hdkv/yuzu
1
Fork 0
Code Issues Pull Requests Releases 1 Activity

early-access version 1509

Browse Source
This commit is contained in:
pineappleEA
2021-03-08 07:51:31 +01:00
parent cf27b36f44
commit c3f9e4a27b
3170 changed files with 9341219 additions and 163 deletions
Download Patch File Download Diff File Expand all files Collapse all files

48
src/video_core/host_shaders/astc_decoder.comp
View File

@@ -151,6 +151,9 @@ uint color_endpoint_data[16];
int color_bitsread = 0;
uint total_color_bitsread = 0;
int color_index = 0;
// Four values, two endpoints, four maximum paritions
uint color_values[32];
int colvals_index = 0;
// Weight data globals
@@ -597,8 +600,7 @@ void DecodeIntegerSequence(uint max_range, uint num_values) {
}
}
void DecodeColorValues(out uint color_values[32], uvec4 modes, uint num_partitions,
uint color_data_bits) {
void DecodeColorValues(uvec4 modes, uint num_partitions, uint color_data_bits) {
uint num_values = 0;
for (uint i = 0; i < num_partitions; i++) {
num_values += ((modes[i] >> 2) + 1) << 1;
@@ -738,8 +740,7 @@ ivec4 BlueContract(int a, int r, int g, int b) {
return ivec4(a, (r + b) >> 1, (g + b) >> 1, b);
}
void ComputeEndpoints(out uvec4 ep1, out uvec4 ep2, inout uint color_values[32],
uint color_endpoint_mode) {
void ComputeEndpoints(out uvec4 ep1, out uvec4 ep2, uint color_endpoint_mode) {
#define READ_UINT_VALUES(N) \
uint v[N]; \
for (uint i = 0; i < N; i++) { \
@@ -782,18 +783,18 @@ void ComputeEndpoints(out uvec4 ep1, out uvec4 ep2, inout uint color_values[32],
v[3] = transferred.x;
v[2] = transferred.y;
ep1 = ClampByte(ivec4(v[2], v[0], v[0], v[0]));
ep2 = ClampByte(ivec4((v[2] + v[3]), v[0] + v[1], v[0] + v[1], v[0] + v[1]));
ep2 = ClampByte(ivec4(v[2] + v[3], v[0] + v[1], v[0] + v[1], v[0] + v[1]));
break;
}
case 6: {
READ_UINT_VALUES(4)
ep1 = uvec4(0xFF, v[0] * v[3] >> 8, v[1] * v[3] >> 8, v[2] * v[3] >> 8);
ep1 = uvec4(0xFF, (v[0] * v[3]) >> 8, (v[1] * v[3]) >> 8, (v[2] * v[3]) >> 8);
ep2 = uvec4(0xFF, v[0], v[1], v[2]);
break;
}
case 8: {
READ_UINT_VALUES(6)
if (v[1] + v[3] + v[5] >= v[0] + v[2] + v[4]) {
if ((v[1] + v[3] + v[5]) >= (v[0] + v[2] + v[4])) {
ep1 = uvec4(0xFF, v[0], v[2], v[4]);
ep2 = uvec4(0xFF, v[1], v[3], v[5]);
} else {
@@ -813,7 +814,7 @@ void ComputeEndpoints(out uvec4 ep1, out uvec4 ep2, inout uint color_values[32],
transferred = BitTransferSigned(v[5], v[4]);
v[5] = transferred.x;
v[4] = transferred.y;
if (v[1] + v[3] + v[5] >= 0) {
if ((v[1] + v[3] + v[5]) >= 0) {
ep1 = ClampByte(ivec4(0xFF, v[0], v[2], v[4]));
ep2 = ClampByte(ivec4(0xFF, v[0] + v[1], v[2] + v[3], v[4] + v[5]));
} else {
@@ -824,13 +825,13 @@ void ComputeEndpoints(out uvec4 ep1, out uvec4 ep2, inout uint color_values[32],
}
case 10: {
READ_UINT_VALUES(6)
ep1 = uvec4(v[4], v[0] * v[3] >> 8, v[1] * v[3] >> 8, v[2] * v[3] >> 8);
ep1 = uvec4(v[4], (v[0] * v[3]) >> 8, (v[1] * v[3]) >> 8, (v[2] * v[3]) >> 8);
ep2 = uvec4(v[5], v[0], v[1], v[2]);
break;
}
case 12: {
READ_UINT_VALUES(8)
if (v[1] + v[3] + v[5] >= v[0] + v[2] + v[4]) {
if ((v[1] + v[3] + v[5]) >= (v[0] + v[2] + v[4])) {
ep1 = uvec4(v[6], v[0], v[2], v[4]);
ep2 = uvec4(v[7], v[1], v[3], v[5]);
} else {
@@ -856,7 +857,7 @@ void ComputeEndpoints(out uvec4 ep1, out uvec4 ep2, inout uint color_values[32],
v[7] = transferred.x;
v[6] = transferred.y;
if (v[1] + v[3] + v[5] >= 0) {
if ((v[1] + v[3] + v[5]) >= 0) {
ep1 = ClampByte(ivec4(v[6], v[0], v[2], v[4]));
ep2 = ClampByte(ivec4(v[7] + v[6], v[0] + v[1], v[2] + v[3], v[4] + v[5]));
} else {
@@ -865,6 +866,12 @@ void ComputeEndpoints(out uvec4 ep1, out uvec4 ep2, inout uint color_values[32],
}
break;
}
default: {
// HDR mode, or more likely a bug computing the color_endpoint_mode
ep1 = uvec4(0xFF, 0xFF, 0, 0);
ep2 = uvec4(0xFF, 0xFF, 0, 0);
break;
}
}
#undef READ_UINT_VALUES
#undef READ_INT_VALUES
@@ -1177,7 +1184,7 @@ void DecompressBlock(ivec3 coord, uint block_index) {
uint ced_pointer = 0;
uint base_cem = 0;
if (num_partitions == 1) {
color_endpoint_mode[0] = StreamBits(4);
color_endpoint_mode.x = StreamBits(4);
partition_index = 0;
} else {
partition_index = StreamBits(10);
@@ -1227,20 +1234,20 @@ void DecompressBlock(ivec3 coord, uint block_index) {
uint extra_cem = StreamBits(extra_cem_bits);
uint cem = (extra_cem << 6) | base_cem;
cem >>= 2;
bvec4 C = bvec4(false);
uvec4 C = uvec4(0);
for (uint i = 0; i < num_partitions; i++) {
C[i] = (cem & 1) == 0;
C[i] = (cem & 1);
cem >>= 1;
}
uint M[4] = {0, 0, 0, 0};
uvec4 M = uvec4(0);
for (uint i = 0; i < num_partitions; i++) {
M[i] = cem & 3;
cem >>= 2;
}
for (uint i = 0; i < num_partitions; i++) {
color_endpoint_mode[i] = base_mode;
if (C[i]) {
color_endpoint_mode[i] -= 1;
if (C[i] == 0) {
--color_endpoint_mode[i];
}
color_endpoint_mode[i] <<= 2;
color_endpoint_mode[i] |= M[i];
@@ -1251,12 +1258,11 @@ void DecompressBlock(ivec3 coord, uint block_index) {
color_endpoint_mode[i] = cem;
}
}
uint color_values[32]; // Four values, two endpoints, four maximum paritions
DecodeColorValues(color_values, color_endpoint_mode, num_partitions, color_data_bits);
DecodeColorValues(color_endpoint_mode, num_partitions, color_data_bits);
uvec4 endpoints[4][2];
for (uint i = 0; i < num_partitions; i++) {
ComputeEndpoints(endpoints[i][0], endpoints[i][1], color_values, color_endpoint_mode[i]);
ComputeEndpoints(endpoints[i][0], endpoints[i][1], color_endpoint_mode[i]);
}
for (uint i = 0; i < 16; i++) {
@@ -1299,7 +1305,7 @@ void DecompressBlock(ivec3 coord, uint block_index) {
}
weight_vec[c] = unquantized_texel_weights[plane_vec[c]][j * block_dims.x + i];
}
vec4 Cf = vec4((C0 * (uvec4(64) - weight_vec) + C1 * weight_vec + uvec4(32)) >> 6);
vec4 Cf = vec4((C0 * (uvec4(64) - weight_vec) + C1 * weight_vec + uvec4(32)) / 64);
p = (Cf / 65535.0);
imageStore(dest_image, coord + ivec3(i, j, 0), p.gbar);
}