diff options
author | Ameer J <52414509+ameerj@users.noreply.github.com> | 2023-08-02 02:52:14 +0200 |
---|---|---|
committer | Ameer J <52414509+ameerj@users.noreply.github.com> | 2023-08-06 20:54:57 +0200 |
commit | f41fb3ec0b12761c723f26e5f44d1c7f15468153 (patch) | |
tree | f4b97703976421cdbc1069747a2a6a647aff5e29 /src/video_core/host_shaders | |
parent | Revert "global endpoints" (diff) | |
download | yuzu-f41fb3ec0b12761c723f26e5f44d1c7f15468153.tar yuzu-f41fb3ec0b12761c723f26e5f44d1c7f15468153.tar.gz yuzu-f41fb3ec0b12761c723f26e5f44d1c7f15468153.tar.bz2 yuzu-f41fb3ec0b12761c723f26e5f44d1c7f15468153.tar.lz yuzu-f41fb3ec0b12761c723f26e5f44d1c7f15468153.tar.xz yuzu-f41fb3ec0b12761c723f26e5f44d1c7f15468153.tar.zst yuzu-f41fb3ec0b12761c723f26e5f44d1c7f15468153.zip |
Diffstat (limited to '')
-rw-r--r-- | src/video_core/host_shaders/astc_decoder.comp | 28 |
1 files changed, 13 insertions, 15 deletions
diff --git a/src/video_core/host_shaders/astc_decoder.comp b/src/video_core/host_shaders/astc_decoder.comp index 5346cba0c..fd38dcfe5 100644 --- a/src/video_core/host_shaders/astc_decoder.comp +++ b/src/video_core/host_shaders/astc_decoder.comp @@ -21,8 +21,6 @@ #endif -#define bfe bitfieldExtract - layout(local_size_x = 8, local_size_y = 8, local_size_z = 1) in; BEGIN_PUSH_CONSTANTS @@ -134,7 +132,7 @@ void ResultEmplaceBack(EncodingData val) { return; } const uint array_index = result_index / 4; - const uint vector_index = bfe(result_index, 0, 2); + const uint vector_index = result_index % 4; result_vector[array_index][vector_index] = val.data; ++result_index; } @@ -386,7 +384,7 @@ uint StreamColorBits(uint num_bits) { EncodingData GetEncodingFromVector(uint index) { const uint array_index = index / 4; - const uint vector_index = bfe(index, 0, 2); + const uint vector_index = index % 4; const uint data = result_vector[array_index][vector_index]; return EncodingData(data); @@ -395,7 +393,7 @@ EncodingData GetEncodingFromVector(uint index) { // Returns the number of bits required to encode n_vals values. uint GetBitLength(uint n_vals, uint encoding_index) { const EncodingData encoding_value = - EncodingData(encoding_values[encoding_index / 4][bfe(encoding_index, 0, 2)]); + EncodingData(encoding_values[encoding_index / 4][encoding_index % 4]); const uint encoding = Encoding(encoding_value); uint total_bits = NumBits(encoding_value) * n_vals; if (encoding == TRIT) { @@ -515,7 +513,7 @@ void DecodeTritBlock(uint num_bits) { } void DecodeIntegerSequence(uint max_range, uint num_values) { - EncodingData val = EncodingData(encoding_values[max_range / 4][bfe(max_range, 0, 2)]); + EncodingData val = EncodingData(encoding_values[max_range / 4][max_range % 4]); const uint encoding = Encoding(val); const uint num_bits = NumBits(val); uint vals_decoded = 0; @@ -567,7 +565,7 @@ void DecodeColorValues(uvec4 modes, uint num_partitions, uint color_data_bits) { A = ReplicateBitTo9((bitval & 1)); switch (encoding) { case JUST_BITS: - color_values[out_index / 4][bfe(out_index, 0, 2)] = FastReplicateTo8(bitval, bitlen); + color_values[out_index / 4][out_index % 4] = FastReplicateTo8(bitval, bitlen); ++out_index; break; case TRIT: { @@ -647,7 +645,7 @@ void DecodeColorValues(uvec4 modes, uint num_partitions, uint color_data_bits) { uint T = (D * C) + B; T ^= A; T = (A & 0x80) | (T >> 2); - color_values[out_index / 4][bfe(out_index, 0, 2)] = T; + color_values[out_index / 4][out_index % 4] = T; ++out_index; } } @@ -678,14 +676,14 @@ 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++) { \ - v[i] = color_values[colvals_index / 4][bfe(colvals_index, 0, 2)]; \ + v[i] = color_values[colvals_index / 4][colvals_index % 4]; \ ++colvals_index; \ } #define READ_INT_VALUES(N) \ int v[N]; \ for (uint i = 0; i < N; i++) { \ - v[i] = int(color_values[colvals_index / 4][bfe(colvals_index, 0, 2)]); \ + v[i] = int(color_values[colvals_index / 4][colvals_index % 4]); \ ++colvals_index; \ } @@ -896,7 +894,7 @@ void UnquantizeTexelWeights(uvec2 size, bool is_dual_plane) { const uint loop_count = min(result_index, area * num_planes); for (uint itr = 0; itr < loop_count; ++itr) { const uint array_index = itr / 4; - const uint vector_index = bfe(itr, 0, 2); + const uint vector_index = itr % 4; result_vector[array_index][vector_index] = UnquantizeTexelWeight(GetEncodingFromVector(itr)); } @@ -923,7 +921,7 @@ void UnquantizeTexelWeights(uvec2 size, bool is_dual_plane) { #define VectorIndicesFromBase(offset_base) \ const uint offset = is_dual_plane ? 2 * offset_base + plane : offset_base; \ const uint array_index = offset / 4; \ - const uint vector_index = bfe(offset, 0, 2); + const uint vector_index = offset % 4; if (v0 < area) { const uint offset_base = v0; @@ -947,7 +945,7 @@ void UnquantizeTexelWeights(uvec2 size, bool is_dual_plane) { } const uint offset = (t * block_dims.x + s) + ARRAY_NUM_ELEMENTS * plane; const uint array_index = offset / 4; - const uint vector_index = bfe(offset, 0, 2); + const uint vector_index = offset % 4; unquantized_texel_weights[array_index][vector_index] = (uint(dot(p, w)) + 8) >> 4; } } @@ -1251,13 +1249,13 @@ void DecompressBlock(ivec3 coord) { const uvec4 C1 = ReplicateByteTo16(endpoints1[local_partition]); const uint weight_offset = (j * block_dims.x + i); const uint array_index = weight_offset / 4; - const uint vector_index = bfe(weight_offset, 0, 2); + const uint vector_index = weight_offset % 4; const uint primary_weight = unquantized_texel_weights[array_index][vector_index]; uvec4 weight_vec = uvec4(primary_weight); if (params.dual_plane) { const uint secondary_weight_offset = (j * block_dims.x + i) + ARRAY_NUM_ELEMENTS; const uint secondary_array_index = secondary_weight_offset / 4; - const uint secondary_vector_index = bfe(secondary_weight_offset, 0, 2); + const uint secondary_vector_index = secondary_weight_offset % 4; const uint secondary_weight = unquantized_texel_weights[secondary_array_index][secondary_vector_index]; for (uint c = 0; c < 4; c++) { |