[android] Another set of QoL changes for Android - 2 (#3886)

Changes:

- Defaults: Set Async GPU and Async Vulkan Presentation to OFF. Stability wasn't worth the trade-off.
- Threading: Lowered default pipeline workers from 7 to 4 to reduce heat and CPU contention.
- Settings: Added a slider for manual pipeline worker count so users can test what works best for their SoC.
- QCOM: Removed SPIRV bans; improves load times and thermals in heavy titles like Jump Force.
- UI: Cleaned up settings descriptions to be less ambiguous.

------------------------
Some games fixed:

-> Trinity Fusion: No longer crashes with Turnip, no longer shows the black dot in the middle of the screen on both QCOM and Turnip drivers.
-> Naruto X Boruto - Ultimate Ninja Storm Connections: Game no longer requires a fixed version of turnip to work (previously requiring Turnip driver from MESA 24.3/ @MrPurple666 EoL v2 driver)

Co-authored-by: lizzie <lizzie@eden-emu.dev>
Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/3886
Reviewed-by: crueter <crueter@eden-emu.dev>
Reviewed-by: Lizzie <lizzie@eden-emu.dev>

src/android/app/src/main/java/org/yuzu/yuzu_emu/features/settings/model/IntSetting.kt

@@ -67,7 +67,8 @@ enum class IntSetting(override val key: String) : AbstractIntSetting {
     MY_PAGE_APPLET("my_page_applet_mode"),
     INPUT_OVERLAY_AUTO_HIDE("input_overlay_auto_hide"),
     OVERLAY_GRID_SIZE("overlay_grid_size"),
-    GPU_LOG_RING_BUFFER_SIZE("gpu_log_ring_buffer_size")
+    GPU_LOG_RING_BUFFER_SIZE("gpu_log_ring_buffer_size"),
+    ANDROID_PIPELINE_WORKERS("pipeline_worker_count")
     ;
 
     override fun getInt(needsGlobal: Boolean): Int = NativeConfig.getInt(key, needsGlobal)

src/android/app/src/main/java/org/yuzu/yuzu_emu/features/settings/model/view/SettingsItem.kt

@@ -582,6 +582,16 @@ abstract class SettingsItem(
                     units = "%"
                 )
             )
+            put(
+                SliderSetting(
+                    IntSetting.ANDROID_PIPELINE_WORKERS,
+                    titleId = R.string.pipeline_worker_cores,
+                    descriptionId = R.string.pipeline_worker_cores_description,
+                    min = 4,
+                    max = 8,
+                    units = "cores"
+                )
+            )
             put(
                 SingleChoiceSetting(
                     IntSetting.RENDERER_ANTI_ALIASING,

src/android/app/src/main/java/org/yuzu/yuzu_emu/features/settings/ui/SettingsFragmentPresenter.kt

@@ -295,6 +295,7 @@ class SettingsFragmentPresenter(
             add(BooleanSetting.EMULATE_BGR565.key)
             add(BooleanSetting.RESCALE_HACK.key)
             add(BooleanSetting.RENDERER_ASYNCHRONOUS_SHADERS.key)
+            add(IntSetting.ANDROID_PIPELINE_WORKERS.key)
             add(BooleanSetting.RENDERER_ASYNCHRONOUS_GPU_EMULATION.key)
             add(BooleanSetting.RENDERER_ASYNC_PRESENTATION.key)
             add(SettingsItem.GPU_UNSWIZZLE_COMBINED)

src/android/app/src/main/jni/android_settings.h

@@ -147,6 +147,13 @@ namespace AndroidSettings {
                                                       &show_performance_overlay};
 
 
+        Settings::Setting<s32> pipeline_worker_count{linkage, 4, "pipeline_worker_count",
+                                 Settings::Category::Android,
+                                 Settings::Specialization::Default,
+                                 true,
+                                 true};
+
+
         Settings::Setting<bool> show_input_overlay{linkage, true, "show_input_overlay",
                                                    Settings::Category::Overlay};
         Settings::Setting<bool> overlay_snap_to_grid{linkage, false, "overlay_snap_to_grid",

src/android/app/src/main/res/values/strings.xml

@@ -71,6 +71,8 @@
     <string name="show_power_info_description">Display current power draw and remaining capacity on battery</string>
     <string name="show_shaders_building">Show Shaders Building</string>
     <string name="show_shaders_building_description">Display current number of shaders being built</string>
+    <string name="pipeline_worker_cores">Pipeline Worker Threads</string>
+    <string name="pipeline_worker_cores_description">Manage the amount of cores used for building Vulkan pipelines, the higher value will improve pipeline compilation performance but temperatures will increase as well.</string>
     <string name="overlay_position">Overlay Position</string>
     <string name="overlay_position_description">Choose where the overlay is displayed on the screen</string>
     <string name="overlay_position_top_left">Top Left</string>
@@ -491,15 +493,15 @@
     <string name="renderer_force_max_clock">Force maximum clocks (Adreno only)</string>
     <string name="renderer_force_max_clock_description">Forces the GPU to run at the maximum possible clocks (thermal constraints will still be applied).</string>
     <string name="renderer_asynchronous_gpu_emulation">GPU async emulation</string>
-    <string name="renderer_asynchronous_gpu_emulation_description">Runs GPU emulation asynchronously to reduce CPU stalls and improve throughput. Disable this only if you run into timing-related issues.</string>
+    <string name="renderer_asynchronous_gpu_emulation_description">This hack can increase performance by running GPU emulation asynchronously at the cost of graphical issues and increased crash rates by timing-related operations.</string>
     <string name="renderer_async_presentation">Asynchronous presentation</string>
-    <string name="renderer_async_presentation_description">Slightly improves performance by moving presentation to a separate CPU thread.</string>
+    <string name="renderer_async_presentation_description">This hack can increase performance by moving presentation to a separate CPU thread at the cost of graphical issues.</string>
     <string name="renderer_reactive_flushing">Use reactive flushing</string>
     <string name="renderer_reactive_flushing_description">Improves rendering accuracy in some games at the cost of performance.</string>
     <string name="enable_buffer_history">Enable buffer history</string>
     <string name="enable_buffer_history_description">Enables access to previous buffer states. This option may improve rendering quality and performance consistency in some games.</string>
     <string name="use_optimized_vertex_buffers">Optimized Vertex Buffers</string>
-    <string name="use_optimized_vertex_buffers_description">Enables optimized vertex buffer binding for improved performance. Requires Mesa 26.0+ Turnip drivers/ QCOM drivers. Will crash on older Turnip drivers.</string>
+    <string name="use_optimized_vertex_buffers_description">Enables optimized vertex buffer binding for improved performance. Requires Mesa 26.0+ Turnip drivers/ QCOM drivers. Will crash on older Turnip drivers (25.3 and below).</string>
 
     <string name="hacks">Hacks</string>
 
@@ -531,10 +533,10 @@
     <string name="extensions">Extensions</string>
 
     <string name="dyna_state">Extended Dynamic State</string>
-    <string name="dyna_state_description">Controls the number of features that can be used in Extended Dynamic State. Higher numbers allow for more features and can increase performance, but may cause issues with some drivers and vendors.</string>
+    <string name="dyna_state_description">Controls the number of features that can be used in ExtendedDynamicState (EDS). The higher value will allow to reduce the amount of pipeline compilations based on the dynamic state supported by driver.</string>
     <string name="disabled">Disabled</string>
     <string name="vertex_input_dynamic_state">Vertex Input Dynamic State</string>
-    <string name="vertex_input_dynamic_state_description">Enables vertex input dynamic state feature for better quality and performance.</string>
+    <string name="vertex_input_dynamic_state_description">Enabling this feature allows for more flexible vertex input handling, potentially reducing pipeline compilation time in vertex/buffer.</string>
     <string name="sample_shading_fraction">Sample Shading</string>
     <string name="sample_shading_fraction_description">Allows the fragment shader to execute per sample in a multi-sampled fragment instead once per fragment. Improves graphics quality at the cost of some performance.</string>
 

src/common/settings.h

@@ -389,7 +389,7 @@ struct Values {
                                                   true};
 
     SwitchableSetting<bool> use_asynchronous_gpu_emulation{
-                                                           linkage, true, "use_asynchronous_gpu_emulation", Category::Renderer};
+                                                           linkage, false, "use_asynchronous_gpu_emulation", Category::Renderer};
     // *nix platforms may have issues with the borderless windowed fullscreen mode.
     // Default to exclusive fullscreen on these platforms for now.
     SwitchableSetting<FullscreenMode, true> fullscreen_mode{linkage,
@@ -542,7 +542,7 @@ struct Values {
                                                         true};
     SwitchableSetting<bool> async_presentation{linkage,
 #ifdef ANDROID
-                                               true,
+                                               false,
 #else
                                                false,
 #endif
@@ -609,7 +609,7 @@ struct Values {
 #if defined (ANDROID)
                                                        false,
 #else
-                                                       true,
+                                                       false,
 #endif
                                                        "vertex_input_dynamic_state", Category::RendererExtensions};
 

src/video_core/renderer_vulkan/vk_pipeline_cache.cpp

@@ -45,6 +45,10 @@
 #include "video_core/vulkan_common/vulkan_wrapper.h"
 #include "video_core/gpu_logging/gpu_logging.h"
 
+#ifdef ANDROID
+#include "../../android/app/src/main/jni/android_settings.h"
+#endif
+
 namespace Vulkan {
 
 namespace {
@@ -325,13 +329,13 @@ size_t GetTotalPipelineWorkers() {
     const size_t max_core_threads =
         std::max<size_t>(static_cast<size_t>(std::thread::hardware_concurrency()), 2ULL) - 1ULL;
 #ifdef ANDROID
-    // Leave at least one core free on Android. Previously we reserved two, but
-    // shipping builds benefit from one extra compilation worker.
-    constexpr size_t free_cores = 1ULL;
-    if (max_core_threads <= free_cores) {
+    const int configured = AndroidSettings::values.pipeline_worker_count.GetValue();
+    const int clamped = std::clamp(configured, 4, 8);
+    const size_t desired = static_cast<size_t>(clamped);
+    if (desired == 0) {
         return 1ULL;
     }
-    return max_core_threads - free_cores;
+    return std::min(max_core_threads, desired);
 #else
     return max_core_threads;
 #endif
@@ -426,13 +430,12 @@ PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
                                        driver_id == VK_DRIVER_ID_INTEL_OPEN_SOURCE_MESA,
 
         .has_broken_spirv_clamp = driver_id == VK_DRIVER_ID_INTEL_PROPRIETARY_WINDOWS,
-        .has_broken_spirv_position_input = driver_id == VK_DRIVER_ID_QUALCOMM_PROPRIETARY,
+        .has_broken_spirv_position_input = driver_id == false,
         .has_broken_unsigned_image_offsets = false,
         .has_broken_signed_operations = false,
         .has_broken_fp16_float_controls = driver_id == VK_DRIVER_ID_NVIDIA_PROPRIETARY,
         .ignore_nan_fp_comparisons = false,
-        .has_broken_spirv_subgroup_mask_vector_extract_dynamic =
-            driver_id == VK_DRIVER_ID_QUALCOMM_PROPRIETARY,
+        .has_broken_spirv_subgroup_mask_vector_extract_dynamic = false,
         .has_broken_robust =
             device.IsNvidia() && device.GetNvidiaArch() <= NvidiaArchitecture::Arch_Pascal,
         .min_ssbo_alignment = device.GetStorageBufferAlignment(),