vk_compute_pass: Implement indexed quads

Implement indexed quads (GL_QUADS used with glDrawElements*) with a
compute pass conversion.

The compute shader converts from uint8/uint16/uint32 indices to uint32.
The format is passed through push constants to avoid having different
variants of the same shader.

- Used by Fast RMX
- Used by Xenoblade Chronicles 2 (it still has graphical due to
synchronization issues on Vulkan)

1 files changed, 50 insertions, 0 deletions

diff --git a/src/video_core/renderer_vulkan/shaders/quad_indexed.comp b/src/video_core/renderer_vulkan/shaders/quad_indexed.comp
new file mode 100644
index 000000000..5a472ba9b
--- /dev/null
+++ b/src/video_core/renderer_vulkan/shaders/quad_indexed.comp
@@ -0,0 +1,50 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+/*
+ * Build instructions:
+ * $ glslangValidator -V quad_indexed.comp -o output.spv
+ * $ spirv-opt -O --strip-debug output.spv -o optimized.spv
+ * $ xxd -i optimized.spv
+ *
+ * Then copy that bytecode to the C++ file
+ */
+
+#version 460 core
+
+layout (local_size_x = 1024) in;
+
+layout (std430, set = 0, binding = 0) readonly buffer InputBuffer {
+    uint input_indexes[];
+};
+
+layout (std430, set = 0, binding = 1) writeonly buffer OutputBuffer {
+    uint output_indexes[];
+};
+
+layout (push_constant) uniform PushConstants {
+    uint base_vertex;
+    int index_shift; // 0: uint8, 1: uint16, 2: uint32
+};
+
+void main() {
+    int primitive = int(gl_GlobalInvocationID.x);
+    if (primitive * 6 >= output_indexes.length()) {
+        return;
+    }
+
+    int index_size = 8 << index_shift;
+    int flipped_shift = 2 - index_shift;
+    int mask = (1 << flipped_shift) - 1;
+
+    const int quad_swizzle[6] = int[](0, 1, 2, 0, 2, 3);
+    for (uint vertex = 0; vertex < 6; ++vertex) {
+        int offset = primitive * 4 + quad_swizzle[vertex];
+        int int_offset = offset >> flipped_shift;
+        int bit_offset = (offset & mask) * index_size;
+        uint packed_input = input_indexes[int_offset];
+        uint index = bitfieldExtract(packed_input, bit_offset, index_size);
+        output_indexes[primitive * 6 + vertex] = index + base_vertex;
+    }
+}