_mutex>;
| ^~~~~~~~~~~
Source/Core/Common/Config/Config.cpp:11:1:
note: 'std::unique_lock' is defined in header '<mutex>';
did you forget to '#include <mutex>'?
```
Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
A small, nonexhaustive set of warning fixes. The DiscIO Volume change
is a workaround for a GCC bug [1] that causes returning an unengaged
std::optional to emit annoying -Wmaybe-uninitialized warnings.
This last change alone fixes pages upon pages of warnings since
Volume.h is included from several files.
-Wstringop-truncation is another irrelevant warning for us, but
unfortunately there seems to be no way to disable it without
adding ugly pragmas wherever the warning appears.
Removed conditional use of std::mutex instead of std::shared_mutex on MacOS.
Because MacOS < 10.12 did not support std::shared_mutex, a previous commit
naïvely substituted std::mutex, which does not have the same behavior.
Reverses PR #8273, which substitues std::mutex for std::shared_mutex on
macOS, and results in several bugs that seem to only affect MacOS
- https://bugs.dolphin-emu.org/issues/11919
- https://bugs.dolphin-emu.org/issues/11842
- https://bugs.dolphin-emu.org/issues/11845
This change eliminates conditional code for MacOS in the core configuration
layer code and enables the use of modern language features that are more
secure and thread-safe.
This is done by:
1) Implementing said protocol in a new controller input class CemuHookUDPServer.
2) Adding functionality in the WiimoteEmu class for pushing that motion input to the emulated Wiimote and MotionPlus.
3) Suitably modifying the UI for configuring an Emulated Wii Remote.
API has been made stricter, layers are now managed with shared pointers,
so using them temporarily increased their reference counters.
Additionally, any s_layers map has been guarded by a read/write lock,
as concurrent write/reads to it were possible.
Only invoke config changed callbacks from Config::Save, not
Layer::Save. The latter results in callbacks being called
once per layer, up to 7 times per save.
Currently, a simple typo in the system name will trigger an assert
message that complains about a "programming error". This is not
user friendly and misleading.
So this changes GetSystemFromName to return an std::optional, which
allows for callers to check whether the system exists and handle
failures better.
Originally, Layer contained a std::map of Sections, which containted a std::map
containing the (key, value) pairs. Here we flattern this structure so that only
one std::map is required, reducing the number of indirections required and
vastly simplifying the code.
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).
The Config::AddLoadLayer functions call Load on the layer
explicitly, but Load is already called in the constructor,
so they'd cause the loader's Load function to be called twice,
which is potentially expensive considering we have to read an INI
from the host filesystem.
This commit removes the Config::AddLoadLayer functions because
they don't appear to be necessary.
JosJuice