Compare commits

..

7 Commits

Author SHA1 Message Date
lizzie 0f8effc22a use scalar 2026-07-03 17:56:36 +00:00
lizzie b930951254 value hint 2026-07-03 17:56:36 +00:00
lizzie 0a38913d9c android 2026-07-03 17:56:36 +00:00
lizzie 481e0e450d [common] make unswizzle settings be able to be configured for all powers of 2 from 2^1 to 2^16
Signed-off-by: lizzie <lizzie@eden-emu.dev>
2026-07-03 17:55:43 +00:00
lizzie 4c65780f11 Revert "[common/dynamic_library] fix AUR build error (#4156)" (#4157)
This reverts commit a9c4c8aefd.

Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/4157
Reviewed-by: CamilleLaVey <camillelavey99@gmail.com>
Reviewed-by: MaranBr <maranbr@eden-emu.dev>
2026-07-02 22:07:37 +02:00
lizzie a9c4c8aefd [common/dynamic_library] fix AUR build error (#4156)
not gonna question why there was an ifdef there

Signed-off-by: lizzie <lizzie@eden-emu.dev>

Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/4156
Reviewed-by: CamilleLaVey <camillelavey99@gmail.com>
Reviewed-by: MaranBr <maranbr@eden-emu.dev>
2026-07-02 19:49:19 +02:00
lizzie a769505a45 [vk, ogl] Remove dedicated precomputed swizzle table; use shorter inline table in block_linear swizzle shaders (#4146)
old table = 64 * 8 * sizeof(u32) = 512 * 4 = 4096 bytes
new table = 8 * sizeof(u32) = 8 * 4 = 32 bytes

the expression
```glsl
    return ((pos & 0x0180) >> 1)
         | ((pos & 0x0040) >> 2)
         | ((pos & 0x0020) << 3)
         | ((pos & 0x0010) << 1)
         | ((pos & 0x000f) << 0);
```
is equivalent for generating the table but idk if we'd want that

Signed-off-by: lizzie <lizzie@eden-emu.dev>

Co-authored-by: CamilleLaVey <camillelavey99@gmail.com>
Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/4146
Reviewed-by: CamilleLaVey <camillelavey99@gmail.com>
Reviewed-by: MaranBr <maranbr@eden-emu.dev>
2026-07-02 19:24:07 +02:00
38 changed files with 201 additions and 1089 deletions
@@ -1,89 +0,0 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
package org.yuzu.yuzu_emu.features.settings.model.view
import androidx.annotation.ArrayRes
import androidx.annotation.StringRes
import org.yuzu.yuzu_emu.features.settings.model.AbstractSetting
import org.yuzu.yuzu_emu.features.settings.model.BooleanSetting
import org.yuzu.yuzu_emu.features.settings.model.IntSetting
class GpuUnswizzleSetting(
@StringRes titleId: Int = 0,
titleString: String = "",
@StringRes descriptionId: Int = 0,
descriptionString: String = "",
@ArrayRes val textureSizeChoicesId: Int,
@ArrayRes val textureSizeValuesId: Int,
@ArrayRes val streamSizeChoicesId: Int,
@ArrayRes val streamSizeValuesId: Int,
@ArrayRes val chunkSizeChoicesId: Int,
@ArrayRes val chunkSizeValuesId: Int
) : SettingsItem(
object : AbstractSetting {
override val key: String = SettingsItem.GPU_UNSWIZZLE_COMBINED
override val defaultValue: Any = false
override val isSaveable = true
override val isRuntimeModifiable = true
override val isSwitchable = true
override val pairedSettingKey: String = ""
override var global: Boolean
get() {
return BooleanSetting.GPU_UNSWIZZLE_ENABLED.global &&
IntSetting.GPU_UNSWIZZLE_TEXTURE_SIZE.global &&
IntSetting.GPU_UNSWIZZLE_STREAM_SIZE.global &&
IntSetting.GPU_UNSWIZZLE_CHUNK_SIZE.global
}
set(value) {
BooleanSetting.GPU_UNSWIZZLE_ENABLED.global = value
IntSetting.GPU_UNSWIZZLE_TEXTURE_SIZE.global = value
IntSetting.GPU_UNSWIZZLE_STREAM_SIZE.global = value
IntSetting.GPU_UNSWIZZLE_CHUNK_SIZE.global = value
}
override fun getValueAsString(needsGlobal: Boolean): String = "combined"
override fun reset() {
BooleanSetting.GPU_UNSWIZZLE_ENABLED.reset()
IntSetting.GPU_UNSWIZZLE_TEXTURE_SIZE.reset()
IntSetting.GPU_UNSWIZZLE_STREAM_SIZE.reset()
IntSetting.GPU_UNSWIZZLE_CHUNK_SIZE.reset()
}
},
titleId,
titleString,
descriptionId,
descriptionString
) {
override val type = SettingsItem.TYPE_GPU_UNSWIZZLE
// Check if GPU unswizzle is enabled via the dedicated boolean setting
fun isEnabled(needsGlobal: Boolean = false): Boolean =
BooleanSetting.GPU_UNSWIZZLE_ENABLED.getBoolean(needsGlobal)
fun setEnabled(value: Boolean) =
BooleanSetting.GPU_UNSWIZZLE_ENABLED.setBoolean(value)
fun enable() = setEnabled(true)
fun disable() = setEnabled(false)
fun getTextureSize(needsGlobal: Boolean = false): Int =
IntSetting.GPU_UNSWIZZLE_TEXTURE_SIZE.getInt(needsGlobal)
fun setTextureSize(value: Int) =
IntSetting.GPU_UNSWIZZLE_TEXTURE_SIZE.setInt(value)
fun getStreamSize(needsGlobal: Boolean = false): Int =
IntSetting.GPU_UNSWIZZLE_STREAM_SIZE.getInt(needsGlobal)
fun setStreamSize(value: Int) =
IntSetting.GPU_UNSWIZZLE_STREAM_SIZE.setInt(value)
fun getChunkSize(needsGlobal: Boolean = false): Int =
IntSetting.GPU_UNSWIZZLE_CHUNK_SIZE.getInt(needsGlobal)
fun setChunkSize(value: Int) =
IntSetting.GPU_UNSWIZZLE_CHUNK_SIZE.setInt(value)
fun reset() = setting.reset()
}
@@ -686,42 +686,33 @@ abstract class SettingsItem(
)
)
put(
SingleChoiceSetting(
SpinBoxSetting(
IntSetting.GPU_UNSWIZZLE_TEXTURE_SIZE,
titleId = R.string.gpu_unswizzle_texture_size,
descriptionId = R.string.gpu_unswizzle_texture_size_description,
choicesId = R.array.gpuTextureSizeSwizzleEntries,
valuesId = R.array.gpuTextureSizeSwizzleValues
valueHint = R.string.gpu_unswizzle_texture_size,
min = 1,
max = 9
)
)
put(
SingleChoiceSetting(
SpinBoxSetting(
IntSetting.GPU_UNSWIZZLE_STREAM_SIZE,
titleId = R.string.gpu_unswizzle_stream_size,
descriptionId = R.string.gpu_unswizzle_stream_size_description,
choicesId = R.array.gpuSwizzleEntries,
valuesId = R.array.gpuSwizzleValues
valueHint = R.string.gpu_unswizzle_stream_size,
min = 1,
max = 9
)
)
put(
SingleChoiceSetting(
SpinBoxSetting(
IntSetting.GPU_UNSWIZZLE_CHUNK_SIZE,
titleId = R.string.gpu_unswizzle_chunk_size,
descriptionId = R.string.gpu_unswizzle_chunk_size_description,
choicesId = R.array.gpuSwizzleChunkEntries,
valuesId = R.array.gpuSwizzleChunkValues
)
)
put(
GpuUnswizzleSetting(
titleId = R.string.gpu_unswizzle_settings,
descriptionId = R.string.gpu_unswizzle_settings_description,
textureSizeChoicesId = R.array.gpuTextureSizeSwizzleEntries,
textureSizeValuesId = R.array.gpuTextureSizeSwizzleValues,
streamSizeChoicesId = R.array.gpuSwizzleEntries,
streamSizeValuesId = R.array.gpuSwizzleValues,
chunkSizeChoicesId = R.array.gpuSwizzleChunkEntries,
chunkSizeValuesId = R.array.gpuSwizzleChunkValues
valueHint = R.string.gpu_unswizzle_chunk_size,
min = 1,
max = 9
)
)
put(
@@ -1,206 +0,0 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
package org.yuzu.yuzu_emu.features.settings.ui
import android.app.Dialog
import android.content.DialogInterface
import android.os.Bundle
import android.view.LayoutInflater
import android.widget.ArrayAdapter
import androidx.fragment.app.DialogFragment
import androidx.fragment.app.activityViewModels
import com.google.android.material.dialog.MaterialAlertDialogBuilder
import org.yuzu.yuzu_emu.R
import org.yuzu.yuzu_emu.databinding.DialogGpuUnswizzleBinding
import org.yuzu.yuzu_emu.features.settings.model.view.GpuUnswizzleSetting
class GpuUnswizzleDialogFragment : DialogFragment() {
private var position = 0
private val settingsViewModel: SettingsViewModel by activityViewModels()
private lateinit var binding: DialogGpuUnswizzleBinding
override fun onCreate(savedInstanceState: Bundle?) {
super.onCreate(savedInstanceState)
position = requireArguments().getInt(POSITION)
if (settingsViewModel.clickedItem == null) dismiss()
}
override fun onCreateDialog(savedInstanceState: Bundle?): Dialog {
binding = DialogGpuUnswizzleBinding.inflate(LayoutInflater.from(requireContext()))
val item = settingsViewModel.clickedItem as GpuUnswizzleSetting
// Setup texture size dropdown
val textureSizeEntries = resources.getStringArray(item.textureSizeChoicesId)
val textureSizeValues = resources.getIntArray(item.textureSizeValuesId)
val textureSizeAdapter = ArrayAdapter(
requireContext(),
android.R.layout.simple_dropdown_item_1line,
textureSizeEntries.toMutableList()
)
binding.dropdownTextureSize.setAdapter(textureSizeAdapter)
// Setup stream size dropdown
val streamSizeEntries = resources.getStringArray(item.streamSizeChoicesId)
val streamSizeValues = resources.getIntArray(item.streamSizeValuesId)
val streamSizeAdapter = ArrayAdapter(
requireContext(),
android.R.layout.simple_dropdown_item_1line,
streamSizeEntries.toMutableList()
)
binding.dropdownStreamSize.setAdapter(streamSizeAdapter)
// Setup chunk size dropdown
val chunkSizeEntries = resources.getStringArray(item.chunkSizeChoicesId)
val chunkSizeValues = resources.getIntArray(item.chunkSizeValuesId)
val chunkSizeAdapter = ArrayAdapter(
requireContext(),
android.R.layout.simple_dropdown_item_1line,
chunkSizeEntries.toMutableList()
)
binding.dropdownChunkSize.setAdapter(chunkSizeAdapter)
// Load current values
val isEnabled = item.isEnabled()
binding.switchEnable.isChecked = isEnabled
if (isEnabled) {
val textureSizeIndex = textureSizeValues.indexOf(item.getTextureSize())
if (textureSizeIndex >= 0) {
binding.dropdownTextureSize.setText(textureSizeEntries[textureSizeIndex], false)
}
val streamSizeIndex = streamSizeValues.indexOf(item.getStreamSize())
if (streamSizeIndex >= 0) {
binding.dropdownStreamSize.setText(streamSizeEntries[streamSizeIndex], false)
}
val chunkSizeIndex = chunkSizeValues.indexOf(item.getChunkSize())
if (chunkSizeIndex >= 0) {
binding.dropdownChunkSize.setText(chunkSizeEntries[chunkSizeIndex], false)
}
} else {
// Set default/recommended values when disabling
binding.dropdownTextureSize.setText(textureSizeEntries[3], false)
binding.dropdownStreamSize.setText(streamSizeEntries[3], false)
binding.dropdownChunkSize.setText(chunkSizeEntries[3], false)
}
// Clear adapter filters after setText to fix rotation bug
textureSizeAdapter.filter.filter(null)
streamSizeAdapter.filter.filter(null)
chunkSizeAdapter.filter.filter(null)
// Enable/disable dropdowns based on switch state
updateDropdownsState(isEnabled)
binding.switchEnable.setOnCheckedChangeListener { _, checked ->
updateDropdownsState(checked)
}
val dialog = MaterialAlertDialogBuilder(requireContext())
.setTitle(item.title)
.setView(binding.root)
.create()
// Setup button listeners
binding.btnDefault.setOnClickListener {
// Reset to defaults
item.reset()
// Refresh values with adapters reset
val textureSizeIndex = textureSizeValues.indexOf(item.getTextureSize())
if (textureSizeIndex >= 0) {
binding.dropdownTextureSize.setText(textureSizeEntries[textureSizeIndex], false)
}
val streamSizeIndex = streamSizeValues.indexOf(item.getStreamSize())
if (streamSizeIndex >= 0) {
binding.dropdownStreamSize.setText(streamSizeEntries[streamSizeIndex], false)
}
val chunkSizeIndex = chunkSizeValues.indexOf(item.getChunkSize())
if (chunkSizeIndex >= 0) {
binding.dropdownChunkSize.setText(chunkSizeEntries[chunkSizeIndex], false)
}
// Clear filters
textureSizeAdapter.filter.filter(null)
streamSizeAdapter.filter.filter(null)
chunkSizeAdapter.filter.filter(null)
settingsViewModel.setAdapterItemChanged(position)
settingsViewModel.setShouldReloadSettingsList(true)
}
binding.btnCancel.setOnClickListener {
dialog.dismiss()
}
binding.btnOk.setOnClickListener {
if (binding.switchEnable.isChecked) {
item.enable()
// Save the selected values
val selectedTextureIndex = textureSizeEntries.indexOf(
binding.dropdownTextureSize.text.toString()
)
if (selectedTextureIndex >= 0) {
item.setTextureSize(textureSizeValues[selectedTextureIndex])
}
val selectedStreamIndex = streamSizeEntries.indexOf(
binding.dropdownStreamSize.text.toString()
)
if (selectedStreamIndex >= 0) {
item.setStreamSize(streamSizeValues[selectedStreamIndex])
}
val selectedChunkIndex = chunkSizeEntries.indexOf(
binding.dropdownChunkSize.text.toString()
)
if (selectedChunkIndex >= 0) {
item.setChunkSize(chunkSizeValues[selectedChunkIndex])
}
} else {
// Disable GPU unswizzle
item.disable()
}
settingsViewModel.setAdapterItemChanged(position)
settingsViewModel.setShouldReloadSettingsList(true)
dialog.dismiss()
}
// Ensure filters are cleared after dialog is shown
binding.root.post {
textureSizeAdapter.filter.filter(null)
streamSizeAdapter.filter.filter(null)
chunkSizeAdapter.filter.filter(null)
}
return dialog
}
private fun updateDropdownsState(enabled: Boolean) {
binding.layoutTextureSize.isEnabled = enabled
binding.dropdownTextureSize.isEnabled = enabled
binding.layoutStreamSize.isEnabled = enabled
binding.dropdownStreamSize.isEnabled = enabled
binding.layoutChunkSize.isEnabled = enabled
binding.dropdownChunkSize.isEnabled = enabled
}
companion object {
const val TAG = "GpuUnswizzleDialogFragment"
const val POSITION = "Position"
fun newInstance(
settingsViewModel: SettingsViewModel,
item: GpuUnswizzleSetting,
position: Int
): GpuUnswizzleDialogFragment {
val dialog = GpuUnswizzleDialogFragment()
val args = Bundle()
args.putInt(POSITION, position)
dialog.arguments = args
settingsViewModel.clickedItem = item
return dialog
}
}
}
@@ -102,10 +102,6 @@ class SettingsAdapter(
PathViewHolder(ListItemSettingBinding.inflate(inflater), this)
}
SettingsItem.TYPE_GPU_UNSWIZZLE -> {
GpuUnswizzleViewHolder(ListItemSettingBinding.inflate(inflater), this)
}
else -> {
HeaderViewHolder(ListItemSettingsHeaderBinding.inflate(inflater), this)
}
@@ -479,14 +475,6 @@ class SettingsAdapter(
settingsViewModel.setShouldShowPathResetDialog(true)
}
fun onGpuUnswizzleClick(item: GpuUnswizzleSetting, position: Int) {
GpuUnswizzleDialogFragment.newInstance(
settingsViewModel,
item,
position
).show(fragment.childFragmentManager, GpuUnswizzleDialogFragment.TAG)
}
private class DiffCallback : DiffUtil.ItemCallback<SettingsItem>() {
override fun areItemsTheSame(oldItem: SettingsItem, newItem: SettingsItem): Boolean {
return oldItem.setting.key == newItem.setting.key
@@ -1,71 +0,0 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
package org.yuzu.yuzu_emu.features.settings.ui.viewholder
import android.view.View
import org.yuzu.yuzu_emu.R
import org.yuzu.yuzu_emu.databinding.ListItemSettingBinding
import org.yuzu.yuzu_emu.features.settings.model.view.GpuUnswizzleSetting
import org.yuzu.yuzu_emu.features.settings.model.view.SettingsItem
import org.yuzu.yuzu_emu.features.settings.ui.SettingsAdapter
import org.yuzu.yuzu_emu.utils.ViewUtils.setVisible
class GpuUnswizzleViewHolder(val binding: ListItemSettingBinding, adapter: SettingsAdapter) :
SettingViewHolder(binding.root, adapter) {
private lateinit var setting: GpuUnswizzleSetting
override fun bind(item: SettingsItem) {
setting = item as GpuUnswizzleSetting
binding.textSettingName.text = setting.title
binding.textSettingDescription.setVisible(item.description.isNotEmpty())
binding.textSettingDescription.text = item.description
binding.textSettingValue.setVisible(true)
val resMgr = binding.root.context.resources
if (setting.isEnabled()) {
// Show a summary of current settings
val textureSizeEntries = resMgr.getStringArray(setting.textureSizeChoicesId)
val textureSizeValues = resMgr.getIntArray(setting.textureSizeValuesId)
val textureSizeIndex = textureSizeValues.indexOf(setting.getTextureSize())
val textureSizeLabel = if (textureSizeIndex >= 0) textureSizeEntries[textureSizeIndex] else "?"
val streamSizeEntries = resMgr.getStringArray(setting.streamSizeChoicesId)
val streamSizeValues = resMgr.getIntArray(setting.streamSizeValuesId)
val streamSizeIndex = streamSizeValues.indexOf(setting.getStreamSize())
val streamSizeLabel = if (streamSizeIndex >= 0) streamSizeEntries[streamSizeIndex] else "?"
val chunkSizeEntries = resMgr.getStringArray(setting.chunkSizeChoicesId)
val chunkSizeValues = resMgr.getIntArray(setting.chunkSizeValuesId)
val chunkSizeIndex = chunkSizeValues.indexOf(setting.getChunkSize())
val chunkSizeLabel = if (chunkSizeIndex >= 0) chunkSizeEntries[chunkSizeIndex] else "?"
binding.textSettingValue.text = "$textureSizeLabel$streamSizeLabel$chunkSizeLabel"
} else {
binding.textSettingValue.text = resMgr.getString(R.string.gpu_unswizzle_disabled)
}
binding.buttonClear.setVisible(setting.clearable)
binding.buttonClear.setOnClickListener {
adapter.onClearClick(setting, bindingAdapterPosition)
}
setStyle(setting.isEditable, binding)
}
override fun onClick(clicked: View) {
if (!setting.isEditable) {
return
}
adapter.onGpuUnswizzleClick(setting, bindingAdapterPosition)
}
override fun onLongClick(clicked: View): Boolean {
if (setting.isEditable) {
return adapter.onLongClick(setting, bindingAdapterPosition)
}
return false
}
}
@@ -516,10 +516,7 @@
<string name="rescale_hack_description">يُمكّن هذا الخيار من التعامل مع عملية إعادة تحجيم الألعاب بطريقة تقليدية باستخدام مسار إعادة التحجيم السريع</string>
<string name="renderer_asynchronous_shaders">استخدم تظليل غير متزامن</string>
<string name="renderer_asynchronous_shaders_description">يقوم بتجميع التظليل بشكل غير متزامن. قد يقلل ذلك من التقطعات ولكنه قد يؤدي أيضًا إلى حدوث أخطاء.</string>
<string name="gpu_unswizzle_settings">إعدادات إلغاء ترتيب بيانات وحدة معالجة الرسومات</string>
<string name="gpu_unswizzle_settings_description">قم بضبط معلمات فكّ تشابك النسيج المستندة إلى وحدة معالجة الرسومات أو تعطيلها تمامًا. اضبط هذه الإعدادات لتحقيق التوازن بين الأداء وجودة تحميل النسيج.</string>
<string name="gpu_unswizzle_enable">تفعيل إلغاء ترتيب بيانات وحدة معالجة الرسومات</string>
<string name="gpu_unswizzle_disabled">تعطيل</string>
<string name="gpu_unswizzle_texture_size">الحد الأقصى لحجم النسيج في وحدة معالجة الرسومات بعد إعادة ترتيب البيانات</string>
<string name="gpu_unswizzle_texture_size_description">يُحدد هذا الخيار الحد الأقصى لحجم (ميغابايت) معالجة الصور باستخدام وحدة معالجة الرسومات. مع أن وحدة معالجة الرسومات أسرع في معالجة الصور المتوسطة والكبيرة، إلا أن وحدة المعالجة المركزية قد تكون أكثر كفاءة في معالجة الصور الصغيرة جدًا. اضبط هذا الخيار لتحقيق التوازن الأمثل بين سرعة معالجة الرسومات واستهلاك وحدة المعالجة المركزية.</string>
<string name="gpu_unswizzle_stream_size">حجم تدفق إلغاء ترتيب بيانات وحدة معالجة الرسومات</string>
@@ -957,25 +954,10 @@
<string name="fast_gpu_high">مرتفع (512)</string>
<!-- GPU swizzle texture size -->
<string name="gpu_texturesizeswizzle_verysmall">صغير جدًا (16 ميغابايت)</string>
<string name="gpu_texturesizeswizzle_small">صغير (32 ميغابايت)</string>
<string name="gpu_texturesizeswizzle_normal">قياسي (128 ميغابايت)</string>
<string name="gpu_texturesizeswizzle_large">كبير (256 ميغابايت)</string>
<string name="gpu_texturesizeswizzle_verylarge">كبير جدًا (512 ميغابايت)</string>
<!-- GPU swizzle streams -->
<string name="gpu_swizzle_verylow">منخفض جدًا (4 ميغابايت)</string>
<string name="gpu_swizzle_low">منخفض (8 ميغابايت)</string>
<string name="gpu_swizzle_normal">قياسي (16 ميغابايت)</string>
<string name="gpu_swizzle_medium">متوسط (32 ميغابايت)</string>
<string name="gpu_swizzle_high">عالي (64 ميغابايت)</string>
<!-- GPU swizzle chunks -->
<string name="gpu_swizzlechunk_verylow">منخفض جدًا (32)</string>
<string name="gpu_swizzlechunk_low">منخفض (64)</string>
<string name="gpu_swizzlechunk_normal">قياسي (128)</string>
<string name="gpu_swizzlechunk_medium">متوسط (256)</string>
<string name="gpu_swizzlechunk_high">عالي (512)</string>
<!-- Temperature Units -->
<string name="temperature_celsius">مئوية</string>
@@ -890,25 +890,10 @@ Wirklich fortfahren?</string>
<string name="fast_gpu_high">Hoch (512)</string>
<!-- GPU swizzle texture size -->
<string name="gpu_texturesizeswizzle_verysmall">Sehr klein (16 MB)</string>
<string name="gpu_texturesizeswizzle_small">Klein (32 MB)</string>
<string name="gpu_texturesizeswizzle_normal">Normal (128 MB)</string>
<string name="gpu_texturesizeswizzle_large">Groß (256 MB)</string>
<string name="gpu_texturesizeswizzle_verylarge">Sehr groß (512 MB)</string>
<!-- GPU swizzle streams -->
<string name="gpu_swizzle_verylow">Sehr niedrig (4 MB)</string>
<string name="gpu_swizzle_low">Niedrig (8 MB)</string>
<string name="gpu_swizzle_normal">Normal (16 MB)</string>
<string name="gpu_swizzle_medium">Mittel (32 MB)</string>
<string name="gpu_swizzle_high">Hoch (64 MB)</string>
<!-- GPU swizzle chunks -->
<string name="gpu_swizzlechunk_verylow">Sehr niedrig (32)</string>
<string name="gpu_swizzlechunk_low">Niedrig (64)</string>
<string name="gpu_swizzlechunk_normal">Normal (128)</string>
<string name="gpu_swizzlechunk_medium">Mittel (256)</string>
<string name="gpu_swizzlechunk_high">Hoch (512)</string>
<!-- Temperature Units -->
<string name="temperature_celsius">Celsius</string>
@@ -510,10 +510,7 @@
<string name="rescale_hack_description">Permite el manejo de versiones anteriores para el paso de configuración de reescalado para juegos mediante el uso de una ruta de reescalado rápida.</string>
<string name="renderer_asynchronous_shaders">Usar sombreadores asíncronos</string>
<string name="renderer_asynchronous_shaders_description">Compila los sombreadores de forma asíncrona. Esto puede reducir los tirones, pero también puede introducir errores gráficos.</string>
<string name="gpu_unswizzle_settings">Ajustes de desentrelazado de la GPU</string>
<string name="gpu_unswizzle_settings_description">Configura los parámetros de desentrelazado de texturas basadas en la GPU o desactívelos por completo. Modifique estos ajustes para equilibrar el rendimiento y la calidad de las texturas cargadas.</string>
<string name="gpu_unswizzle_enable">Activar desentrelazado de la GPU</string>
<string name="gpu_unswizzle_disabled">Desactivado</string>
<string name="gpu_unswizzle_texture_size">Tamaño máximo de textura de desentrelazado de la GPU</string>
<string name="gpu_unswizzle_texture_size_description">Establece el tamaño máximo (en MB) para el desentrelazado de texturas basada en GPU. Aunque la GPU es más rápida para texturas medianas y grandes, la CPU puede ser más eficiente para texturas muy pequeñas. Ajuste este valor para encontrar el equilibrio entre la aceleración de la GPU y la sobrecarga de la CPU.</string>
<string name="gpu_unswizzle_stream_size">Tamaño del flujo de desentrelazado de la GPU</string>
@@ -951,25 +948,10 @@
<string name="fast_gpu_high">Alto (512)</string>
<!-- GPU swizzle texture size -->
<string name="gpu_texturesizeswizzle_verysmall">Muy pequeño (16 MB)</string>
<string name="gpu_texturesizeswizzle_small">Pequeño (32 MB)</string>
<string name="gpu_texturesizeswizzle_normal">Normal (128 MB)</string>
<string name="gpu_texturesizeswizzle_large">Grande (256 MB)</string>
<string name="gpu_texturesizeswizzle_verylarge">Muy grande (512 MB)</string>
<!-- GPU swizzle streams -->
<string name="gpu_swizzle_verylow">Muy bajo (4 MB)</string>
<string name="gpu_swizzle_low">Bajo (8 MB)</string>
<string name="gpu_swizzle_normal">Normal (16 MB)</string>
<string name="gpu_swizzle_medium">Medio (32 MB)</string>
<string name="gpu_swizzle_high">Alto (64 MB)</string>
<!-- GPU swizzle chunks -->
<string name="gpu_swizzlechunk_verylow">Muy bajo (32)</string>
<string name="gpu_swizzlechunk_low">Bajo (64)</string>
<string name="gpu_swizzlechunk_normal">Normal (128)</string>
<string name="gpu_swizzlechunk_medium">Medio (256)</string>
<string name="gpu_swizzlechunk_high">Alto (512)</string>
<!-- Temperature Units -->
<string name="temperature_celsius">Celsius</string>
@@ -509,10 +509,7 @@
<string name="rescale_hack_description">Включает старый метод обработки этапа перенастройки масштабирования для игр за счёт использования быстрого алгоритма перемасштабирования.</string>
<string name="renderer_asynchronous_shaders">Использовать асинхронные шейдеры</string>
<string name="renderer_asynchronous_shaders_description">Компилирует шейдеры асинхронно. Это может уменьшить подтормаживания, но также может вызвать графические артефакты.</string>
<string name="gpu_unswizzle_settings">Настройки распаковки текстур (Unswizzle)</string>
<string name="gpu_unswizzle_settings_description">Настройте параметры распаковки текстур на стороне ГПУ либо полностью отключите эту функцию. Изменение этих параметров позволяет найти баланс между производительностью и качеством загрузки текстур.</string>
<string name="gpu_unswizzle_enable">Включить распаковку текстур (Unswizzle)</string>
<string name="gpu_unswizzle_disabled">Отключено</string>
<string name="gpu_unswizzle_texture_size">Макс. размер текстуры Unswizzle</string>
<string name="gpu_unswizzle_texture_size_description">Задает максимальный размер (в МБ) текстур для преобразования формата (unswizzle) на ГПУ. Хотя ГПУ быстрее работает со средними и большими текстурами, ЦП может быть эффективнее для очень маленьких. Настройте это значение, чтобы найти баланс между ускорением на ГПУ и нагрузкой на ЦП.</string>
<string name="gpu_unswizzle_stream_size">Размер потока Unswizzle</string>
@@ -950,25 +947,10 @@
<string name="fast_gpu_high">Высокое (512)</string>
<!-- GPU swizzle texture size -->
<string name="gpu_texturesizeswizzle_verysmall">Очень малый (16 МБ)</string>
<string name="gpu_texturesizeswizzle_small">Малый (32 МБ)</string>
<string name="gpu_texturesizeswizzle_normal">Обычный (128 МБ)</string>
<string name="gpu_texturesizeswizzle_large">Большой (256 МБ)</string>
<string name="gpu_texturesizeswizzle_verylarge">Очень большой (512 МБ)</string>
<!-- GPU swizzle streams -->
<string name="gpu_swizzle_verylow">Очень низкий (4 МБ)</string>
<string name="gpu_swizzle_low">Низкий (8 МБ)</string>
<string name="gpu_swizzle_normal">Обычный (16 МБ)</string>
<string name="gpu_swizzle_medium">Средний (32 МБ)</string>
<string name="gpu_swizzle_high">Высокий (64 МБ)</string>
<!-- GPU swizzle chunks -->
<string name="gpu_swizzlechunk_verylow">Очень малый (32)</string>
<string name="gpu_swizzlechunk_low">Малый (64)</string>
<string name="gpu_swizzlechunk_normal">Обычный (128)</string>
<string name="gpu_swizzlechunk_medium">Средний (256)</string>
<string name="gpu_swizzlechunk_high">Большой (512)</string>
<!-- Temperature Units -->
<string name="temperature_celsius">Цельсий</string>
@@ -512,10 +512,7 @@
<string name="rescale_hack_description">Вмикає застарілу обробку масштабування для ігор, використовуючи швидкий шлях масштабування</string>
<string name="renderer_asynchronous_shaders">Асинхронні шейдери</string>
<string name="renderer_asynchronous_shaders_description">Компілює шейдери асинхронно. Це може зменшити затримки, але також може спричинити графічні баги.</string>
<string name="gpu_unswizzle_settings">Налаштування розпакування за допомогою ГП</string>
<string name="gpu_unswizzle_settings_description">Налаштуйте розпакування текстур за допомогою ГП або повністю вимкнути його. Відкоригуйте ці налаштування, щоб урівноважити продуктивність і якість завантаження текстур.</string>
<string name="gpu_unswizzle_enable">Увімкнути розпакування за допомогою ГП</string>
<string name="gpu_unswizzle_disabled">Вимкнено</string>
<string name="gpu_unswizzle_texture_size">Максимальний розмір текстур для розпакування ГП за допомогою ГП</string>
<string name="gpu_unswizzle_texture_size_description">Встановлює максимальний розмір (МБ) для розпакування текстур за допомогою ГП. ГП швидше справляється з текстурами середніх і великих розмірів, а ЦП ефективніший для дуже маленьких. Налаштуйте, щоб збалансувати ГП-прискоренням і навантаженням на ЦП.</string>
<string name="gpu_unswizzle_stream_size">Розмір потоку розпакування за допомогою ГП</string>
@@ -953,25 +950,10 @@
<string name="fast_gpu_high">Високо (512)</string>
<!-- GPU swizzle texture size -->
<string name="gpu_texturesizeswizzle_verysmall">Дуже малий (16 МБ)</string>
<string name="gpu_texturesizeswizzle_small">Малий (32 МБ)</string>
<string name="gpu_texturesizeswizzle_normal">Нормальний (128 МБ)</string>
<string name="gpu_texturesizeswizzle_large">Великий (256 МБ)</string>
<string name="gpu_texturesizeswizzle_verylarge">Дуже великий (512 МБ)</string>
<!-- GPU swizzle streams -->
<string name="gpu_swizzle_verylow">Дуже низький (4 МБ)</string>
<string name="gpu_swizzle_low">Низький (8 МБ)</string>
<string name="gpu_swizzle_normal">Нормальний (16 МБ)</string>
<string name="gpu_swizzle_medium">Середній (32 МБ)</string>
<string name="gpu_swizzle_high">Високий (64 МБ)</string>
<!-- GPU swizzle chunks -->
<string name="gpu_swizzlechunk_verylow">Дуже низький (32)</string>
<string name="gpu_swizzlechunk_low">Низький (64)</string>
<string name="gpu_swizzlechunk_normal">Нормальний (128)</string>
<string name="gpu_swizzlechunk_medium">Середній (256)</string>
<string name="gpu_swizzlechunk_high">Високий (512)</string>
<!-- Temperature Units -->
<string name="temperature_celsius">Цельсій</string>
@@ -947,25 +947,10 @@
<string name="fast_gpu_high">高 (512)</string>
<!-- GPU swizzle texture size -->
<string name="gpu_texturesizeswizzle_verysmall">极小 (16 MB)</string>
<string name="gpu_texturesizeswizzle_small">较小 (32 MB)</string>
<string name="gpu_texturesizeswizzle_normal">正常 (128 MB)</string>
<string name="gpu_texturesizeswizzle_large">较大 (256 MB)</string>
<string name="gpu_texturesizeswizzle_verylarge">极大 (512 MB)</string>
<!-- GPU swizzle streams -->
<string name="gpu_swizzle_verylow">极低 (4 MB)</string>
<string name="gpu_swizzle_low">低 (8 MB)</string>
<string name="gpu_swizzle_normal">正常 (16 MB)</string>
<string name="gpu_swizzle_medium">中 (32 MB)</string>
<string name="gpu_swizzle_high">高 (64 MB)</string>
<!-- GPU swizzle chunks -->
<string name="gpu_swizzlechunk_verylow">极低 (32)</string>
<string name="gpu_swizzlechunk_low">低 (64)</string>
<string name="gpu_swizzlechunk_normal">正常 (128)</string>
<string name="gpu_swizzlechunk_medium">中 (256)</string>
<string name="gpu_swizzlechunk_high">高 (512)</string>
<!-- Temperature Units -->
<string name="temperature_celsius">摄氏度</string>
@@ -545,54 +545,6 @@
<item>2</item>
</integer-array>
<string-array name="gpuTextureSizeSwizzleEntries">
<item>@string/gpu_texturesizeswizzle_verysmall</item>
<item>@string/gpu_texturesizeswizzle_small</item>
<item>@string/gpu_texturesizeswizzle_normal</item>
<item>@string/gpu_texturesizeswizzle_large</item>
<item>@string/gpu_texturesizeswizzle_verylarge</item>
</string-array>
<integer-array name="gpuTextureSizeSwizzleValues">
<item>0</item>
<item>1</item>
<item>2</item>
<item>3</item>
<item>4</item>
</integer-array>
<string-array name="gpuSwizzleEntries">
<item>@string/gpu_swizzle_verylow</item>
<item>@string/gpu_swizzle_low</item>
<item>@string/gpu_swizzle_normal</item>
<item>@string/gpu_swizzle_medium</item>
<item>@string/gpu_swizzle_high</item>
</string-array>
<integer-array name="gpuSwizzleValues">
<item>0</item>
<item>1</item>
<item>2</item>
<item>3</item>
<item>4</item>
</integer-array>
<string-array name="gpuSwizzleChunkEntries">
<item>@string/gpu_swizzlechunk_verylow</item>
<item>@string/gpu_swizzlechunk_low</item>
<item>@string/gpu_swizzlechunk_normal</item>
<item>@string/gpu_swizzlechunk_medium</item>
<item>@string/gpu_swizzlechunk_high</item>
</string-array>
<integer-array name="gpuSwizzleChunkValues">
<item>0</item>
<item>1</item>
<item>2</item>
<item>3</item>
<item>4</item>
</integer-array>
<string-array name="temperatureUnitEntries">
<item>@string/temperature_celsius</item>
<item>@string/temperature_fahrenheit</item>
@@ -524,10 +524,7 @@
<string name="rescale_hack_description">Enables a legacy handling for the rescale configuration pass for games by using a quick rescale path</string>
<string name="renderer_asynchronous_shaders">Use asynchronous shaders</string>
<string name="renderer_asynchronous_shaders_description">Compiles shaders asynchronously. This may reduce stutters but may also introduce glitches.</string>
<string name="gpu_unswizzle_settings">GPU Unswizzle Settings</string>
<string name="gpu_unswizzle_settings_description">Configure GPU-based texture unswizzling parameters or disable it entirely. Adjust these settings to balance performance and texture loading quality.</string>
<string name="gpu_unswizzle_enable">Enable GPU Unswizzle</string>
<string name="gpu_unswizzle_disabled">Disabled</string>
<string name="gpu_unswizzle_texture_size">GPU Unswizzle Max Texture Size</string>
<string name="gpu_unswizzle_texture_size_description">Sets the maximum size (MB) for GPU-based texture unswizzling. While the GPU is faster for medium and large textures, the CPU may be more efficient for very small ones. Adjust this to find the balance between GPU acceleration and CPU overhead.</string>
<string name="gpu_unswizzle_stream_size">GPU Unswizzle Stream Size</string>
@@ -978,25 +975,10 @@
<string name="fast_gpu_high">High (512)</string>
<!-- GPU swizzle texture size -->
<string name="gpu_texturesizeswizzle_verysmall">Very Small (16 MB)</string>
<string name="gpu_texturesizeswizzle_small">Small (32 MB)</string>
<string name="gpu_texturesizeswizzle_normal">Normal (128 MB)</string>
<string name="gpu_texturesizeswizzle_large">Large (256 MB)</string>
<string name="gpu_texturesizeswizzle_verylarge">Very Large (512 MB)</string>
<!-- GPU swizzle streams -->
<string name="gpu_swizzle_verylow">Very Low (4 MB)</string>
<string name="gpu_swizzle_low">Low (8 MB)</string>
<string name="gpu_swizzle_normal">Normal (16 MB)</string>
<string name="gpu_swizzle_medium">Medium (32 MB)</string>
<string name="gpu_swizzle_high">High (64 MB)</string>
<!-- GPU swizzle chunks -->
<string name="gpu_swizzlechunk_verylow">Very Low (32)</string>
<string name="gpu_swizzlechunk_low">Low (64)</string>
<string name="gpu_swizzlechunk_normal">Normal (128)</string>
<string name="gpu_swizzlechunk_medium">Medium (256)</string>
<string name="gpu_swizzlechunk_high">High (512)</string>
<!-- Temperature Units -->
<string name="temperature_celsius">Celsius</string>
+15 -9
View File
@@ -584,23 +584,29 @@ struct Values {
SwitchableSetting<bool> use_asynchronous_shaders{linkage, false, "use_asynchronous_shaders",
Category::RendererHacks};
SwitchableSetting<GpuUnswizzleSize> gpu_unswizzle_texture_size{linkage,
GpuUnswizzleSize::Large,
SwitchableSetting<u32, true> gpu_unswizzle_texture_size{linkage,
6,
1,
9,
"gpu_unswizzle_texture_size",
Category::RendererHacks,
Specialization::Default};
Specialization::Scalar};
SwitchableSetting<GpuUnswizzle> gpu_unswizzle_stream_size{linkage,
GpuUnswizzle::Medium,
SwitchableSetting<u32, true> gpu_unswizzle_stream_size{linkage,
4,
1,
9,
"gpu_unswizzle_stream_size",
Category::RendererHacks,
Specialization::Default};
Specialization::Scalar};
SwitchableSetting<GpuUnswizzleChunk> gpu_unswizzle_chunk_size{linkage,
GpuUnswizzleChunk::Medium,
SwitchableSetting<u32, true> gpu_unswizzle_chunk_size{linkage,
7,
1,
9,
"gpu_unswizzle_chunk_size",
Category::RendererHacks,
Specialization::Default};
Specialization::Scalar};
SwitchableSetting<bool> gpu_unswizzle_enabled{linkage, false, "gpu_unswizzle_enabled",
Category::RendererHacks};
-3
View File
@@ -152,9 +152,6 @@ ENUM(ConsoleMode, Handheld, Docked);
ENUM(AppletMode, HLE, LLE);
ENUM(SpirvOptimizeMode, Never, OnLoad, Always);
ENUM(GpuOverclock, Normal, Medium, High)
ENUM(GpuUnswizzleSize, VerySmall, Small, Normal, Large, VeryLarge)
ENUM(GpuUnswizzle, VeryLow, Low, Normal, Medium, High)
ENUM(GpuUnswizzleChunk, VeryLow, Low, Normal, Medium, High)
ENUM(TemperatureUnits, Celsius, Fahrenheit)
ENUM(ExtendedDynamicState, Disabled, EDS1, EDS2, EDS3);
ENUM(GpuLogLevel, Off, Errors, Standard, Verbose, All)
+3 -28
View File
@@ -240,16 +240,16 @@ std::unique_ptr<TranslationMap> InitializeTranslations(QObject* parent) {
tr("Accelerates BCn 3D texture decoding using GPU compute.\n"
"Disable if experiencing crashes or graphical glitches."));
INSERT(Settings, gpu_unswizzle_texture_size, tr("GPU Unswizzle Max Texture Size"),
tr("Sets the maximum size (MiB) for GPU-based texture unswizzling.\n"
tr("Sets the maximum size (power of 2 * 1MiB) for GPU-based texture unswizzling.\n"
"While the GPU is faster for medium and large textures, the CPU may be more "
"efficient for very small ones.\n"
"Adjust this to find the balance between GPU acceleration and CPU overhead."));
INSERT(Settings, gpu_unswizzle_stream_size, tr("GPU Unswizzle Stream Size"),
tr("Sets the maximum amount of texture data (in MiB) processed per frame.\n"
tr("Sets the maximum amount of texture data (power of 2 * 1MiB) processed per frame.\n"
"Higher values can reduce stutter during texture loading but may impact frame "
"consistency."));
INSERT(Settings, gpu_unswizzle_chunk_size, tr("GPU Unswizzle Chunk Size"),
tr("Determines the number of depth slices processed in a single dispatch.\n"
tr("Determines the number of depth slices (power of 2) processed in a single dispatch.\n"
"Increasing this can improve throughput on high-end GPUs but may cause TDR or driver "
"timeouts on weaker hardware."));
@@ -652,31 +652,6 @@ std::unique_ptr<ComboboxTranslationMap> ComboboxEnumeration(QObject* parent) {
PAIR(GpuOverclock, Medium, tr("Medium (256)")),
PAIR(GpuOverclock, High, tr("High (512)")),
}});
translations->insert({Settings::EnumMetadata<Settings::GpuUnswizzleSize>::Index(),
{
PAIR(GpuUnswizzleSize, VerySmall, tr("Very Small (16 MB)")),
PAIR(GpuUnswizzleSize, Small, tr("Small (32 MB)")),
PAIR(GpuUnswizzleSize, Normal, tr("Normal (128 MB)")),
PAIR(GpuUnswizzleSize, Large, tr("Large (256 MB)")),
PAIR(GpuUnswizzleSize, VeryLarge, tr("Very Large (512 MB)")),
}});
translations->insert({Settings::EnumMetadata<Settings::GpuUnswizzle>::Index(),
{
PAIR(GpuUnswizzle, VeryLow, tr("Very Low (4 MB)")),
PAIR(GpuUnswizzle, Low, tr("Low (8 MB)")),
PAIR(GpuUnswizzle, Normal, tr("Normal (16 MB)")),
PAIR(GpuUnswizzle, Medium, tr("Medium (32 MB)")),
PAIR(GpuUnswizzle, High, tr("High (64 MB)")),
}});
translations->insert({Settings::EnumMetadata<Settings::GpuUnswizzleChunk>::Index(),
{
PAIR(GpuUnswizzleChunk, VeryLow, tr("Very Low (32)")),
PAIR(GpuUnswizzleChunk, Low, tr("Low (64)")),
PAIR(GpuUnswizzleChunk, Normal, tr("Normal (128)")),
PAIR(GpuUnswizzleChunk, Medium, tr("Medium (256)")),
PAIR(GpuUnswizzleChunk, High, tr("High (512)")),
}});
translations->insert({Settings::EnumMetadata<Settings::ExtendedDynamicState>::Index(),
{
PAIR(ExtendedDynamicState, Disabled, tr("Disabled")),
@@ -332,9 +332,6 @@ void DefineEntryPoint(const IR::Program& program, EmitContext& ctx, Id main) {
void SetupDenormControl(const Profile& profile, const IR::Program& program, EmitContext& ctx,
Id main_func) {
const Info& info{program.info};
if (profile.has_broken_fp16_float_controls && info.uses_fp16) {
return;
}
if (info.uses_fp32_denorms_flush && info.uses_fp32_denorms_preserve) {
LOG_DEBUG(Shader_SPIRV, "Fp32 denorm flush and preserve on the same shader");
} else if (info.uses_fp32_denorms_flush) {
@@ -435,7 +432,7 @@ void SetupCapabilities(const Profile& profile, const Info& info, EmitContext& ct
}
if ((info.uses_subgroup_vote || info.uses_subgroup_invocation_id ||
info.uses_subgroup_shuffles) &&
profile.support_vote && profile.SupportsSubgroupStage(ctx.stage)) {
profile.support_vote) {
ctx.AddCapability(spv::Capability::GroupNonUniformBallot);
ctx.AddCapability(spv::Capability::GroupNonUniformShuffle);
if (!profile.warp_size_potentially_larger_than_guest) {
@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -18,7 +15,7 @@ Id SharedPointer(EmitContext& ctx, Id offset, u32 index_offset = 0) {
if (index_offset > 0) {
index = ctx.OpIAdd(ctx.U32[1], index, ctx.Const(index_offset));
}
return ctx.uses_explicit_workgroup_layout
return ctx.profile.support_explicit_workgroup_layout
? ctx.OpAccessChain(ctx.shared_u32, ctx.shared_memory_u32, ctx.u32_zero_value, index)
: ctx.OpAccessChain(ctx.shared_u32, ctx.shared_memory_u32, index);
}
@@ -158,7 +155,7 @@ Id EmitSharedAtomicExchange32(EmitContext& ctx, Id offset, Id value) {
}
Id EmitSharedAtomicExchange64(EmitContext& ctx, Id offset, Id value) {
if (ctx.profile.support_shared_int64_atomics && ctx.uses_explicit_workgroup_layout) {
if (ctx.profile.support_int64_atomics && ctx.profile.support_explicit_workgroup_layout) {
const Id shift_id{ctx.Const(3U)};
const Id index{ctx.OpShiftRightArithmetic(ctx.U32[1], offset, shift_id)};
const Id pointer{
@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -31,7 +28,7 @@ std::pair<Id, Id> ExtractArgs(EmitContext& ctx, Id offset, u32 mask, u32 count)
} // Anonymous namespace
Id EmitLoadSharedU8(EmitContext& ctx, Id offset) {
if (ctx.uses_explicit_workgroup_layout) {
if (ctx.profile.support_explicit_workgroup_layout) {
const Id pointer{
ctx.OpAccessChain(ctx.shared_u8, ctx.shared_memory_u8, ctx.u32_zero_value, offset)};
return ctx.OpUConvert(ctx.U32[1], ctx.OpLoad(ctx.U8, pointer));
@@ -42,7 +39,7 @@ Id EmitLoadSharedU8(EmitContext& ctx, Id offset) {
}
Id EmitLoadSharedS8(EmitContext& ctx, Id offset) {
if (ctx.uses_explicit_workgroup_layout) {
if (ctx.profile.support_explicit_workgroup_layout) {
const Id pointer{
ctx.OpAccessChain(ctx.shared_u8, ctx.shared_memory_u8, ctx.u32_zero_value, offset)};
return ctx.OpSConvert(ctx.U32[1], ctx.OpLoad(ctx.U8, pointer));
@@ -53,7 +50,7 @@ Id EmitLoadSharedS8(EmitContext& ctx, Id offset) {
}
Id EmitLoadSharedU16(EmitContext& ctx, Id offset) {
if (ctx.uses_explicit_workgroup_layout) {
if (ctx.profile.support_explicit_workgroup_layout) {
const Id pointer{Pointer(ctx, ctx.shared_u16, ctx.shared_memory_u16, offset, 1)};
return ctx.OpUConvert(ctx.U32[1], ctx.OpLoad(ctx.U16, pointer));
} else {
@@ -63,7 +60,7 @@ Id EmitLoadSharedU16(EmitContext& ctx, Id offset) {
}
Id EmitLoadSharedS16(EmitContext& ctx, Id offset) {
if (ctx.uses_explicit_workgroup_layout) {
if (ctx.profile.support_explicit_workgroup_layout) {
const Id pointer{Pointer(ctx, ctx.shared_u16, ctx.shared_memory_u16, offset, 1)};
return ctx.OpSConvert(ctx.U32[1], ctx.OpLoad(ctx.U16, pointer));
} else {
@@ -73,7 +70,7 @@ Id EmitLoadSharedS16(EmitContext& ctx, Id offset) {
}
Id EmitLoadSharedU32(EmitContext& ctx, Id offset) {
if (ctx.uses_explicit_workgroup_layout) {
if (ctx.profile.support_explicit_workgroup_layout) {
const Id pointer{Pointer(ctx, ctx.shared_u32, ctx.shared_memory_u32, offset, 2)};
return ctx.OpLoad(ctx.U32[1], pointer);
} else {
@@ -82,7 +79,7 @@ Id EmitLoadSharedU32(EmitContext& ctx, Id offset) {
}
Id EmitLoadSharedU64(EmitContext& ctx, Id offset) {
if (ctx.uses_explicit_workgroup_layout) {
if (ctx.profile.support_explicit_workgroup_layout) {
const Id pointer{Pointer(ctx, ctx.shared_u32x2, ctx.shared_memory_u32x2, offset, 3)};
return ctx.OpLoad(ctx.U32[2], pointer);
} else {
@@ -97,7 +94,7 @@ Id EmitLoadSharedU64(EmitContext& ctx, Id offset) {
}
Id EmitLoadSharedU128(EmitContext& ctx, Id offset) {
if (ctx.uses_explicit_workgroup_layout) {
if (ctx.profile.support_explicit_workgroup_layout) {
const Id pointer{Pointer(ctx, ctx.shared_u32x4, ctx.shared_memory_u32x4, offset, 4)};
return ctx.OpLoad(ctx.U32[4], pointer);
}
@@ -113,7 +110,7 @@ Id EmitLoadSharedU128(EmitContext& ctx, Id offset) {
}
void EmitWriteSharedU8(EmitContext& ctx, Id offset, Id value) {
if (ctx.uses_explicit_workgroup_layout) {
if (ctx.profile.support_explicit_workgroup_layout) {
const Id pointer{
ctx.OpAccessChain(ctx.shared_u8, ctx.shared_memory_u8, ctx.u32_zero_value, offset)};
ctx.OpStore(pointer, ctx.OpUConvert(ctx.U8, value));
@@ -123,7 +120,7 @@ void EmitWriteSharedU8(EmitContext& ctx, Id offset, Id value) {
}
void EmitWriteSharedU16(EmitContext& ctx, Id offset, Id value) {
if (ctx.uses_explicit_workgroup_layout) {
if (ctx.profile.support_explicit_workgroup_layout) {
const Id pointer{Pointer(ctx, ctx.shared_u16, ctx.shared_memory_u16, offset, 1)};
ctx.OpStore(pointer, ctx.OpUConvert(ctx.U16, value));
} else {
@@ -133,7 +130,7 @@ void EmitWriteSharedU16(EmitContext& ctx, Id offset, Id value) {
void EmitWriteSharedU32(EmitContext& ctx, Id offset, Id value) {
Id pointer{};
if (ctx.uses_explicit_workgroup_layout) {
if (ctx.profile.support_explicit_workgroup_layout) {
pointer = Pointer(ctx, ctx.shared_u32, ctx.shared_memory_u32, offset, 2);
} else {
const Id shift{ctx.Const(2U)};
@@ -144,7 +141,7 @@ void EmitWriteSharedU32(EmitContext& ctx, Id offset, Id value) {
}
void EmitWriteSharedU64(EmitContext& ctx, Id offset, Id value) {
if (ctx.uses_explicit_workgroup_layout) {
if (ctx.profile.support_explicit_workgroup_layout) {
const Id pointer{Pointer(ctx, ctx.shared_u32x2, ctx.shared_memory_u32x2, offset, 3)};
ctx.OpStore(pointer, value);
return;
@@ -159,7 +156,7 @@ void EmitWriteSharedU64(EmitContext& ctx, Id offset, Id value) {
}
void EmitWriteSharedU128(EmitContext& ctx, Id offset, Id value) {
if (ctx.uses_explicit_workgroup_layout) {
if (ctx.profile.support_explicit_workgroup_layout) {
const Id pointer{Pointer(ctx, ctx.shared_u32x4, ctx.shared_memory_u32x4, offset, 4)};
ctx.OpStore(pointer, value);
return;
@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -13,14 +10,7 @@ Id SubgroupScope(EmitContext& ctx) {
return ctx.Const(static_cast<u32>(spv::Scope::Subgroup));
}
bool StageSupportsSubgroups(EmitContext& ctx) {
return ctx.profile.SupportsSubgroupStage(ctx.stage);
}
Id GetThreadId(EmitContext& ctx) {
if (!StageSupportsSubgroups(ctx)) {
return ctx.u32_zero_value;
}
return ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id);
}
@@ -78,9 +68,6 @@ Id GetMaxThreadId(EmitContext& ctx, Id thread_id, Id clamp, Id segmentation_mask
}
Id SelectValue(EmitContext& ctx, Id in_range, Id value, Id src_thread_id) {
if (!StageSupportsSubgroups(ctx)) {
return value;
}
return ctx.OpSelect(
ctx.U32[1], in_range,
ctx.OpGroupNonUniformShuffle(ctx.U32[1], SubgroupScope(ctx), value, src_thread_id), value);
@@ -102,9 +89,6 @@ Id EmitLaneId(EmitContext& ctx) {
}
Id EmitVoteAll(EmitContext& ctx, Id pred) {
if (!StageSupportsSubgroups(ctx)) {
return pred;
}
if (!ctx.profile.warp_size_potentially_larger_than_guest) {
return ctx.OpGroupNonUniformAll(ctx.U1, SubgroupScope(ctx), pred);
}
@@ -118,9 +102,6 @@ Id EmitVoteAll(EmitContext& ctx, Id pred) {
}
Id EmitVoteAny(EmitContext& ctx, Id pred) {
if (!StageSupportsSubgroups(ctx)) {
return pred;
}
if (!ctx.profile.warp_size_potentially_larger_than_guest) {
return ctx.OpGroupNonUniformAny(ctx.U1, SubgroupScope(ctx), pred);
}
@@ -134,9 +115,6 @@ Id EmitVoteAny(EmitContext& ctx, Id pred) {
}
Id EmitVoteEqual(EmitContext& ctx, Id pred) {
if (!StageSupportsSubgroups(ctx)) {
return ctx.true_value;
}
if (!ctx.profile.warp_size_potentially_larger_than_guest) {
return ctx.OpGroupNonUniformAllEqual(ctx.U1, SubgroupScope(ctx), pred);
}
@@ -151,9 +129,6 @@ Id EmitVoteEqual(EmitContext& ctx, Id pred) {
}
Id EmitSubgroupBallot(EmitContext& ctx, Id pred) {
if (!StageSupportsSubgroups(ctx)) {
return ctx.OpSelect(ctx.U32[1], pred, ctx.Const(1u), ctx.u32_zero_value);
}
const Id ballot{ctx.OpGroupNonUniformBallot(ctx.U32[4], SubgroupScope(ctx), pred)};
if (!ctx.profile.warp_size_potentially_larger_than_guest) {
return ctx.OpCompositeExtract(ctx.U32[1], ballot, 0U);
@@ -162,37 +137,22 @@ Id EmitSubgroupBallot(EmitContext& ctx, Id pred) {
}
Id EmitSubgroupEqMask(EmitContext& ctx) {
if (!StageSupportsSubgroups(ctx)) {
return ctx.Const(1u);
}
return LoadMask(ctx, ctx.subgroup_mask_eq);
}
Id EmitSubgroupLtMask(EmitContext& ctx) {
if (!StageSupportsSubgroups(ctx)) {
return ctx.u32_zero_value;
}
return LoadMask(ctx, ctx.subgroup_mask_lt);
}
Id EmitSubgroupLeMask(EmitContext& ctx) {
if (!StageSupportsSubgroups(ctx)) {
return ctx.Const(1u);
}
return LoadMask(ctx, ctx.subgroup_mask_le);
}
Id EmitSubgroupGtMask(EmitContext& ctx) {
if (!StageSupportsSubgroups(ctx)) {
return ctx.u32_zero_value;
}
return LoadMask(ctx, ctx.subgroup_mask_gt);
}
Id EmitSubgroupGeMask(EmitContext& ctx) {
if (!StageSupportsSubgroups(ctx)) {
return ctx.Const(1u);
}
return LoadMask(ctx, ctx.subgroup_mask_ge);
}
@@ -262,7 +222,7 @@ Id EmitShuffleButterfly(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id
Id EmitFSwizzleAdd(EmitContext& ctx, Id op_a, Id op_b, Id swizzle) {
const Id three{ctx.Const(3U)};
Id mask{GetThreadId(ctx)};
Id mask{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
mask = ctx.OpBitwiseAnd(ctx.U32[1], mask, three);
mask = ctx.OpShiftLeftLogical(ctx.U32[1], mask, ctx.Const(1U));
mask = ctx.OpShiftRightLogical(ctx.U32[1], swizzle, mask);
@@ -371,7 +371,7 @@ Id CasFunction(EmitContext& ctx, Operation operation, Id value_type) {
Id CasLoop(EmitContext& ctx, Operation operation, Id array_pointer, Id element_pointer,
Id value_type, Id memory_type, spv::Scope scope) {
const bool is_shared{scope == spv::Scope::Workgroup};
const bool is_struct{!is_shared || ctx.uses_explicit_workgroup_layout};
const bool is_struct{!is_shared || ctx.profile.support_explicit_workgroup_layout};
const Id cas_func{CasFunction(ctx, operation, value_type)};
const Id zero{ctx.u32_zero_value};
const Id scope_id{ctx.Const(static_cast<u32>(scope))};
@@ -620,11 +620,7 @@ void EmitContext::DefineSharedMemory(const IR::Program& program) {
return std::make_tuple(variable, element_pointer, pointer);
}};
uses_explicit_workgroup_layout =
profile.support_explicit_workgroup_layout &&
(!program.info.uses_int8 || profile.support_workgroup_layout_8bit_access) &&
(!program.info.uses_int16 || profile.support_workgroup_layout_16bit_access);
if (uses_explicit_workgroup_layout) {
if (profile.support_explicit_workgroup_layout) {
AddExtension("SPV_KHR_workgroup_memory_explicit_layout");
AddCapability(spv::Capability::WorkgroupMemoryExplicitLayoutKHR);
if (program.info.uses_int8) {
@@ -942,10 +938,6 @@ void EmitContext::DefineGlobalMemoryFunctions(const Info& info) {
if (!info.uses_global_memory || !profile.support_int64) {
return;
}
if (!profile.support_descriptor_aliasing) {
DefineGlobalMemoryFunctionsU32Fallback(info);
return;
}
using DefPtr = Id StorageDefinitions::*;
const Id zero{u32_zero_value};
const auto define_body{[&](DefPtr ssbo_member, Id addr, Id element_pointer, u32 shift,
@@ -1023,111 +1015,6 @@ void EmitContext::DefineGlobalMemoryFunctions(const Info& info) {
define(&StorageDefinitions::U32x4, storage_types.U32x4, U32[4], sizeof(u32[4]));
}
void EmitContext::DefineGlobalMemoryFunctionsU32Fallback(const Info& info) {
const Id zero{u32_zero_value};
const auto define_body{[&](Id addr, u32 num_words, auto&& callback) {
AddLabel();
const size_t num_buffers{info.storage_buffers_descriptors.size()};
for (size_t index = 0; index < num_buffers; ++index) {
if (!info.nvn_buffer_used[index]) {
continue;
}
const auto& ssbo{info.storage_buffers_descriptors[index]};
const u32 addr_word{ssbo.cbuf_offset / 4};
const u32 addr_lo_comp{addr_word % 4};
const Id cbuf{cbufs[ssbo.cbuf_index].U32x4};
const Id addr_vec_pointer{
OpAccessChain(uniform_types.U32x4, cbuf, zero, Const(addr_word / 4))};
const Id addr_vec{OpLoad(U32[4], addr_vec_pointer)};
const Id addr_lo{OpCompositeExtract(U32[1], addr_vec, addr_lo_comp)};
const Id addr_hi{OpCompositeExtract(U32[1], addr_vec, addr_lo_comp + 1U)};
const Id unaligned_addr{
OpBitcast(U64, OpCompositeConstruct(U32[2], addr_lo, addr_hi))};
const u64 ssbo_align_mask{~(profile.min_ssbo_alignment - 1U)};
const Id ssbo_addr{OpBitwiseAnd(U64, unaligned_addr, Constant(U64, ssbo_align_mask))};
const u32 size_word{addr_word + 2};
Id size_vec{addr_vec};
if (size_word / 4 != addr_word / 4) {
const Id size_vec_pointer{
OpAccessChain(uniform_types.U32x4, cbuf, zero, Const(size_word / 4))};
size_vec = OpLoad(U32[4], size_vec_pointer);
}
const Id ssbo_size{
OpUConvert(U64, OpCompositeExtract(U32[1], size_vec, size_word % 4))};
const Id ssbo_end{OpIAdd(U64, ssbo_addr, ssbo_size)};
const Id cond{OpLogicalAnd(U1, OpUGreaterThanEqual(U1, addr, ssbo_addr),
OpULessThan(U1, addr, ssbo_end))};
const Id then_label{OpLabel()};
const Id else_label{OpLabel()};
OpSelectionMerge(else_label, spv::SelectionControlMask::MaskNone);
OpBranchConditional(cond, then_label, else_label);
AddLabel(then_label);
const Id ssbo_id{ssbos[index].U32};
const Id ssbo_offset{OpUConvert(U32[1], OpISub(U64, addr, ssbo_addr))};
const Id base_word{OpShiftRightLogical(U32[1], ssbo_offset, Const(2U))};
std::array<Id, 4> word_pointers{};
for (u32 word = 0; word < num_words; ++word) {
const Id word_index{word == 0 ? base_word
: OpIAdd(U32[1], base_word, Const(word))};
word_pointers[word] =
OpAccessChain(storage_types.U32.element, ssbo_id, zero, word_index);
}
callback(word_pointers);
AddLabel(else_label);
}
}};
const auto define_load{[&](Id type, u32 num_words) {
const Id function_type{TypeFunction(type, U64)};
const Id func_id{OpFunction(type, spv::FunctionControlMask::MaskNone, function_type)};
const Id addr{OpFunctionParameter(U64)};
define_body(addr, num_words, [&](const std::array<Id, 4>& pointers) {
std::array<Id, 4> words{};
for (u32 word = 0; word < num_words; ++word) {
words[word] = OpLoad(U32[1], pointers[word]);
}
switch (num_words) {
case 1:
OpReturnValue(words[0]);
break;
case 2:
OpReturnValue(OpCompositeConstruct(type, words[0], words[1]));
break;
default:
OpReturnValue(
OpCompositeConstruct(type, words[0], words[1], words[2], words[3]));
break;
}
});
OpReturnValue(ConstantNull(type));
OpFunctionEnd();
return func_id;
}};
const auto define_write{[&](Id type, u32 num_words) {
const Id function_type{TypeFunction(void_id, U64, type)};
const Id func_id{OpFunction(void_id, spv::FunctionControlMask::MaskNone, function_type)};
const Id addr{OpFunctionParameter(U64)};
const Id data{OpFunctionParameter(type)};
define_body(addr, num_words, [&](const std::array<Id, 4>& pointers) {
for (u32 word = 0; word < num_words; ++word) {
const Id value{num_words == 1 ? data : OpCompositeExtract(U32[1], data, word)};
OpStore(pointers[word], value);
}
OpReturn();
});
OpReturn();
OpFunctionEnd();
return func_id;
}};
load_global_func_u32 = define_load(U32[1], 1);
load_global_func_u32x2 = define_load(U32[2], 2);
load_global_func_u32x4 = define_load(U32[4], 4);
write_global_func_u32 = define_write(U32[1], 1);
write_global_func_u32x2 = define_write(U32[2], 2);
write_global_func_u32x4 = define_write(U32[4], 4);
}
void EmitContext::DefineRescalingInput(const Info& info) {
if (!info.uses_rescaling_uniform) {
return;
@@ -1551,7 +1438,7 @@ void EmitContext::DefineInputs(const IR::Program& program) {
if (info.uses_is_helper_invocation) {
is_helper_invocation = DefineInput(*this, U1, false, spv::BuiltIn::HelperInvocation);
}
if (info.uses_subgroup_mask && profile.SupportsSubgroupStage(stage)) {
if (info.uses_subgroup_mask) {
subgroup_mask_eq = DefineInput(*this, U32[4], false, spv::BuiltIn::SubgroupEqMaskKHR);
subgroup_mask_lt = DefineInput(*this, U32[4], false, spv::BuiltIn::SubgroupLtMaskKHR);
subgroup_mask_le = DefineInput(*this, U32[4], false, spv::BuiltIn::SubgroupLeMaskKHR);
@@ -1565,10 +1452,9 @@ void EmitContext::DefineInputs(const IR::Program& program) {
Decorate(subgroup_mask_ge, spv::Decoration::Flat);
}
}
if ((info.uses_fswzadd || info.uses_subgroup_invocation_id || info.uses_subgroup_shuffles ||
(profile.warp_size_potentially_larger_than_guest &&
(info.uses_subgroup_vote || info.uses_subgroup_mask))) &&
profile.SupportsSubgroupStage(stage)) {
if (info.uses_fswzadd || info.uses_subgroup_invocation_id || info.uses_subgroup_shuffles ||
(profile.warp_size_potentially_larger_than_guest &&
(info.uses_subgroup_vote || info.uses_subgroup_mask))) {
AddCapability(spv::Capability::GroupNonUniform);
subgroup_local_invocation_id =
DefineInput(*this, U32[1], false, spv::BuiltIn::SubgroupLocalInvocationId);
@@ -310,10 +310,6 @@ public:
Id local_memory{};
/// True when this shader's shared memory uses SPV_KHR_workgroup_memory_explicit_layout.
/// False when the host lacks the extension or a width this shader accesses natively.
bool uses_explicit_workgroup_layout{};
Id shared_memory_u8{};
Id shared_memory_u16{};
Id shared_memory_u32{};
@@ -391,7 +387,6 @@ private:
void DefineAttributeMemAccess(const Info& info);
void DefineWriteStorageCasLoopFunction(const Info& info);
void DefineGlobalMemoryFunctions(const Info& info);
void DefineGlobalMemoryFunctionsU32Fallback(const Info& info);
void DefineRescalingInput(const Info& info);
void DefineRescalingInputPushConstant();
void DefineRescalingInputUniformConstant();
-10
View File
@@ -10,8 +10,6 @@
namespace Shader {
enum class Stage : u32;
struct Profile {
u32 supported_spirv{0x00010000};
bool unified_descriptor_binding{};
@@ -31,16 +29,12 @@ struct Profile {
bool support_fp32_signed_zero_nan_preserve{};
bool support_fp64_signed_zero_nan_preserve{};
bool support_explicit_workgroup_layout{};
bool support_workgroup_layout_8bit_access{};
bool support_workgroup_layout_16bit_access{};
bool support_vote{};
u32 supported_subgroup_stages{0x7F};
bool support_viewport_index_layer_non_geometry{};
bool support_viewport_mask{};
bool support_typeless_image_loads{};
bool support_demote_to_helper_invocation{};
bool support_int64_atomics{};
bool support_shared_int64_atomics{};
bool support_derivative_control{};
bool support_geometry_shader_passthrough{};
bool support_native_ndc{};
@@ -99,10 +93,6 @@ struct Profile {
u64 min_ssbo_alignment{};
u32 max_user_clip_distances{};
bool SupportsSubgroupStage(Stage stage) const {
return (supported_subgroup_stages & (1u << static_cast<u32>(stage))) != 0;
}
};
} // namespace Shader
@@ -11,9 +11,8 @@
#define BEGIN_PUSH_CONSTANTS layout(push_constant) uniform PushConstants {
#define END_PUSH_CONSTANTS };
#define UNIFORM(n)
#define BINDING_SWIZZLE_BUFFER 0
#define BINDING_INPUT_BUFFER 1
#define BINDING_OUTPUT_IMAGE 2
#define BINDING_INPUT_BUFFER 0
#define BINDING_OUTPUT_IMAGE 1
#else // ^^^ Vulkan ^^^ // vvv OpenGL vvv
@@ -26,8 +25,7 @@
#define BEGIN_PUSH_CONSTANTS
#define END_PUSH_CONSTANTS
#define UNIFORM(n) layout (location = n) uniform
#define BINDING_SWIZZLE_BUFFER 0
#define BINDING_INPUT_BUFFER 1
#define BINDING_INPUT_BUFFER 0
#define BINDING_OUTPUT_IMAGE 0
#endif
@@ -43,10 +41,6 @@ UNIFORM(6) uint block_height;
UNIFORM(7) uint block_height_mask;
END_PUSH_CONSTANTS
layout(binding = BINDING_SWIZZLE_BUFFER, std430) readonly buffer SwizzleTable {
uint swizzle_table[];
};
#if HAS_EXTENDED_TYPES
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU8 { uint8_t u8data[]; };
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU16 { uint16_t u16data[]; };
@@ -71,9 +65,19 @@ const uint GOB_SIZE_SHIFT = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_SHI
const uvec2 SWIZZLE_MASK = uvec2(GOB_SIZE_X - 1, GOB_SIZE_Y - 1);
uint SwizzleTable(uint pos) {
const uint t[8] = uint[](
0x12100200, 0x13110301, 0x16140604, 0x17150705,
0x1a180a08, 0x1b190b09, 0x1e1c0e0c, 0x1f1d0f0d
);
const uint i = pos >> 4;
const uint h = (t[i / 4] >> ((i % 4) * 8)) & 0xff;
return (h << 4) | (pos & 0xf);
}
uint SwizzleOffset(uvec2 pos) {
pos = pos & SWIZZLE_MASK;
return swizzle_table[pos.y * 64 + pos.x];
return SwizzleTable(pos.y * 64 + pos.x);
}
uvec4 ReadTexel(uint offset) {
@@ -11,9 +11,8 @@
#define BEGIN_PUSH_CONSTANTS layout(push_constant) uniform PushConstants {
#define END_PUSH_CONSTANTS };
#define UNIFORM(n)
#define BINDING_SWIZZLE_BUFFER 0
#define BINDING_INPUT_BUFFER 1
#define BINDING_OUTPUT_IMAGE 2
#define BINDING_INPUT_BUFFER 0
#define BINDING_OUTPUT_IMAGE 1
#else // ^^^ Vulkan ^^^ // vvv OpenGL vvv
@@ -26,8 +25,7 @@
#define BEGIN_PUSH_CONSTANTS
#define END_PUSH_CONSTANTS
#define UNIFORM(n) layout (location = n) uniform
#define BINDING_SWIZZLE_BUFFER 0
#define BINDING_INPUT_BUFFER 1
#define BINDING_INPUT_BUFFER 0
#define BINDING_OUTPUT_IMAGE 0
#endif
@@ -45,10 +43,6 @@ UNIFORM(8) uint block_depth;
UNIFORM(9) uint block_depth_mask;
END_PUSH_CONSTANTS
layout(binding = BINDING_SWIZZLE_BUFFER, std430) readonly buffer SwizzleTable {
uint swizzle_table[];
};
#if HAS_EXTENDED_TYPES
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU8 { uint8_t u8data[]; };
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU16 { uint16_t u16data[]; };
@@ -73,9 +67,19 @@ const uint GOB_SIZE_SHIFT = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_SHI
const uvec2 SWIZZLE_MASK = uvec2(GOB_SIZE_X - 1, GOB_SIZE_Y - 1);
uint SwizzleTable(uint pos) {
const uint t[8] = uint[](
0x12100200, 0x13110301, 0x16140604, 0x17150705,
0x1a180a08, 0x1b190b09, 0x1e1c0e0c, 0x1f1d0f0d
);
const uint i = pos >> 4;
const uint h = (t[i / 4] >> ((i % 4) * 8)) & 0xff;
return (h << 4) | (pos & 0xf);
}
uint SwizzleOffset(uvec2 pos) {
pos = pos & SWIZZLE_MASK;
return swizzle_table[pos.y * 64 + pos.x];
return SwizzleTable(pos.y * 64 + pos.x);
}
uvec4 ReadTexel(uint offset) {
@@ -10,9 +10,8 @@
#define BEGIN_PUSH_CONSTANTS layout(push_constant) uniform PushConstants {
#define END_PUSH_CONSTANTS };
#define UNIFORM(n)
#define BINDING_SWIZZLE_BUFFER 0
#define BINDING_INPUT_BUFFER 1
#define BINDING_OUTPUT_BUFFER 2
#define BINDING_INPUT_BUFFER 0
#define BINDING_OUTPUT_BUFFER 1
#else
#extension GL_NV_gpu_shader5 : enable
#ifdef GL_NV_gpu_shader5
@@ -23,7 +22,6 @@
#define BEGIN_PUSH_CONSTANTS
#define END_PUSH_CONSTANTS
#define UNIFORM(n) layout(location = n) uniform
#define BINDING_SWIZZLE_BUFFER 0
#define BINDING_INPUT_BUFFER 1
#define BINDING_OUTPUT_BUFFER 0
#endif
@@ -66,13 +64,9 @@ END_PUSH_CONSTANTS
#endif
// --- Buffers ---
layout(binding = BINDING_SWIZZLE_BUFFER, std430) readonly buffer SwizzleTable {
uint swizzle_table[];
};
#if HAS_EXTENDED_TYPES
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU8 { uint8_t u8data[]; };
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU16 { uint16_t u16data[]; };
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU8 { uint8_t u8data[]; };
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU16 { uint16_t u16data[]; };
#endif
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU32 { uint u32data[]; };
layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU64 { uvec2 u64data[]; };
@@ -96,10 +90,20 @@ const uint GOB_SIZE_Z_SHIFT = 0;
const uint GOB_SIZE_SHIFT = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_SHIFT;
const uvec2 SWIZZLE_MASK = uvec2(GOB_SIZE_X - 1u, GOB_SIZE_Y - 1u);
uint SwizzleTable(uint pos) {
const uint t[8] = uint[](
0x12100200, 0x13110301, 0x16140604, 0x17150705,
0x1a180a08, 0x1b190b09, 0x1e1c0e0c, 0x1f1d0f0d
);
const uint i = pos >> 4;
const uint h = (t[i / 4] >> ((i % 4) * 8)) & 0xff;
return (h << 4) | (pos & 0xf);
}
// --- Helpers ---
uint SwizzleOffset(uvec2 pos) {
pos &= SWIZZLE_MASK;
return swizzle_table[pos.y * 64u + pos.x];
return SwizzleTable(pos.y * 64u + pos.x);
}
uvec4 ReadTexel(uint offset) {
@@ -1,3 +1,6 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -56,10 +59,8 @@ UtilShaders::UtilShaders(ProgramManager& program_manager_)
copy_bc4_program(MakeProgram(OPENGL_COPY_BC4_COMP)),
convert_s8d24_program(MakeProgram(OPENGL_CONVERT_S8D24_COMP)),
convert_ms_to_nonms_program(MakeProgram(CONVERT_MSAA_TO_NON_MSAA_COMP)),
convert_nonms_to_ms_program(MakeProgram(CONVERT_NON_MSAA_TO_MSAA_COMP)) {
const auto swizzle_table = Tegra::Texture::MakeSwizzleTable();
swizzle_table_buffer.Create();
glNamedBufferStorage(swizzle_table_buffer.handle, sizeof(swizzle_table), &swizzle_table, 0);
convert_nonms_to_ms_program(MakeProgram(CONVERT_NON_MSAA_TO_MSAA_COMP))
{
}
UtilShaders::~UtilShaders() = default;
@@ -116,13 +117,11 @@ void UtilShaders::ASTCDecode(Image& image, const StagingBufferMap& map,
void UtilShaders::BlockLinearUpload2D(Image& image, const StagingBufferMap& map,
std::span<const SwizzleParameters> swizzles) {
static constexpr Extent3D WORKGROUP_SIZE{32, 32, 1};
static constexpr GLuint BINDING_SWIZZLE_BUFFER = 0;
static constexpr GLuint BINDING_INPUT_BUFFER = 1;
static constexpr GLuint BINDING_INPUT_BUFFER = 0;
static constexpr GLuint BINDING_OUTPUT_IMAGE = 0;
program_manager.BindComputeProgram(block_linear_unswizzle_2d_program.handle);
glFlushMappedNamedBufferRange(map.buffer, map.offset, image.guest_size_bytes);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, BINDING_SWIZZLE_BUFFER, swizzle_table_buffer.handle);
const GLenum store_format = StoreFormat(BytesPerBlock(image.info.format));
for (const SwizzleParameters& swizzle : swizzles) {
@@ -153,14 +152,11 @@ void UtilShaders::BlockLinearUpload2D(Image& image, const StagingBufferMap& map,
void UtilShaders::BlockLinearUpload3D(Image& image, const StagingBufferMap& map,
std::span<const SwizzleParameters> swizzles) {
static constexpr Extent3D WORKGROUP_SIZE{16, 8, 8};
static constexpr GLuint BINDING_SWIZZLE_BUFFER = 0;
static constexpr GLuint BINDING_INPUT_BUFFER = 1;
static constexpr GLuint BINDING_INPUT_BUFFER = 0;
static constexpr GLuint BINDING_OUTPUT_IMAGE = 0;
glFlushMappedNamedBufferRange(map.buffer, map.offset, image.guest_size_bytes);
program_manager.BindComputeProgram(block_linear_unswizzle_3d_program.handle);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, BINDING_SWIZZLE_BUFFER, swizzle_table_buffer.handle);
const GLenum store_format = StoreFormat(BytesPerBlock(image.info.format));
for (const SwizzleParameters& swizzle : swizzles) {
@@ -1,3 +1,6 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -45,9 +48,6 @@ public:
private:
ProgramManager& program_manager;
OGLBuffer swizzle_table_buffer;
OGLProgram astc_decoder_program;
OGLProgram block_linear_unswizzle_2d_program;
OGLProgram block_linear_unswizzle_3d_program;
@@ -653,71 +653,8 @@ void ASTCDecoderPass::Assemble(Image& image, const StagingBufferRef& map,
scheduler.Finish();
}
constexpr u32 BL3D_BINDING_SWIZZLE_TABLE = 0;
constexpr u32 BL3D_BINDING_INPUT_BUFFER = 1;
constexpr u32 BL3D_BINDING_OUTPUT_BUFFER = 2;
constexpr std::array<VkDescriptorSetLayoutBinding, 3> BL3D_DESCRIPTOR_SET_BINDINGS{{
{
.binding = BL3D_BINDING_SWIZZLE_TABLE,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER, // swizzle_table[]
.descriptorCount = 1,
.stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
.pImmutableSamplers = nullptr,
},
{
.binding = BL3D_BINDING_INPUT_BUFFER,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER, // block-linear input
.descriptorCount = 1,
.stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
.pImmutableSamplers = nullptr,
},
{
.binding = BL3D_BINDING_OUTPUT_BUFFER,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
.descriptorCount = 1,
.stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
.pImmutableSamplers = nullptr,
},
}};
constexpr DescriptorBankInfo BL3D_BANK_INFO{
.uniform_buffers = 0,
.storage_buffers = 3,
.texture_buffers = 0,
.image_buffers = 0,
.textures = 0,
.images = 0,
.score = 3,
};
constexpr std::array<VkDescriptorUpdateTemplateEntry, 3>
BL3D_DESCRIPTOR_UPDATE_TEMPLATE_ENTRY{{
{
.dstBinding = BL3D_BINDING_SWIZZLE_TABLE,
.dstArrayElement = 0,
.descriptorCount = 1,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
.offset = BL3D_BINDING_SWIZZLE_TABLE * sizeof(DescriptorUpdateEntry),
.stride = sizeof(DescriptorUpdateEntry),
},
{
.dstBinding = BL3D_BINDING_INPUT_BUFFER,
.dstArrayElement = 0,
.descriptorCount = 1,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
.offset = BL3D_BINDING_INPUT_BUFFER * sizeof(DescriptorUpdateEntry),
.stride = sizeof(DescriptorUpdateEntry),
},
{
.dstBinding = BL3D_BINDING_OUTPUT_BUFFER,
.dstArrayElement = 0,
.descriptorCount = 1,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
.offset = BL3D_BINDING_OUTPUT_BUFFER * sizeof(DescriptorUpdateEntry),
.stride = sizeof(DescriptorUpdateEntry),
}
}};
constexpr u32 BL3D_BINDING_INPUT_BUFFER = 0;
constexpr u32 BL3D_BINDING_OUTPUT_BUFFER = 1;
struct alignas(16) BlockLinearUnswizzle3DPushConstants {
u32 blocks_dim[3]; // Offset 0
@@ -745,11 +682,50 @@ BlockLinearUnswizzle3DPass::BlockLinearUnswizzle3DPass(
DescriptorPool& descriptor_pool_,
StagingBufferPool& staging_buffer_pool_,
ComputePassDescriptorQueue& compute_pass_descriptor_queue_)
: ComputePass(
device_, scheduler_, descriptor_pool_,
BL3D_DESCRIPTOR_SET_BINDINGS,
BL3D_DESCRIPTOR_UPDATE_TEMPLATE_ENTRY,
BL3D_BANK_INFO,
: ComputePass(device_, scheduler_, descriptor_pool_,
std::array<VkDescriptorSetLayoutBinding, 2>{{
{
.binding = BL3D_BINDING_INPUT_BUFFER,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER, // block-linear input
.descriptorCount = 1,
.stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
.pImmutableSamplers = nullptr,
},
{
.binding = BL3D_BINDING_OUTPUT_BUFFER,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
.descriptorCount = 1,
.stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
.pImmutableSamplers = nullptr,
},
}},
std::array<VkDescriptorUpdateTemplateEntry, 2>{{
{
.dstBinding = BL3D_BINDING_INPUT_BUFFER,
.dstArrayElement = 0,
.descriptorCount = 1,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
.offset = BL3D_BINDING_INPUT_BUFFER * sizeof(DescriptorUpdateEntry),
.stride = sizeof(DescriptorUpdateEntry),
},
{
.dstBinding = BL3D_BINDING_OUTPUT_BUFFER,
.dstArrayElement = 0,
.descriptorCount = 1,
.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
.offset = BL3D_BINDING_OUTPUT_BUFFER * sizeof(DescriptorUpdateEntry),
.stride = sizeof(DescriptorUpdateEntry),
}
}},
DescriptorBankInfo{
.uniform_buffers = 0,
.storage_buffers = 2,
.texture_buffers = 0,
.image_buffers = 0,
.textures = 0,
.images = 0,
.score = 2,
},
COMPUTE_PUSH_CONSTANT_RANGE<sizeof(BlockLinearUnswizzle3DPushConstants)>,
BLOCK_LINEAR_UNSWIZZLE_3D_BCN_COMP_SPV),
scheduler{scheduler_},
@@ -822,8 +798,6 @@ void BlockLinearUnswizzle3DPass::UnswizzleChunk(
pc.blocks_dim[2] = z_count; // Only process the count
compute_pass_descriptor_queue.Acquire(scheduler, 3);
compute_pass_descriptor_queue.AddBuffer(*image.runtime->swizzle_table_buffer, 0,
image.runtime->swizzle_table_size);
compute_pass_descriptor_queue.AddBuffer(swizzled.buffer,
sw.buffer_offset + swizzled.offset,
image.guest_size_bytes - sw.buffer_offset);
@@ -378,20 +378,6 @@ PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
serialization_thread(1, "VkPipelineSerialization") {
const auto& float_control{device.FloatControlProperties()};
const VkDriverId driver_id{device.GetDriverID()};
const VkShaderStageFlags subgroup_stages{device.GetSubgroupSupportedStages()};
const auto subgroup_stage_bit{[subgroup_stages](VkShaderStageFlags flag, Shader::Stage stage) {
return (subgroup_stages & flag) != 0 ? (1u << static_cast<u32>(stage)) : 0u;
}};
const u32 supported_subgroup_stages{
subgroup_stage_bit(VK_SHADER_STAGE_VERTEX_BIT, Shader::Stage::VertexA) |
subgroup_stage_bit(VK_SHADER_STAGE_VERTEX_BIT, Shader::Stage::VertexB) |
subgroup_stage_bit(VK_SHADER_STAGE_TESSELLATION_CONTROL_BIT,
Shader::Stage::TessellationControl) |
subgroup_stage_bit(VK_SHADER_STAGE_TESSELLATION_EVALUATION_BIT,
Shader::Stage::TessellationEval) |
subgroup_stage_bit(VK_SHADER_STAGE_GEOMETRY_BIT, Shader::Stage::Geometry) |
subgroup_stage_bit(VK_SHADER_STAGE_FRAGMENT_BIT, Shader::Stage::Fragment) |
subgroup_stage_bit(VK_SHADER_STAGE_COMPUTE_BIT, Shader::Stage::Compute)};
profile = Shader::Profile{
.supported_spirv = device.SupportedSpirvVersion(),
.unified_descriptor_binding = true,
@@ -416,12 +402,7 @@ PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
.support_fp64_signed_zero_nan_preserve =
float_control.shaderSignedZeroInfNanPreserveFloat64 != VK_FALSE,
.support_explicit_workgroup_layout = device.IsKhrWorkgroupMemoryExplicitLayoutSupported(),
.support_workgroup_layout_8bit_access =
device.IsWorkgroupMemoryExplicitLayout8BitSupported(),
.support_workgroup_layout_16bit_access =
device.IsWorkgroupMemoryExplicitLayout16BitSupported(),
.support_vote = device.IsSubgroupFeatureSupported(VK_SUBGROUP_FEATURE_VOTE_BIT),
.supported_subgroup_stages = supported_subgroup_stages,
.support_viewport_index_layer_non_geometry =
device.IsExtShaderViewportIndexLayerSupported(),
.support_viewport_mask = device.IsNvViewportArray2Supported(),
@@ -429,7 +410,6 @@ PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
.support_demote_to_helper_invocation =
device.IsExtShaderDemoteToHelperInvocationSupported(),
.support_int64_atomics = device.IsExtShaderAtomicInt64Supported(),
.support_shared_int64_atomics = device.IsSharedInt64AtomicsSupported(),
.support_derivative_control = true,
.support_geometry_shader_passthrough = device.IsNvGeometryShaderPassthroughSupported(),
.support_native_ndc = device.IsExtDepthClipControlSupported(),
@@ -453,8 +433,7 @@ PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
.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 ||
driver_id == VK_DRIVER_ID_QUALCOMM_PROPRIETARY,
.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 = false,
.has_broken_robust =
@@ -925,8 +904,12 @@ std::unique_ptr<ComputePipeline> PipelineCache::CreateComputePipeline(
}
auto program{TranslateProgram(pools.inst, pools.block, env, cfg, host_info)};
const VkDriverIdKHR driver_id = device.GetDriverID();
const bool needs_shared_mem_clamp =
driver_id == VK_DRIVER_ID_QUALCOMM_PROPRIETARY ||
driver_id == VK_DRIVER_ID_ARM_PROPRIETARY;
const u32 max_shared_memory = device.GetMaxComputeSharedMemorySize();
if (program.shared_memory_size > max_shared_memory) {
if (needs_shared_mem_clamp && program.shared_memory_size > max_shared_memory) {
LOG_WARNING(Render_Vulkan,
"Compute shader 0x{:016x} requests {}KB shared memory but device max is {}KB - clamping",
key.unique_hash,
@@ -1255,18 +1255,6 @@ void RasterizerVulkan::UpdateDepthBias(Tegra::Engines::Maxwell3D::Regs& regs) {
}
}
const bool is_float_depth =
regs.zeta.format == Tegra::DepthFormat::Z32_FLOAT ||
regs.zeta.format == Tegra::DepthFormat::Z32_FLOAT_X24S8_UINT;
if (is_float_depth && units != 0.0f && !device.IsExtDepthBiasControlSupported()) {
static bool logged_float_bias_warning = false;
if (!logged_float_bias_warning) {
logged_float_bias_warning = true;
LOG_WARNING(Render_Vulkan,
"Depth bias on a float depth target without VK_EXT_depth_bias_control");
}
}
scheduler.Record([constant = units, clamp = regs.depth_bias_clamp,
factor = regs.slope_scale_depth_bias, this](vk::CommandBuffer cmdbuf) {
if (device.IsExtDepthBiasControlSupported()) {
@@ -909,40 +909,6 @@ TextureCacheRuntime::TextureCacheRuntime(const Device& device_, Scheduler& sched
bl3d_unswizzle_pass.emplace(device, scheduler, descriptor_pool,
staging_buffer_pool, compute_pass_descriptor_queue);
}
// --- Create swizzle table buffer ---
{
auto table = Tegra::Texture::MakeSwizzleTable();
swizzle_table_size = static_cast<VkDeviceSize>(table.size() * sizeof(table[0]));
auto staging = staging_buffer_pool.Request(swizzle_table_size, MemoryUsage::Upload);
std::memcpy(staging.mapped_span.data(), table.data(), static_cast<size_t>(swizzle_table_size));
VkBufferCreateInfo ci{
.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
.size = swizzle_table_size,
.usage = VK_BUFFER_USAGE_STORAGE_BUFFER_BIT |
VK_BUFFER_USAGE_TRANSFER_DST_BIT |
VK_BUFFER_USAGE_TRANSFER_SRC_BIT,
.sharingMode = VK_SHARING_MODE_EXCLUSIVE,
};
swizzle_table_buffer = memory_allocator.CreateBuffer(ci, MemoryUsage::DeviceLocal);
scheduler.RequestOutsideRenderPassOperationContext();
scheduler.Record([staging_buf = staging.buffer,
dst_buf = *swizzle_table_buffer,
size = swizzle_table_size,
src_off = staging.offset](vk::CommandBuffer cmdbuf) {
const VkBufferCopy region{
.srcOffset = src_off,
.dstOffset = 0,
.size = size,
};
cmdbuf.CopyBuffer(staging_buf, dst_buf, region);
});
}
}
void TextureCacheRuntime::Finish() {
@@ -2484,18 +2450,13 @@ void TextureCacheRuntime::AccelerateImageUpload(
if (!Settings::values.gpu_unswizzle_enabled.GetValue() || !bl3d_unswizzle_pass) {
if (IsPixelFormatBCn(image.info.format) && image.info.type == ImageType::e3D) {
ASSERT_MSG(false, "GPU unswizzle is disabled for BCn 3D texture");
ASSERT(false && "GPU unswizzle is disabled for BCn 3D texture");
}
ASSERT(false);
return;
}
if (bl3d_unswizzle_pass &&
IsPixelFormatBCn(image.info.format) &&
image.info.type == ImageType::e3D &&
image.info.resources.levels == 1 &&
image.info.resources.layers == 1) {
if (bl3d_unswizzle_pass && IsPixelFormatBCn(image.info.format) && image.info.type == ImageType::e3D && image.info.resources.levels == 1 && image.info.resources.layers == 1) {
return bl3d_unswizzle_pass->Unswizzle(image, map, swizzles, z_start, z_count);
}
@@ -1,4 +1,4 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2019 yuzu Emulator Project
@@ -130,9 +130,6 @@ public:
std::optional<ASTCDecoderPass> astc_decoder_pass;
std::optional<BlockLinearUnswizzle3DPass> bl3d_unswizzle_pass;
vk::Buffer swizzle_table_buffer;
VkDeviceSize swizzle_table_size = 0;
std::optional<MSAACopyPass> msaa_copy_pass;
const Settings::ResolutionScalingInfo& resolution;
std::array<std::vector<VkFormat>, VideoCore::Surface::MaxPixelFormat> view_formats;
+5 -33
View File
@@ -77,35 +77,10 @@ TextureCache<P>::TextureCache(Runtime& runtime_, Tegra::MaxwellDeviceMemoryManag
minimum_memory = 0;
}
const bool gpu_unswizzle_enabled = Settings::values.gpu_unswizzle_enabled.GetValue();
if (gpu_unswizzle_enabled) {
switch (Settings::values.gpu_unswizzle_texture_size.GetValue()) {
case Settings::GpuUnswizzleSize::VerySmall: gpu_unswizzle_maxsize = 16_MiB; break;
case Settings::GpuUnswizzleSize::Small: gpu_unswizzle_maxsize = 32_MiB; break;
case Settings::GpuUnswizzleSize::Normal: gpu_unswizzle_maxsize = 128_MiB; break;
case Settings::GpuUnswizzleSize::Large: gpu_unswizzle_maxsize = 256_MiB; break;
case Settings::GpuUnswizzleSize::VeryLarge: gpu_unswizzle_maxsize = 512_MiB; break;
default: gpu_unswizzle_maxsize = 128_MiB; break;
}
switch (Settings::values.gpu_unswizzle_stream_size.GetValue()) {
case Settings::GpuUnswizzle::VeryLow: swizzle_chunk_size = 4_MiB; break;
case Settings::GpuUnswizzle::Low: swizzle_chunk_size = 8_MiB; break;
case Settings::GpuUnswizzle::Normal: swizzle_chunk_size = 16_MiB; break;
case Settings::GpuUnswizzle::Medium: swizzle_chunk_size = 32_MiB; break;
case Settings::GpuUnswizzle::High: swizzle_chunk_size = 64_MiB; break;
default: swizzle_chunk_size = 16_MiB;
}
switch (Settings::values.gpu_unswizzle_chunk_size.GetValue()) {
case Settings::GpuUnswizzleChunk::VeryLow: swizzle_slices_per_batch = 32; break;
case Settings::GpuUnswizzleChunk::Low: swizzle_slices_per_batch = 64; break;
case Settings::GpuUnswizzleChunk::Normal: swizzle_slices_per_batch = 128; break;
case Settings::GpuUnswizzleChunk::Medium: swizzle_slices_per_batch = 256; break;
case Settings::GpuUnswizzleChunk::High: swizzle_slices_per_batch = 512; break;
default: swizzle_slices_per_batch = 128;
}
if (Settings::values.gpu_unswizzle_enabled.GetValue()) {
gpu_unswizzle_maxsize = (1 << u32(Settings::values.gpu_unswizzle_texture_size.GetValue())) * 1_MiB;
swizzle_chunk_size = (1 << u32(Settings::values.gpu_unswizzle_stream_size.GetValue())) * 1_MiB;
swizzle_slices_per_batch = 1 << u32(Settings::values.gpu_unswizzle_chunk_size.GetValue());
} else {
gpu_unswizzle_maxsize = 0;
swizzle_chunk_size = 0;
@@ -1915,16 +1890,13 @@ void TextureCache<P>::TrimInactiveSamplers(size_t budget) {
ankerl::unordered_dense::set<SamplerId> active_sampler_ids;
for (auto const& e : channel_state->sampler_ids)
active_sampler_ids.insert(e.second);
if constexpr (requires { runtime.Finish(); }) {
runtime.Finish();
}
// Elements in the map must be necesarily valid
size_t removed = 0;
for (auto it = channel_state->samplers.begin(); it != channel_state->samplers.end();) {
const SamplerId sampler_id = it->second;
if (!sampler_id || sampler_id == CORRUPT_ID) {
it = channel_state->samplers.erase(it);
} else if (active_sampler_ids.contains(sampler_id)) {
} else if (std::ranges::find(active_sampler_ids, sampler_id) != active_sampler_ids.end()) {
++it;
} else {
slot_samplers.erase(sampler_id);
+3 -18
View File
@@ -1,3 +1,6 @@
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -23,24 +26,6 @@ constexpr u32 GOB_SIZE_SHIFT = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_
constexpr u32 SWIZZLE_X_BITS = 0b100101111;
constexpr u32 SWIZZLE_Y_BITS = 0b011010000;
using SwizzleTable = std::array<std::array<u32, GOB_SIZE_X>, GOB_SIZE_Y>;
/**
* This table represents the internal swizzle of a gob, in format 16 bytes x 2 sector packing.
* Calculates the offset of an (x, y) position within a swizzled texture.
* Taken from the Tegra X1 Technical Reference Manual. pages 1187-1188
*/
constexpr SwizzleTable MakeSwizzleTable() {
SwizzleTable table{};
for (u32 y = 0; y < table.size(); ++y) {
for (u32 x = 0; x < table[0].size(); ++x) {
table[y][x] = ((x % 64) / 32) * 256 + ((y % 8) / 2) * 64 + ((x % 32) / 16) * 32 +
(y % 2) * 16 + (x % 16);
}
}
return table;
}
/// Unswizzles a block linear texture into linear memory.
void UnswizzleTexture(std::span<u8> output, std::span<const u8> input, u32 bytes_per_pixel,
u32 width, u32 height, u32 depth, u32 block_height, u32 block_depth,
+31 -5
View File
@@ -501,6 +501,14 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
LOG_WARNING(Render_Vulkan,
"Qualcomm drivers have slow push descriptor implementation");
RemoveExtension(extensions.push_descriptor, VK_KHR_PUSH_DESCRIPTOR_EXTENSION_NAME);
LOG_WARNING(Render_Vulkan,
"Disabling shader float controls and 64-bit integer features on Qualcomm proprietary drivers");
RemoveExtension(extensions.shader_float_controls, VK_KHR_SHADER_FLOAT_CONTROLS_EXTENSION_NAME);
RemoveExtensionFeature(extensions.shader_atomic_int64, features.shader_atomic_int64,
VK_KHR_SHADER_ATOMIC_INT64_EXTENSION_NAME);
features.shader_atomic_int64.shaderBufferInt64Atomics = false;
features.shader_atomic_int64.shaderSharedInt64Atomics = false;
features.features.shaderInt64 = false;
#if defined(__ANDROID__) && defined(ARCHITECTURE_arm64)
// BCn patching only safe on Android 9+ (API 28+). Older versions crash on driver load.
@@ -553,6 +561,7 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
features.shader_float16_int8.shaderFloat16 = false;
}
// Mali/ NVIDIA proprietary drivers: Shader stencil export not supported
// Use hardware depth/stencil blits instead when available
if (!extensions.shader_stencil_export) {
LOG_INFO(Render_Vulkan,
@@ -565,8 +574,8 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
if (!is_blit_depth24_stencil8_supported && !is_blit_depth32_stencil8_supported) {
LOG_WARNING(Render_Vulkan,
"Neither shader export nor hardware blits available for "
"depth/stencil.");
"NVIDIA: Neither shader export nor hardware blits available for "
"depth/stencil. Performance may be degraded.");
}
}
}
@@ -644,13 +653,21 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
if (is_turnip || is_qualcomm) {
LOG_WARNING(Render_Vulkan, "Driver requires higher-than-reported binding limits");
properties.properties.limits.maxVertexInputBindings =
(std::max)(properties.properties.limits.maxVertexInputBindings, 32U);
properties.properties.limits.maxVertexInputBindings = 32;
}
const auto dyna_state = Settings::values.dyna_state.GetValue();
// Base dynamic states (VIEWPORT, SCISSOR, DEPTH_BIAS, etc.) are ALWAYS active in vk_graphics_pipeline.cpp
// This slider controls EXTENDED dynamic states with accumulative levels per Vulkan specs:
// Level 0 = Core Dynamic States only (Vulkan 1.0)
// Level 1 = Core + VK_EXT_extended_dynamic_state
// Level 2 = Core + VK_EXT_extended_dynamic_state + VK_EXT_extended_dynamic_state2
// Level 3 = Core + VK_EXT_extended_dynamic_state + VK_EXT_extended_dynamic_state2 + VK_EXT_extended_dynamic_state3
switch (dyna_state) {
case Settings::ExtendedDynamicState::Disabled:
// Level 0: Disable all extended dynamic state extensions
RemoveExtensionFeature(extensions.extended_dynamic_state, features.extended_dynamic_state,
VK_EXT_EXTENDED_DYNAMIC_STATE_EXTENSION_NAME);
RemoveExtensionFeature(extensions.extended_dynamic_state2, features.extended_dynamic_state2,
@@ -661,6 +678,7 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
dynamic_state3_enables = false;
break;
case Settings::ExtendedDynamicState::EDS1:
// Level 1: Enable EDS1, disable EDS2 and EDS3
RemoveExtensionFeature(extensions.extended_dynamic_state2, features.extended_dynamic_state2,
VK_EXT_EXTENDED_DYNAMIC_STATE_2_EXTENSION_NAME);
RemoveExtensionFeature(extensions.extended_dynamic_state3, features.extended_dynamic_state3,
@@ -669,6 +687,7 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
dynamic_state3_enables = false;
break;
case Settings::ExtendedDynamicState::EDS2:
// Level 2: Enable EDS1 + EDS2, disable EDS3
RemoveExtensionFeature(extensions.extended_dynamic_state3, features.extended_dynamic_state3,
VK_EXT_EXTENDED_DYNAMIC_STATE_3_EXTENSION_NAME);
dynamic_state3_blending = false;
@@ -676,9 +695,12 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
break;
case Settings::ExtendedDynamicState::EDS3:
default:
// Level 3: Enable all (EDS1 + EDS2 + EDS3)
break;
}
// VK_EXT_vertex_input_dynamic_state is independent from EDS
// It can be enabled even without extended_dynamic_state
if (!Settings::values.vertex_input_dynamic_state.GetValue()) {
RemoveExtensionFeature(extensions.vertex_input_dynamic_state, features.vertex_input_dynamic_state, VK_EXT_VERTEX_INPUT_DYNAMIC_STATE_EXTENSION_NAME);
}
@@ -1260,7 +1282,8 @@ void Device::RemoveUnsuitableExtensions() {
VK_EXT_IMAGE_ROBUSTNESS_EXTENSION_NAME);
// VK_KHR_shader_atomic_int64
extensions.shader_atomic_int64 = features.shader_atomic_int64.shaderBufferInt64Atomics;
extensions.shader_atomic_int64 = features.shader_atomic_int64.shaderBufferInt64Atomics &&
features.shader_atomic_int64.shaderSharedInt64Atomics;
RemoveExtensionFeatureIfUnsuitable(extensions.shader_atomic_int64, features.shader_atomic_int64,
VK_KHR_SHADER_ATOMIC_INT64_EXTENSION_NAME);
@@ -1309,7 +1332,10 @@ void Device::RemoveUnsuitableExtensions() {
// VK_KHR_workgroup_memory_explicit_layout
extensions.workgroup_memory_explicit_layout =
features.features.shaderInt16 &&
features.workgroup_memory_explicit_layout.workgroupMemoryExplicitLayout &&
features.workgroup_memory_explicit_layout.workgroupMemoryExplicitLayout8BitAccess &&
features.workgroup_memory_explicit_layout.workgroupMemoryExplicitLayout16BitAccess &&
features.workgroup_memory_explicit_layout.workgroupMemoryExplicitLayoutScalarBlockLayout;
RemoveExtensionFeatureIfUnsuitable(extensions.workgroup_memory_explicit_layout,
features.workgroup_memory_explicit_layout,
+2 -26
View File
@@ -403,11 +403,6 @@ FN_MAX_LIMIT_LIST
return properties.subgroup_properties.supportedOperations & feature;
}
/// Returns the shader stages that support subgroup operations.
VkShaderStageFlags GetSubgroupSupportedStages() const {
return properties.subgroup_properties.supportedStages;
}
/// Returns the maximum number of push descriptors.
u32 MaxPushDescriptors() const {
return properties.push_descriptor.maxPushDescriptors;
@@ -488,18 +483,6 @@ FN_MAX_LIMIT_LIST
return extensions.workgroup_memory_explicit_layout;
}
/// Returns true if the device supports 8-bit accesses to workgroup explicit layout memory.
bool IsWorkgroupMemoryExplicitLayout8BitSupported() const {
return extensions.workgroup_memory_explicit_layout &&
features.workgroup_memory_explicit_layout.workgroupMemoryExplicitLayout8BitAccess;
}
/// Returns true if the device supports 16-bit accesses to workgroup explicit layout memory.
bool IsWorkgroupMemoryExplicitLayout16BitSupported() const {
return extensions.workgroup_memory_explicit_layout && features.features.shaderInt16 &&
features.workgroup_memory_explicit_layout.workgroupMemoryExplicitLayout16BitAccess;
}
/// Returns true if the device supports VK_KHR_image_format_list.
bool IsKhrImageFormatListSupported() const {
return extensions.image_format_list || instance_version >= VK_API_VERSION_1_2;
@@ -728,16 +711,9 @@ FN_MAX_LIMIT_LIST
features.provoking_vertex.transformFeedbackPreservesProvokingVertex;
}
/// Returns true if the device supports int64 atomics on storage buffers.
/// Returns true if the device supports VK_KHR_shader_atomic_int64.
bool IsExtShaderAtomicInt64Supported() const {
return extensions.shader_atomic_int64 &&
features.shader_atomic_int64.shaderBufferInt64Atomics;
}
/// Returns true if the device supports int64 atomics on workgroup (shared) memory.
bool IsSharedInt64AtomicsSupported() const {
return extensions.shader_atomic_int64 &&
features.shader_atomic_int64.shaderSharedInt64Atomics;
return extensions.shader_atomic_int64;
}
bool IsExtConditionalRendering() const {