alex/vello
1
0
Fork 0
mirror of https://github.com/italicsjenga/vello.git synced 2024-10-18 07:21:30 +11:00
Code Issues Packages Projects Releases Wiki Activity
vello/piet-gpu/shader/gen/transform_root.hlsl

95 lines
2.6 KiB
HLSL
Raw Normal View History

Start work on new element pipeline There's a bit of reorganizing as well. Shader stages are made available from piet-gpu to the test rig, config is now a proper structure (marshaled with bytemuck). This commit just has the transform stage, which is a simple monoid scan of affine transforms. Progress toward #119
2021-11-24 02:28:50 +11:00
struct Transform
{
float4 mat;
float2 translate;
};
Adjust workgroup sizes Make max workgroup size 256 and respect LG_WG_FACTOR. Because the monoid scans only support a height of 2, this will reduce the maximum scene complexity we can render. But it also increases compatibility. Supporting larger scans is a TODO.
2021-12-09 05:42:35 +11:00
static const uint3 gl_WorkGroupSize = uint3(256u, 1u, 1u);
Start work on new element pipeline There's a bit of reorganizing as well. Shader stages are made available from piet-gpu to the test rig, config is now a proper structure (marshaled with bytemuck). This commit just has the transform stage, which is a simple monoid scan of affine transforms. Progress toward #119
2021-11-24 02:28:50 +11:00
static const Transform _23 = { float4(1.0f, 0.0f, 0.0f, 1.0f), 0.0f.xx };
Make shaders cross-platform Translate all piet-gpu shaders into DXIL and MSL; move generated files into the shader/gen directory.
2021-12-04 10:49:58 +11:00
RWByteAddressBuffer _89 : register(u0, space0);
Start work on new element pipeline There's a bit of reorganizing as well. Shader stages are made available from piet-gpu to the test rig, config is now a proper structure (marshaled with bytemuck). This commit just has the transform stage, which is a simple monoid scan of affine transforms. Progress toward #119
2021-11-24 02:28:50 +11:00
static uint3 gl_LocalInvocationID;
static uint3 gl_GlobalInvocationID;
struct SPIRV_Cross_Input
{
uint3 gl_LocalInvocationID : SV_GroupThreadID;
uint3 gl_GlobalInvocationID : SV_DispatchThreadID;
};
Adjust workgroup sizes Make max workgroup size 256 and respect LG_WG_FACTOR. Because the monoid scans only support a height of 2, this will reduce the maximum scene complexity we can render. But it also increases compatibility. Supporting larger scans is a TODO.
2021-12-09 05:42:35 +11:00
groupshared Transform sh_scratch[256];
Start work on new element pipeline There's a bit of reorganizing as well. Shader stages are made available from piet-gpu to the test rig, config is now a proper structure (marshaled with bytemuck). This commit just has the transform stage, which is a simple monoid scan of affine transforms. Progress toward #119
2021-11-24 02:28:50 +11:00
Transform combine_monoid(Transform a, Transform b)
{
Transform c;
c.mat = (a.mat.xyxy * b.mat.xxzz) + (a.mat.zwzw * b.mat.yyww);
c.translate = ((a.mat.xy * b.translate.x) + (a.mat.zw * b.translate.y)) + a.translate;
return c;
}
Transform monoid_identity()
{
return _23;
}
void comp_main()
{
uint ix = gl_GlobalInvocationID.x * 8u;
Transform _93;
_93.mat = asfloat(_89.Load4(ix * 32 + 0));
_93.translate = asfloat(_89.Load2(ix * 32 + 16));
Transform local[8];
local[0].mat = _93.mat;
local[0].translate = _93.translate;
Transform param_1;
for (uint i = 1u; i < 8u; i++)
{
Transform param = local[i - 1u];
Transform _119;
_119.mat = asfloat(_89.Load4((ix + i) * 32 + 0));
_119.translate = asfloat(_89.Load2((ix + i) * 32 + 16));
param_1.mat = _119.mat;
param_1.translate = _119.translate;
local[i] = combine_monoid(param, param_1);
}
Transform agg = local[7];
sh_scratch[gl_LocalInvocationID.x] = agg;
Adjust workgroup sizes Make max workgroup size 256 and respect LG_WG_FACTOR. Because the monoid scans only support a height of 2, this will reduce the maximum scene complexity we can render. But it also increases compatibility. Supporting larger scans is a TODO.
2021-12-09 05:42:35 +11:00
for (uint i_1 = 0u; i_1 < 8u; i_1++)
Start work on new element pipeline There's a bit of reorganizing as well. Shader stages are made available from piet-gpu to the test rig, config is now a proper structure (marshaled with bytemuck). This commit just has the transform stage, which is a simple monoid scan of affine transforms. Progress toward #119
2021-11-24 02:28:50 +11:00
{
GroupMemoryBarrierWithGroupSync();
if (gl_LocalInvocationID.x >= (1u << i_1))
{
Transform other = sh_scratch[gl_LocalInvocationID.x - (1u << i_1)];
Transform param_2 = other;
Transform param_3 = agg;
agg = combine_monoid(param_2, param_3);
}
GroupMemoryBarrierWithGroupSync();
sh_scratch[gl_LocalInvocationID.x] = agg;
}
GroupMemoryBarrierWithGroupSync();
Transform row = monoid_identity();
if (gl_LocalInvocationID.x > 0u)
{
row = sh_scratch[gl_LocalInvocationID.x - 1u];
}
for (uint i_2 = 0u; i_2 < 8u; i_2++)
{
Transform param_4 = row;
Transform param_5 = local[i_2];
Transform m = combine_monoid(param_4, param_5);
Adjust workgroup sizes Make max workgroup size 256 and respect LG_WG_FACTOR. Because the monoid scans only support a height of 2, this will reduce the maximum scene complexity we can render. But it also increases compatibility. Supporting larger scans is a TODO.
2021-12-09 05:42:35 +11:00
uint _208 = ix + i_2;
_89.Store4(_208 * 32 + 0, asuint(m.mat));
_89.Store2(_208 * 32 + 16, asuint(m.translate));
Start work on new element pipeline There's a bit of reorganizing as well. Shader stages are made available from piet-gpu to the test rig, config is now a proper structure (marshaled with bytemuck). This commit just has the transform stage, which is a simple monoid scan of affine transforms. Progress toward #119
2021-11-24 02:28:50 +11:00
}
}
Adjust workgroup sizes Make max workgroup size 256 and respect LG_WG_FACTOR. Because the monoid scans only support a height of 2, this will reduce the maximum scene complexity we can render. But it also increases compatibility. Supporting larger scans is a TODO.
2021-12-09 05:42:35 +11:00
[numthreads(256, 1, 1)]
Start work on new element pipeline There's a bit of reorganizing as well. Shader stages are made available from piet-gpu to the test rig, config is now a proper structure (marshaled with bytemuck). This commit just has the transform stage, which is a simple monoid scan of affine transforms. Progress toward #119
2021-11-24 02:28:50 +11:00
void main(SPIRV_Cross_Input stage_input)
{
gl_LocalInvocationID = stage_input.gl_LocalInvocationID;
gl_GlobalInvocationID = stage_input.gl_GlobalInvocationID;
comp_main();
}
Reference in a new issue Copy permalink