feat: add Low GPU Accuracy setting for maximum performance

Implements a new "Low" GPU accuracy level that prioritizes performance
over accuracy by aggressively cutting corners in GPU emulation.

Changes:
- Add GpuAccuracy::Low enum and setting infrastructure
- Implement IsGPULevelNormal() helper function
- Skip texture cache checks and query operations
- Use unsafe memory reads for DMA operations
- Disable fence delays and query precision
- Add UI support for desktop (Qt) and Android

Performance optimizations:
- Skips texture cache coherency checks (vk/gl_rasterizer.cpp)
- Non-blocking query synchronization (query_cache.h)
- Unsafe memory operations (dma_pusher.cpp)
- No macro parameter refresh (maxwell_3d.cpp)
- Immediate fence signaling (fence_manager.h)
- Non-precise Vulkan queries (vk_query_cache.cpp)

Ideal for lower-end hardware and users prioritizing FPS over accuracy.
Works on both desktop and Android platforms.

Signed-off-by: Zephyron <zephyron@citron-emu.org>

src/video_core/dma_pusher.cpp

@@ -98,7 +98,8 @@ bool DmaPusher::Step() {
                         &command_headers);
             ProcessCommands(headers);
         };
-        if (Settings::IsGPULevelHigh()) {
+        if (Settings::IsGPULevelNormal()) {
+            // Normal/High/Extreme: Use safe reads for most operations
             if (dma_state.method >= MacroRegistersStart) {
                 unsafe_process();
                 return true;
@@ -106,6 +107,10 @@ bool DmaPusher::Step() {
             safe_process();
             return true;
         }
+        // Low accuracy: Use unsafe reads for maximum performance everywhere
+        unsafe_process();
+        return true;
+        // Note: The code below is unreachable for Low, but kept for reference
         // Even in normal accuracy, use safe reads for KeplerCompute inline methods
         if (subchannel_type[dma_state.subchannel] == Engines::EngineTypes::KeplerCompute &&
             dma_state.method == ComputeInline) {

src/video_core/engines/maxwell_3d.cpp

@@ -221,7 +221,8 @@ void Maxwell3D::ProcessMacro(u32 method, const u32* base_start, u32 amount, bool
 }
 
 void Maxwell3D::RefreshParametersImpl() {
-    if (!Settings::IsGPULevelHigh()) {
+    if (!Settings::IsGPULevelNormal()) {
+        // Skip parameter refresh for Low accuracy - ultimate performance
         return;
     }
     size_t current_index = 0;

src/video_core/fence_manager.h

@@ -72,7 +72,7 @@ public:
     }
 
     void SignalFence(std::function<void()>&& func) {
-        bool delay_fence = Settings::IsGPULevelHigh();
+        bool delay_fence = Settings::IsGPULevelNormal();
         if constexpr (!can_async_check) {
             TryReleasePendingFences<false>();
         }

src/video_core/query_cache/query_cache.h

@@ -257,7 +257,7 @@ void QueryCacheBase<Traits>::CounterReport(GPUVAddr addr, QueryType counter_type
     };
     u8* pointer = impl->device_memory.template GetPointer<u8>(cpu_addr);
     u8* pointer_timestamp = impl->device_memory.template GetPointer<u8>(cpu_addr + 8);
-    bool is_synced = !Settings::IsGPULevelHigh() && is_fence;
+    bool is_synced = !Settings::IsGPULevelNormal() && is_fence;
     std::function<void()> operation([this, is_synced, streamer, query_base = query, query_location,
                                      pointer, pointer_timestamp] {
         if (True(query_base->flags & QueryFlagBits::IsInvalidated)) {
@@ -287,7 +287,8 @@ void QueryCacheBase<Traits>::CounterReport(GPUVAddr addr, QueryType counter_type
     if (is_fence) {
         impl->rasterizer.SignalFence(std::move(operation));
     } else {
-        if (!Settings::IsGPULevelHigh() && counter_type == QueryType::Payload) {
+        if (!Settings::IsGPULevelNormal() && counter_type == QueryType::Payload) {
+            // Low accuracy: Immediately write payload for ultimate performance
             if (has_timestamp) {
                 u64 timestamp = impl->gpu.GetTicks();
                 u64 value = static_cast<u64>(payload);

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -545,7 +545,8 @@ bool RasterizerOpenGL::MustFlushRegion(DAddr addr, u64 size, VideoCommon::CacheT
             return true;
         }
     }
-    if (!Settings::IsGPULevelHigh()) {
+    if (!Settings::IsGPULevelNormal()) {
+        // Skip texture cache checks for Low accuracy - ultimate performance
         return false;
     }
     if (True(which & VideoCommon::CacheType::TextureCache)) {
@@ -740,7 +741,8 @@ bool RasterizerOpenGL::AccelerateConditionalRendering() {
         // Reimplement Host conditional rendering.
         return false;
     }
-    // Medium / Low Hack: stub any checks on queries written into the buffer cache.
+    // Normal / Low Hack: stub any checks on queries written into the buffer cache.
+    // Low accuracy: Always stub for maximum performance
     const GPUVAddr condition_address{maxwell3d->regs.render_enable.Address()};
     Maxwell::ReportSemaphore::Compare cmp;
     if (gpu_memory->IsMemoryDirty(condition_address, sizeof(cmp),

src/video_core/renderer_vulkan/vk_query_cache.cpp

@@ -153,7 +153,7 @@ public:
         ReserveHostQuery();
         scheduler.Record([query_pool = current_query_pool,
                           query_index = current_bank_slot](vk::CommandBuffer cmdbuf) {
-            const bool use_precise = Settings::IsGPULevelHigh();
+            const bool use_precise = Settings::IsGPULevelNormal();
             cmdbuf.BeginQuery(query_pool, static_cast<u32>(query_index),
                               use_precise ? VK_QUERY_CONTROL_PRECISE_BIT : 0);
         });
@@ -1415,8 +1415,9 @@ bool QueryCacheRuntime::HostConditionalRenderingCompareValues(VideoCommon::Looku
         return false;
     }
 
-    const bool is_gpu_high = Settings::IsGPULevelHigh();
+    const bool is_gpu_high = Settings::IsGPULevelNormal();
     if (!is_gpu_high && impl->device.GetDriverID() == VK_DRIVER_ID_INTEL_PROPRIETARY_WINDOWS) {
+        // Low accuracy: stub conditional rendering on Intel for performance
         return true;
     }
 

src/video_core/renderer_vulkan/vk_rasterizer.cpp

@@ -565,7 +565,8 @@ bool RasterizerVulkan::MustFlushRegion(DAddr addr, u64 size, VideoCommon::CacheT
             return true;
         }
     }
-    if (!Settings::IsGPULevelHigh()) {
+    if (!Settings::IsGPULevelNormal()) {
+        // Skip texture cache checks for Low accuracy - ultimate performance
         return false;
     }
     if (True(which & VideoCommon::CacheType::TextureCache)) {

src/video_core/renderer_vulkan/vk_scheduler.cpp

@@ -260,11 +260,14 @@ void Scheduler::AllocateNewContext() {
     if (query_cache) {
 #if ANDROID
         if (Settings::IsGPULevelHigh()) {
-            // This is problematic on Android, disable on GPU Normal.
+            // This is problematic on Android, disable on GPU Normal and Low.
             query_cache->NotifySegment(true);
         }
 #else
-        query_cache->NotifySegment(true);
+        if (Settings::IsGPULevelNormal()) {
+            // Skip query cache operations for Low accuracy
+            query_cache->NotifySegment(true);
+        }
 #endif
     }
 }
@@ -278,13 +281,16 @@ void Scheduler::InvalidateState() {
 void Scheduler::EndPendingOperations() {
 #if ANDROID
     if (Settings::IsGPULevelHigh()) {
-        // This is problematic on Android, disable on GPU Normal.
+        // This is problematic on Android, disable on GPU Normal and Low.
         // query_cache->DisableStreams();
     }
 #else
     // query_cache->DisableStreams();
 #endif
-    query_cache->NotifySegment(false);
+    if (Settings::IsGPULevelNormal()) {
+        // Skip query cache operations for Low accuracy
+        query_cache->NotifySegment(false);
+    }
     EndRenderPass();
 }