Fixes a few crashes:
- fixes a crash related to waking threads (priorities were wrong).
- fixes a crash from reusing the SetRenderTargets texture array (left-over data causing issues).
- fixes a mistake and an oversight in the buffer system.
- buffers were getting updated wrong causing bad data to be stored or some times cut.
- modified ranges would extend past their old buffers, crashing on syncs. Old buffers are now skipped as the new buffers already sync instead.
Introduces pooling in a few more places to increase memory efficiency.
simplified RangeList item logic.
- removed RangeItem by making all the range objects use the I(NonOverlapping)Range interface.
- BufferCache class no longer locks its RangeList, as the list is only ever accessed synchronously.
Small change to how keyboard snapshots are stored.
Increase ThreadedRenderer SpanPool size to fit slightly more data (4MB -> 8MB).
General memory improvements to decrease GC pressure and frequency.
Pool big arrays and objects that are created and deleted often.
Skip data copies when they aren't needed.
Inline flag checks to skip unneeded allocations.
From my testing the performance is about the same, but the GC frequency is much lower and collection is faster causing less and smaller spikes.
- Move the message handler into its debugger class part,
- Move all message types into one file and collapse 3 of the ones with no data into a generic, stateless message with a single property being its type,
- Add an Fpscr helper property on IExecutionContext along with a comment about what Fpscr is (similar to the other registers in there)
- Moved the Rcmd helpers (such as GetRegisters, GetMinidump, etc) into a dedicated Debugger class part,
- Fixed the double-collection (ToArray being called twice) in GetThreadUids & GetThread in KProcess
Using resolution mods, heaps past 8 GiB work in some games (like LoZ: TotK) and not in others (like SMP Jamboree). Setting the heap to a hard limit of 8 GiB on the 10 & 12 GiB DRAM options seems to be the safe bet right now until a better solution is found.
The heap was limited to 6 GiB no matter the memory setting, causing memory configurations above 8 GiB to not actually affect the heap size.
Now when the memory config is set to [10 or] 12 GiB the heap also allocates 12 GiB.
The SetHeapSize SysCall will now allow heap sizes up to 12 GiB (technically slightly less).
Co-authored-by: KeatonTheBot <keaton@ryujinx.app>
Currently, some logs can be missing when a fatal error occurs (especially GuestBrokeExecutionException).
This MR attempts to flush logs to console and file before process exit.
After commit 51ca73cb, using a non-blocking socket (e.g. poll(..., timeout=0)) will still result in a blocking socket.
It appears the above commit inverted the behavior.
This merge request fixes the blocking flag: blocking will be set to false when the flag is 0x800.
A few more internal changes to the RangeList systems.
* No longer using a QuickAccess dictionary.
* The performance of the dictionary wasn't much faster than just doing binary searches.
* Using just binary searches allows us to take advantage of span and array returns as they're are faster than linked lists when iterating or copying the overlaps.
Small code optimizations.
Fixes a few leftover crashes.
* Slightly refactors RangeLists from the last Memory Changes MR, which fixes issue 61.
* Convert as many const size array iterators to span iterators as possible. When iterating over a const size array, every iteration created a Span, now only the first iteration does in most places.
* Now using object pooling for a few object types that were rapidly deleted and recreated.
* Converted a few flag checks to binary operations to save memory allocations.
* Refactors the RangeList and derivative classes used for handling lists of regions
* The Binary searches are now more performant, relying on edge searches instead of just returning the first matching hit and manually iterating until the edge is found
* Most look-ups now return a RangeItem, which acts as a linked list node now, instead where possible, moving away from Array copies. This should help with some specific lag spikes.
* Made IntrusiveRedBlackTreeNodes act like linked list nodes too to improve the lookup time of minimums, maximums and successors.
* Changed a few cases of HasFlag() into binary operations to save on memory allocations.
In general, [these changes] should increase frame time stability and lag spikes, but at the cost of some overhead to memory look-ups, the result being a very slightly better average fps from my testing (~1-2%).
* GetSerialNumber in ISystemSettingsServer
* Bionic qualifiers
* Revert application pool size to 3285 MB for 4 GB DRAM
* Fix PPTC recompilation if mod or patch is applied
This PR changes the core count to be defined in the device instead of
being a const value.
This is mostly a change for future features I want to implement and
should not impact any functionality.
The console will now log the range of cores requested from the
application, and for now, if the requested range is not 0 to 2 (the 3
cores used for application emulation), it will give an error message
which tells the user to contact me on discord. I'm doing this because
I'm interested in finding applications/games that don't use 3 cores and
the error will be removed in the future once I've gotten enough data.
* Kernel: Wake cores from idle directly rather than through a host thread
Right now when a core enters an idle state, leaving that idle state requires us to first signal the core's idle thread, which then signals the correct thread that we want to run on the core. This means that in a lot of cases, we're paying double for a thread to be woken from an idle state.
This PR moves this process to happen on the thread that is waking others out of idle, instead of an idle thread that needs to be woken first.
For compatibility the process has been kept as similar as possible - the process for IdleThreadLoop has been migrated to TryLeaveIdle, and is gated by a condition variable that lets it run only once at a time for each core. A core is only considered for wake from idle if idle is both active and has been signalled - the signal is consumed and the active state is cleared when the core leaves idle.
Dummy threads (just the idle thread at the moment) have been changed to have no host thread, as the work is now done by threads entering idle and signalling out of it.
This could put a bit of extra work on threads that would have triggered `_idleInterruptEvent` before, but I'd expect less work than signalling all those reset events and the OS overhead that follows. Worst case is that other threads performing these signals at the same time will have to wait for each other, but it's still going to be a very short amount of time.
Improvements are best seen in games with heavy (or very misguided) multithreading, such as Pokemon: Legends Arceus. Improvements are expected in Scarlet/Violet and TOTK, but are harder to measure.
Testing on Linux/MacOS still to be done, definitely need to test more games as this affects all of them (obviously) and any issues might be rare to encounter.
* Remove _idleThread entirely
* Use spinwait so we don't completely blast the CPU with cmpxchg
* Didn't I already do this
* Cleanup
KeatonTheBot