Fully support all 16-bit packed texture formats
Vulkan lists the following formats as 16-bit packed formats:
* VK_FORMAT_R4G4B4A4_UNORM_PACK16
* VK_FORMAT_B4G4R4A4_UNORM_PACK16
* VK_FORMAT_R5G6B5_UNORM_PACK16
* VK_FORMAT_B5G6R5_UNORM_PACK16
* VK_FORMAT_R5G5B5A1_UNORM_PACK16
* VK_FORMAT_B5G5R5A1_UNORM_PACK16
* VK_FORMAT_A1R5G5B5_UNORM_PACK16
We already support VK_FORMAT_A1R5G5B5_UNORM_PACK16, so we can use that
as a crossreference to see which tests we expect to pass.
Bug: b/139351376
Tests: dEQP-VK.*pack16*
Change-Id: Ia7b0b9efe92c7f67c803579fee7c8e0d62ef081f
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/57048
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Kokoro-Result: kokoro <noreply+kokoro@google.com>
Tested-by: Sean Risser <srisser@google.com>
Commit-Queue: Sean Risser <srisser@google.com>
diff --git a/src/Pipeline/SamplerCore.cpp b/src/Pipeline/SamplerCore.cpp
index d821e46..9043038 100644
--- a/src/Pipeline/SamplerCore.cpp
+++ b/src/Pipeline/SamplerCore.cpp
@@ -152,16 +152,20 @@
switch(state.textureFormat)
{
case VK_FORMAT_R5G6B5_UNORM_PACK16:
+ case VK_FORMAT_B5G6R5_UNORM_PACK16:
c.x *= Float4(1.0f / 0xF800);
c.y *= Float4(1.0f / 0xFC00);
c.z *= Float4(1.0f / 0xF800);
break;
+ case VK_FORMAT_R4G4B4A4_UNORM_PACK16:
case VK_FORMAT_B4G4R4A4_UNORM_PACK16:
c.x *= Float4(1.0f / 0xF000);
c.y *= Float4(1.0f / 0xF000);
c.z *= Float4(1.0f / 0xF000);
c.w *= Float4(1.0f / 0xF000);
break;
+ case VK_FORMAT_R5G5B5A1_UNORM_PACK16:
+ case VK_FORMAT_B5G5R5A1_UNORM_PACK16:
case VK_FORMAT_A1R5G5B5_UNORM_PACK16:
c.x *= Float4(1.0f / 0xF800);
c.y *= Float4(1.0f / 0xF800);
@@ -214,16 +218,20 @@
switch(state.textureFormat)
{
case VK_FORMAT_R5G6B5_UNORM_PACK16:
+ case VK_FORMAT_B5G6R5_UNORM_PACK16:
c.x = Float4(As<UShort4>(cs.x)) * Float4(1.0f / 0xF800);
c.y = Float4(As<UShort4>(cs.y)) * Float4(1.0f / 0xFC00);
c.z = Float4(As<UShort4>(cs.z)) * Float4(1.0f / 0xF800);
break;
+ case VK_FORMAT_R4G4B4A4_UNORM_PACK16:
case VK_FORMAT_B4G4R4A4_UNORM_PACK16:
c.x = Float4(As<UShort4>(cs.x)) * Float4(1.0f / 0xF000);
c.y = Float4(As<UShort4>(cs.y)) * Float4(1.0f / 0xF000);
c.z = Float4(As<UShort4>(cs.z)) * Float4(1.0f / 0xF000);
c.w = Float4(As<UShort4>(cs.w)) * Float4(1.0f / 0xF000);
break;
+ case VK_FORMAT_R5G5B5A1_UNORM_PACK16:
+ case VK_FORMAT_B5G5R5A1_UNORM_PACK16:
case VK_FORMAT_A1R5G5B5_UNORM_PACK16:
c.x = Float4(As<UShort4>(cs.x)) * Float4(1.0f / 0xF800);
c.y = Float4(As<UShort4>(cs.y)) * Float4(1.0f / 0xF800);
@@ -1526,12 +1534,35 @@
c.y = (c.x & Short4(0x07E0u)) << 5;
c.x = (c.x & Short4(0xF800u));
break;
+ case VK_FORMAT_B5G6R5_UNORM_PACK16:
+ c.z = (c.x & Short4(0xF800u));
+ c.y = (c.x & Short4(0x07E0u)) << 5;
+ c.x = (c.x & Short4(0x001Fu)) << 11;
+ break;
+ case VK_FORMAT_R4G4B4A4_UNORM_PACK16:
+ c.w = (c.x << 12) & Short4(0xF000u);
+ c.z = (c.x << 8) & Short4(0xF000u);
+ c.y = (c.x << 4) & Short4(0xF000u);
+ c.x = (c.x) & Short4(0xF000u);
+ break;
case VK_FORMAT_B4G4R4A4_UNORM_PACK16:
c.w = (c.x << 12) & Short4(0xF000u);
c.z = (c.x) & Short4(0xF000u);
c.y = (c.x << 4) & Short4(0xF000u);
c.x = (c.x << 8) & Short4(0xF000u);
break;
+ case VK_FORMAT_R5G5B5A1_UNORM_PACK16:
+ c.w = (c.x << 15) & Short4(0x8000u);
+ c.z = (c.x << 10) & Short4(0xF800u);
+ c.y = (c.x << 5) & Short4(0xF800u);
+ c.x = (c.x) & Short4(0xF800u);
+ break;
+ case VK_FORMAT_B5G5R5A1_UNORM_PACK16:
+ c.w = (c.x << 15) & Short4(0x8000u);
+ c.z = (c.x) & Short4(0xF800u);
+ c.y = (c.x << 5) & Short4(0xF800u);
+ c.x = (c.x << 10) & Short4(0xF800u);
+ break;
case VK_FORMAT_A1R5G5B5_UNORM_PACK16:
c.w = (c.x) & Short4(0x8000u);
c.z = (c.x << 11) & Short4(0xF800u);
