VERY experimental Nintendo Switch 2 emulator
C++
more packing
Ameer J827cb407652f
diff --git a/src/video_core/host_shaders/astc_decoder.comp b/src/video_core/host_shaders/astc_decoder.compindex ca93dc2a20..0e611ede50 100644--- a/src/video_core/host_shaders/astc_decoder.comp+++ b/src/video_core/host_shaders/astc_decoder.comp@@ -88,7 +88,7 @@ int color_bitsread = 0;#define VECTOR_ARRAY_SIZE DIVCEIL(ARRAY_NUM_ELEMENTS * 2, 4)uvec4 result_vector[VECTOR_ARRAY_SIZE];-int result_index = 0;+uint result_index = 0;uint result_vector_max_index;bool result_limit_reached = false;@@ -131,8 +131,8 @@ void ResultEmplaceBack(EncodingData val) {result_limit_reached = true;return;}- const uint array_index = result_index / 4;- const uint vector_index = result_index % 4;+ const uint array_index = result_index / 4u;+ const uint vector_index = result_index % 4u;result_vector[array_index][vector_index] = val.data;++result_index;}@@ -428,69 +428,68 @@ uint BitsOp(uint bits, uint start, uint end) {void DecodeQuintBlock(uint num_bits) {uvec3 m;- uvec3 q;- uint Q;+ uvec4 qQ;m[0] = StreamColorBits(num_bits);- Q = StreamColorBits(3);+ qQ.w = StreamColorBits(3);m[1] = StreamColorBits(num_bits);- Q |= StreamColorBits(2) << 3;+ qQ.w |= StreamColorBits(2) << 3;m[2] = StreamColorBits(num_bits);- Q |= StreamColorBits(2) << 5;- if (BitsOp(Q, 1, 2) == 3 && BitsOp(Q, 5, 6) == 0) {- q.x = 4;- q.y = 4;- q.z = (BitsBracket(Q, 0) << 2) | ((BitsBracket(Q, 4) & ~BitsBracket(Q, 0)) << 1) |- (BitsBracket(Q, 3) & ~BitsBracket(Q, 0));+ qQ.w |= StreamColorBits(2) << 5;+ if (BitsOp(qQ.w, 1, 2) == 3 && BitsOp(qQ.w, 5, 6) == 0) {+ qQ.x = 4;+ qQ.y = 4;+ qQ.z = (BitsBracket(qQ.w, 0) << 2) | ((BitsBracket(qQ.w, 4) & ~BitsBracket(qQ.w, 0)) << 1) |+ (BitsBracket(qQ.w, 3) & ~BitsBracket(qQ.w, 0));} else {uint C = 0;- if (BitsOp(Q, 1, 2) == 3) {- q.z = 4;- C = (BitsOp(Q, 3, 4) << 3) | ((~BitsOp(Q, 5, 6) & 3) << 1) | BitsBracket(Q, 0);+ if (BitsOp(qQ.w, 1, 2) == 3) {+ qQ.z = 4;+ C = (BitsOp(qQ.w, 3, 4) << 3) | ((~BitsOp(qQ.w, 5, 6) & 3) << 1) | BitsBracket(qQ.w, 0);} else {- q.z = BitsOp(Q, 5, 6);- C = BitsOp(Q, 0, 4);+ qQ.z = BitsOp(qQ.w, 5, 6);+ C = BitsOp(qQ.w, 0, 4);}if (BitsOp(C, 0, 2) == 5) {- q.y = 4;- q.x = BitsOp(C, 3, 4);+ qQ.y = 4;+ qQ.x = BitsOp(C, 3, 4);} else {- q.y = BitsOp(C, 3, 4);- q.x = BitsOp(C, 0, 2);+ qQ.y = BitsOp(C, 3, 4);+ qQ.x = BitsOp(C, 0, 2);}}for (uint i = 0; i < 3; i++) {- const EncodingData val = CreateEncodingData(QUINT, num_bits, m[i], q[i]);+ const EncodingData val = CreateEncodingData(QUINT, num_bits, m[i], qQ[i]);ResultEmplaceBack(val);}}void DecodeTritBlock(uint num_bits) {- uint m[5];- uint t[5];- uint T;+ uvec4 m;+ uvec4 t;+ uvec3 Tm5t5;m[0] = StreamColorBits(num_bits);- T = StreamColorBits(2);+ Tm5t5.x = StreamColorBits(2);m[1] = StreamColorBits(num_bits);- T |= StreamColorBits(2) << 2;+ Tm5t5.x |= StreamColorBits(2) << 2;m[2] = StreamColorBits(num_bits);- T |= StreamColorBits(1) << 4;+ Tm5t5.x |= StreamColorBits(1) << 4;m[3] = StreamColorBits(num_bits);- T |= StreamColorBits(2) << 5;- m[4] = StreamColorBits(num_bits);- T |= StreamColorBits(1) << 7;+ Tm5t5.x |= StreamColorBits(2) << 5;+ Tm5t5.y = StreamColorBits(num_bits);+ Tm5t5.x |= StreamColorBits(1) << 7;uint C = 0;- if (BitsOp(T, 2, 4) == 7) {- C = (BitsOp(T, 5, 7) << 2) | BitsOp(T, 0, 1);- t[4] = 2;+ if (BitsOp(Tm5t5.x, 2, 4) == 7) {+ C = (BitsOp(Tm5t5.x, 5, 7) << 2) | BitsOp(Tm5t5.x, 0, 1);+ Tm5t5.z = 2;t[3] = 2;} else {- C = BitsOp(T, 0, 4);- if (BitsOp(T, 5, 6) == 3) {- t[4] = 2;- t[3] = BitsBracket(T, 7);+ C = BitsOp(Tm5t5.x, 0, 4);+ if (BitsOp(Tm5t5.x, 5, 6) == 3) {+ Tm5t5.z = 2;+ t[3] = BitsBracket(Tm5t5.x, 7);} else {- t[4] = BitsBracket(T, 7);- t[3] = BitsOp(T, 5, 6);+ Tm5t5.z = BitsBracket(Tm5t5.x, 7);+ t[3] = BitsOp(Tm5t5.x, 5, 6);}}if (BitsOp(C, 0, 1) == 3) {@@ -506,10 +505,12 @@ void DecodeTritBlock(uint num_bits) {t[1] = BitsOp(C, 2, 3);t[0] = (BitsBracket(C, 1) << 1) | (BitsBracket(C, 0) & ~BitsBracket(C, 1));}- for (uint i = 0; i < 5; i++) {+ for (uint i = 0; i < 4; i++) {const EncodingData val = CreateEncodingData(TRIT, num_bits, m[i], t[i]);ResultEmplaceBack(val);}+ const EncodingData val = CreateEncodingData(TRIT, num_bits, Tm5t5.y, Tm5t5.z);+ ResultEmplaceBack(val);}void DecodeIntegerSequence(uint max_range, uint num_values) {@@ -674,129 +675,128 @@ ivec4 BlueContract(int a, int r, int g, int b) {void ComputeEndpoints(out uvec4 ep1, out uvec4 ep2, uint color_endpoint_mode,inout uint colvals_index) {#define READ_UINT_VALUES(N) \- uint v[N]; \+ uvec4 V[2]; \for (uint i = 0; i < N; i++) { \- v[i] = color_values[colvals_index / 4][colvals_index % 4]; \+ V[i / 4][i % 4] = color_values[colvals_index / 4][colvals_index % 4]; \++colvals_index; \}-#define READ_INT_VALUES(N) \- int v[N]; \+ ivec4 V[2]; \for (uint i = 0; i < N; i++) { \- v[i] = int(color_values[colvals_index / 4][colvals_index % 4]); \+ V[i / 4][i % 4] = int(color_values[colvals_index / 4][colvals_index % 4]); \++colvals_index; \}switch (color_endpoint_mode) {case 0: {READ_UINT_VALUES(2)- ep1 = uvec4(0xFF, v[0], v[0], v[0]);- ep2 = uvec4(0xFF, v[1], v[1], v[1]);+ ep1 = uvec4(0xFF, V[0].x, V[0].x, V[0].x);+ ep2 = uvec4(0xFF, V[0].y, V[0].y, V[0].y);break;}case 1: {READ_UINT_VALUES(2)- const uint L0 = (v[0] >> 2) | (v[1] & 0xC0);- const uint L1 = min(L0 + (v[1] & 0x3F), 0xFFU);+ const uint L0 = (V[0].x >> 2) | (V[0].y & 0xC0);+ const uint L1 = min(L0 + (V[0].y & 0x3F), 0xFFU);ep1 = uvec4(0xFF, L0, L0, L0);ep2 = uvec4(0xFF, L1, L1, L1);break;}case 4: {READ_UINT_VALUES(4)- ep1 = uvec4(v[2], v[0], v[0], v[0]);- ep2 = uvec4(v[3], v[1], v[1], v[1]);+ ep1 = uvec4(V[0].z, V[0].x, V[0].x, V[0].x);+ ep2 = uvec4(V[0].w, V[0].y, V[0].y, V[0].y);break;}case 5: {READ_INT_VALUES(4)- ivec2 transferred = BitTransferSigned(v[1], v[0]);- v[1] = transferred.x;- v[0] = transferred.y;- transferred = BitTransferSigned(v[3], v[2]);- 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]));+ ivec2 transferred = BitTransferSigned(V[0].y, V[0].x);+ V[0].y = transferred.x;+ V[0].x = transferred.y;+ transferred = BitTransferSigned(V[0].w, V[0].z);+ V[0].w = transferred.x;+ V[0].z = transferred.y;+ ep1 = ClampByte(ivec4(V[0].z, V[0].x, V[0].x, V[0].x));+ ep2 = ClampByte(ivec4(V[0].z + V[0].w, V[0].x + V[0].y, V[0].x + V[0].y, V[0].x + V[0].y));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);- ep2 = uvec4(0xFF, v[0], v[1], v[2]);+ ep1 = uvec4(0xFF, (V[0].x * V[0].w) >> 8, (V[0].y * V[0].w) >> 8, (V[0].z * V[0].w) >> 8);+ ep2 = uvec4(0xFF, V[0].x, V[0].y, V[0].z);break;}case 8: {READ_UINT_VALUES(6)- 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]);+ if ((V[0].y + V[0].w + V[1].y) >= (V[0].x + V[0].z + V[1].x)) {+ ep1 = uvec4(0xFF, V[0].x, V[0].z, V[1].x);+ ep2 = uvec4(0xFF, V[0].y, V[0].w, V[1].y);} else {- ep1 = uvec4(BlueContract(0xFF, int(v[1]), int(v[3]), int(v[5])));- ep2 = uvec4(BlueContract(0xFF, int(v[0]), int(v[2]), int(v[4])));+ ep1 = uvec4(BlueContract(0xFF, int(V[0].y), int(V[0].w), int(V[1].y)));+ ep2 = uvec4(BlueContract(0xFF, int(V[0].x), int(V[0].z), int(V[1].x)));}break;}case 9: {READ_INT_VALUES(6)- ivec2 transferred = BitTransferSigned(v[1], v[0]);- v[1] = transferred.x;- v[0] = transferred.y;- transferred = BitTransferSigned(v[3], v[2]);- v[3] = transferred.x;- v[2] = transferred.y;- transferred = BitTransferSigned(v[5], v[4]);- v[5] = transferred.x;- v[4] = transferred.y;- 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]));+ ivec2 transferred = BitTransferSigned(V[0].y, V[0].x);+ V[0].y = transferred.x;+ V[0].x = transferred.y;+ transferred = BitTransferSigned(V[0].w, V[0].z);+ V[0].w = transferred.x;+ V[0].z = transferred.y;+ transferred = BitTransferSigned(V[1].y, V[1].x);+ V[1].y = transferred.x;+ V[1].x = transferred.y;+ if ((V[0].y + V[0].w + V[1].y) >= 0) {+ ep1 = ClampByte(ivec4(0xFF, V[0].x, V[0].z, V[1].x));+ ep2 = ClampByte(ivec4(0xFF, V[0].x + V[0].y, V[0].z + V[0].w, V[1].x + V[1].y));} else {- ep1 = ClampByte(BlueContract(0xFF, v[0] + v[1], v[2] + v[3], v[4] + v[5]));- ep2 = ClampByte(BlueContract(0xFF, v[0], v[2], v[4]));+ ep1 = ClampByte(BlueContract(0xFF, V[0].x + V[0].y, V[0].z + V[0].w, V[1].x + V[1].y));+ ep2 = ClampByte(BlueContract(0xFF, V[0].x, V[0].z, V[1].x));}break;}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);- ep2 = uvec4(v[5], v[0], v[1], v[2]);+ ep1 = uvec4(V[1].x, (V[0].x * V[0].w) >> 8, (V[0].y * V[0].w) >> 8, (V[0].z * V[0].w) >> 8);+ ep2 = uvec4(V[1].y, V[0].x, V[0].y, V[0].z);break;}case 12: {READ_UINT_VALUES(8)- 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]);+ if ((V[0].y + V[0].w + V[1].y) >= (V[0].x + V[0].z + V[1].x)) {+ ep1 = uvec4(V[1].z, V[0].x, V[0].z, V[1].x);+ ep2 = uvec4(V[1].w, V[0].y, V[0].w, V[1].y);} else {- ep1 = uvec4(BlueContract(int(v[7]), int(v[1]), int(v[3]), int(v[5])));- ep2 = uvec4(BlueContract(int(v[6]), int(v[0]), int(v[2]), int(v[4])));+ ep1 = uvec4(BlueContract(int(V[1].w), int(V[0].y), int(V[0].w), int(V[1].y)));+ ep2 = uvec4(BlueContract(int(V[1].z), int(V[0].x), int(V[0].z), int(V[1].x)));}break;}case 13: {READ_INT_VALUES(8)- ivec2 transferred = BitTransferSigned(v[1], v[0]);- v[1] = transferred.x;- v[0] = transferred.y;- transferred = BitTransferSigned(v[3], v[2]);- v[3] = transferred.x;- v[2] = transferred.y;-- transferred = BitTransferSigned(v[5], v[4]);- v[5] = transferred.x;- v[4] = transferred.y;-- transferred = BitTransferSigned(v[7], v[6]);- v[7] = transferred.x;- v[6] = transferred.y;-- 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]));+ ivec2 transferred = BitTransferSigned(V[0].y, V[0].x);+ V[0].y = transferred.x;+ V[0].x = transferred.y;+ transferred = BitTransferSigned(V[0].w, V[0].z);+ V[0].w = transferred.x;+ V[0].z = transferred.y;++ transferred = BitTransferSigned(V[1].y, V[1].x);+ V[1].y = transferred.x;+ V[1].x = transferred.y;++ transferred = BitTransferSigned(V[1].w, V[1].z);+ V[1].w = transferred.x;+ V[1].z = transferred.y;++ if ((V[0].y + V[0].w + V[1].y) >= 0) {+ ep1 = ClampByte(ivec4(V[1].z, V[0].x, V[0].z, V[1].x));+ ep2 = ClampByte(ivec4(V[1].w + V[1].z, V[0].x + V[0].y, V[0].z + V[0].w, V[1].x + V[1].y));} else {- ep1 = ClampByte(BlueContract(v[6] + v[7], v[0] + v[1], v[2] + v[3], v[4] + v[5]));- ep2 = ClampByte(BlueContract(v[6], v[0], v[2], v[4]));+ ep1 = ClampByte(BlueContract(V[1].z + V[1].w, V[0].x + V[0].y, V[0].z + V[0].w, V[1].x + V[1].y));+ ep2 = ClampByte(BlueContract(V[1].z, V[0].x, V[0].z, V[1].x));}break;}