You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It look like if several constant buffer values get packed in a single "constant buffer register" the hlslcc addresses only the first value in the register
for example in the following shader the buffer is all packed in a single register ("cb[0]")
EDIT: the problem only occurs when 2 or more "variables" are accessed at once (by the D3D HLSL optimizer) using a swizzle like "cb[0].xy" (when the optimizer doesn't pack them in a swizzle it works fine)
INPUT
cbuffer RenderTargetParams {
int Width;
int Height;
float AspectRatio;
}
It look like if several constant buffer values get packed in a single "constant buffer register" the hlslcc addresses only the first value in the register
for example in the following shader the buffer is all packed in a single register ("cb[0]")
EDIT: the problem only occurs when 2 or more "variables" are accessed at once (by the D3D HLSL optimizer) using a swizzle like "cb[0].xy" (when the optimizer doesn't pack them in a swizzle it works fine)
INPUT
cbuffer RenderTargetParams {
int Width;
int Height;
float AspectRatio;
}
struct VS_model0_ZNORMAL_Input {
float4 ObjectPosition : POSITION;
float3 Normal : NORMAL;
};
struct VS_model0_GATHER_Output {
float4 PixelPosition : SV_Position;
};
VS_model0_GATHER_Output main(VS_model0_ZNORMAL_Input input)
{
VS_model0_GATHER_Output output = (VS_model0_GATHER_Output)0;
float2 _screenuvnormalized;
}
ASM
// Buffer Definitions:
//
// cbuffer RenderTargetParams
// {
//
// int Width; // Offset: 0 Size: 4
// int Height; // Offset: 4 Size: 4
// float AspectRatio; // Offset: 8 Size: 4 [unused]
//
// }
//
//
// Resource Bindings:
//
// Name Type Format Dim Slot Elements
// ------------------------------ ---------- ------- ----------- ---- --------
// RenderTargetParams cbuffer NA NA 0 1
//
//
//
// Input signature:
//
// Name Index Mask Register SysValue Format Used
// -------------------- ----- ------ -------- -------- ------- ------
// POSITION 0 xyzw 0 NONE float xy
// NORMAL 0 xyz 1 NONE float
//
//
// Output signature:
//
// Name Index Mask Register SysValue Format Used
// -------------------- ----- ------ -------- -------- ------- ------
// SV_Position 0 xyzw 0 POS float xyzw
//
vs_5_0
dcl_globalFlags refactoringAllowed
dcl_constantbuffer cb0[1], immediateIndexed
dcl_input v0.xy
dcl_output_siv o0.xyzw, position
dcl_temps 1
itof r0.xy, cb0[0].xyxx
div o0.xy, v0.xyxx, r0.xyxx
ishl r0.xy, cb0[0].xyxx, l(1, 1, 0, 0)
itof o0.zw, r0.xxxy
ret
OUTPUT
version 430
extension GL_ARB_shader_bit_encoding : require
struct vec1 {
float x;
};
struct uvec1 {
uint x;
};
struct ivec1 {
int x;
};
out gl_PerVertex {
vec4 gl_Position;
float gl_PointSize;
float gl_ClipDistance[];};
subroutine void SubroutineType();
layout(location = 0) uniform struct RenderTargetParamsVS_Type {
int Width;
int Height;
float AspectRatio;
} RenderTargetParamsVS;
layout(location = 0) in vec4 dcl_Input0;
vec4 Input0;
undef Output0
define Output0 phase0_Output0
vec4 phase0_Output0;
vec4 Temp[1];
ivec4 Temp_int[1];
uvec4 Temp_uint[1];
void main()
{
Input0 = dcl_Input0;
Temp_int[0].xy = floatBitsToInt(vec4(RenderTargetParamsVS.Width).xy);
Output0.xy = (vec4(Input0.xyxx / intBitsToFloat(Temp_int[0]).xyxx)).xy;
Temp_int[0].xy = (RenderTargetParamsVS.Width << ivec4(0x1, 0x1, 0x0, 0x0)).xy;
Output0.zw = (vec4(Temp_int[0].xxxy).zw);
gl_Position = vec4(phase0_Output0);
return;
}
the correct result should be more like
Temp_int[0].xy = (vec4(RenderTargetParamsVS.Width,RenderTargetParamsVS.Height,0,0) << ivec4(0x1, 0x1, 0x0, 0x0)).xy;
and
Temp_int[0].xy = floatBitsToInt(vec4(RenderTargetParamsVS.Width,RenderTargetParamsVS.Height,0,0).xy);
thank you for you time
Alberto
The text was updated successfully, but these errors were encountered: