audio_core: interpolate: Improvements to fix audio crackling.

- Fixes audio crackling in Crash Team Racing Nitro-Fueled, Super Mario Odyssey, and others. - Addresses followup issues from #3310.
Merge pull request #3444 from bunnei/linux-audio-fix
2020-02-22 22:26:16 -05:00 · 2020-02-22 03:08:05 -05:00 · 2020-02-22 02:29:41 -05:00 · 2020-02-22 01:54:40 -05:00 · 2020-02-21 22:08:57 -05:00 · 2020-02-21 15:57:27 -03:00
20 changed files with 2047 additions and 838 deletions
@@ -1,4 +1,4 @@
-From https://github.com/yhirose/cpp-httplib/commit/d9479bc0b12e8a1e8bce2d34da4feeef488581f3
+From https://github.com/yhirose/cpp-httplib/tree/fce8e6fefdab4ad48bc5b25c98e5ebfda4f3cf53

 MIT License

@@ -5,21 +5,141 @@
 #define _USE_MATH_DEFINES

 #include <algorithm>
+#include <climits>
 #include <cmath>
 #include <vector>
+
 #include "audio_core/algorithm/interpolate.h"
 #include "common/common_types.h"
 #include "common/logging/log.h"

 namespace AudioCore {

-/// The Lanczos kernel
-static double Lanczos(std::size_t a, double x) {
-    if (x == 0.0)
-        return 1.0;
-    const double px = M_PI * x;
-    return a * std::sin(px) * std::sin(px / a) / (px * px);
-}
+constexpr std::array<s16, 512> curve_lut0{
+    6600,  19426, 6722,  3,     6479,  19424, 6845,  9,     6359,  19419, 6968,  15,    6239,
+    19412, 7093,  22,    6121,  19403, 7219,  28,    6004,  19391, 7345,  34,    5888,  19377,
+    7472,  41,    5773,  19361, 7600,  48,    5659,  19342, 7728,  55,    5546,  19321, 7857,
+    62,    5434,  19298, 7987,  69,    5323,  19273, 8118,  77,    5213,  19245, 8249,  84,
+    5104,  19215, 8381,  92,    4997,  19183, 8513,  101,   4890,  19148, 8646,  109,   4785,
+    19112, 8780,  118,   4681,  19073, 8914,  127,   4579,  19031, 9048,  137,   4477,  18988,
+    9183,  147,   4377,  18942, 9318,  157,   4277,  18895, 9454,  168,   4179,  18845, 9590,
+    179,   4083,  18793, 9726,  190,   3987,  18738, 9863,  202,   3893,  18682, 10000, 215,
+    3800,  18624, 10137, 228,   3709,  18563, 10274, 241,   3618,  18500, 10411, 255,   3529,
+    18436, 10549, 270,   3441,  18369, 10687, 285,   3355,  18300, 10824, 300,   3269,  18230,
+    10962, 317,   3186,  18157, 11100, 334,   3103,  18082, 11238, 351,   3022,  18006, 11375,
+    369,   2942,  17927, 11513, 388,   2863,  17847, 11650, 408,   2785,  17765, 11788, 428,
+    2709,  17681, 11925, 449,   2635,  17595, 12062, 471,   2561,  17507, 12198, 494,   2489,
+    17418, 12334, 517,   2418,  17327, 12470, 541,   2348,  17234, 12606, 566,   2280,  17140,
+    12741, 592,   2213,  17044, 12876, 619,   2147,  16946, 13010, 647,   2083,  16846, 13144,
+    675,   2020,  16745, 13277, 704,   1958,  16643, 13409, 735,   1897,  16539, 13541, 766,
+    1838,  16434, 13673, 798,   1780,  16327, 13803, 832,   1723,  16218, 13933, 866,   1667,
+    16109, 14062, 901,   1613,  15998, 14191, 937,   1560,  15885, 14318, 975,   1508,  15772,
+    14445, 1013,  1457,  15657, 14571, 1052,  1407,  15540, 14695, 1093,  1359,  15423, 14819,
+    1134,  1312,  15304, 14942, 1177,  1266,  15185, 15064, 1221,  1221,  15064, 15185, 1266,
+    1177,  14942, 15304, 1312,  1134,  14819, 15423, 1359,  1093,  14695, 15540, 1407,  1052,
+    14571, 15657, 1457,  1013,  14445, 15772, 1508,  975,   14318, 15885, 1560,  937,   14191,
+    15998, 1613,  901,   14062, 16109, 1667,  866,   13933, 16218, 1723,  832,   13803, 16327,
+    1780,  798,   13673, 16434, 1838,  766,   13541, 16539, 1897,  735,   13409, 16643, 1958,
+    704,   13277, 16745, 2020,  675,   13144, 16846, 2083,  647,   13010, 16946, 2147,  619,
+    12876, 17044, 2213,  592,   12741, 17140, 2280,  566,   12606, 17234, 2348,  541,   12470,
+    17327, 2418,  517,   12334, 17418, 2489,  494,   12198, 17507, 2561,  471,   12062, 17595,
+    2635,  449,   11925, 17681, 2709,  428,   11788, 17765, 2785,  408,   11650, 17847, 2863,
+    388,   11513, 17927, 2942,  369,   11375, 18006, 3022,  351,   11238, 18082, 3103,  334,
+    11100, 18157, 3186,  317,   10962, 18230, 3269,  300,   10824, 18300, 3355,  285,   10687,
+    18369, 3441,  270,   10549, 18436, 3529,  255,   10411, 18500, 3618,  241,   10274, 18563,
+    3709,  228,   10137, 18624, 3800,  215,   10000, 18682, 3893,  202,   9863,  18738, 3987,
+    190,   9726,  18793, 4083,  179,   9590,  18845, 4179,  168,   9454,  18895, 4277,  157,
+    9318,  18942, 4377,  147,   9183,  18988, 4477,  137,   9048,  19031, 4579,  127,   8914,
+    19073, 4681,  118,   8780,  19112, 4785,  109,   8646,  19148, 4890,  101,   8513,  19183,
+    4997,  92,    8381,  19215, 5104,  84,    8249,  19245, 5213,  77,    8118,  19273, 5323,
+    69,    7987,  19298, 5434,  62,    7857,  19321, 5546,  55,    7728,  19342, 5659,  48,
+    7600,  19361, 5773,  41,    7472,  19377, 5888,  34,    7345,  19391, 6004,  28,    7219,
+    19403, 6121,  22,    7093,  19412, 6239,  15,    6968,  19419, 6359,  9,     6845,  19424,
+    6479,  3,     6722,  19426, 6600};
+
+constexpr std::array<s16, 512> curve_lut1{
+    -68,   32639, 69,    -5,    -200,  32630, 212,   -15,   -328,  32613, 359,   -26,   -450,
+    32586, 512,   -36,   -568,  32551, 669,   -47,   -680,  32507, 832,   -58,   -788,  32454,
+    1000,  -69,   -891,  32393, 1174,  -80,   -990,  32323, 1352,  -92,   -1084, 32244, 1536,
+    -103,  -1173, 32157, 1724,  -115,  -1258, 32061, 1919,  -128,  -1338, 31956, 2118,  -140,
+    -1414, 31844, 2322,  -153,  -1486, 31723, 2532,  -167,  -1554, 31593, 2747,  -180,  -1617,
+    31456, 2967,  -194,  -1676, 31310, 3192,  -209,  -1732, 31157, 3422,  -224,  -1783, 30995,
+    3657,  -240,  -1830, 30826, 3897,  -256,  -1874, 30649, 4143,  -272,  -1914, 30464, 4393,
+    -289,  -1951, 30272, 4648,  -307,  -1984, 30072, 4908,  -325,  -2014, 29866, 5172,  -343,
+    -2040, 29652, 5442,  -362,  -2063, 29431, 5716,  -382,  -2083, 29203, 5994,  -403,  -2100,
+    28968, 6277,  -424,  -2114, 28727, 6565,  -445,  -2125, 28480, 6857,  -468,  -2133, 28226,
+    7153,  -490,  -2139, 27966, 7453,  -514,  -2142, 27700, 7758,  -538,  -2142, 27428, 8066,
+    -563,  -2141, 27151, 8378,  -588,  -2136, 26867, 8694,  -614,  -2130, 26579, 9013,  -641,
+    -2121, 26285, 9336,  -668,  -2111, 25987, 9663,  -696,  -2098, 25683, 9993,  -724,  -2084,
+    25375, 10326, -753,  -2067, 25063, 10662, -783,  -2049, 24746, 11000, -813,  -2030, 24425,
+    11342, -844,  -2009, 24100, 11686, -875,  -1986, 23771, 12033, -907,  -1962, 23438, 12382,
+    -939,  -1937, 23103, 12733, -972,  -1911, 22764, 13086, -1005, -1883, 22422, 13441, -1039,
+    -1855, 22077, 13798, -1072, -1825, 21729, 14156, -1107, -1795, 21380, 14516, -1141, -1764,
+    21027, 14877, -1176, -1732, 20673, 15239, -1211, -1700, 20317, 15602, -1246, -1667, 19959,
+    15965, -1282, -1633, 19600, 16329, -1317, -1599, 19239, 16694, -1353, -1564, 18878, 17058,
+    -1388, -1530, 18515, 17423, -1424, -1495, 18151, 17787, -1459, -1459, 17787, 18151, -1495,
+    -1424, 17423, 18515, -1530, -1388, 17058, 18878, -1564, -1353, 16694, 19239, -1599, -1317,
+    16329, 19600, -1633, -1282, 15965, 19959, -1667, -1246, 15602, 20317, -1700, -1211, 15239,
+    20673, -1732, -1176, 14877, 21027, -1764, -1141, 14516, 21380, -1795, -1107, 14156, 21729,
+    -1825, -1072, 13798, 22077, -1855, -1039, 13441, 22422, -1883, -1005, 13086, 22764, -1911,
+    -972,  12733, 23103, -1937, -939,  12382, 23438, -1962, -907,  12033, 23771, -1986, -875,
+    11686, 24100, -2009, -844,  11342, 24425, -2030, -813,  11000, 24746, -2049, -783,  10662,
+    25063, -2067, -753,  10326, 25375, -2084, -724,  9993,  25683, -2098, -696,  9663,  25987,
+    -2111, -668,  9336,  26285, -2121, -641,  9013,  26579, -2130, -614,  8694,  26867, -2136,
+    -588,  8378,  27151, -2141, -563,  8066,  27428, -2142, -538,  7758,  27700, -2142, -514,
+    7453,  27966, -2139, -490,  7153,  28226, -2133, -468,  6857,  28480, -2125, -445,  6565,
+    28727, -2114, -424,  6277,  28968, -2100, -403,  5994,  29203, -2083, -382,  5716,  29431,
+    -2063, -362,  5442,  29652, -2040, -343,  5172,  29866, -2014, -325,  4908,  30072, -1984,
+    -307,  4648,  30272, -1951, -289,  4393,  30464, -1914, -272,  4143,  30649, -1874, -256,
+    3897,  30826, -1830, -240,  3657,  30995, -1783, -224,  3422,  31157, -1732, -209,  3192,
+    31310, -1676, -194,  2967,  31456, -1617, -180,  2747,  31593, -1554, -167,  2532,  31723,
+    -1486, -153,  2322,  31844, -1414, -140,  2118,  31956, -1338, -128,  1919,  32061, -1258,
+    -115,  1724,  32157, -1173, -103,  1536,  32244, -1084, -92,   1352,  32323, -990,  -80,
+    1174,  32393, -891,  -69,   1000,  32454, -788,  -58,   832,   32507, -680,  -47,   669,
+    32551, -568,  -36,   512,   32586, -450,  -26,   359,   32613, -328,  -15,   212,   32630,
+    -200,  -5,    69,    32639, -68};
+
+constexpr std::array<s16, 512> curve_lut2{
+    3195,  26287, 3329,  -32,   3064,  26281, 3467,  -34,   2936,  26270, 3608,  -38,   2811,
+    26253, 3751,  -42,   2688,  26230, 3897,  -46,   2568,  26202, 4046,  -50,   2451,  26169,
+    4199,  -54,   2338,  26130, 4354,  -58,   2227,  26085, 4512,  -63,   2120,  26035, 4673,
+    -67,   2015,  25980, 4837,  -72,   1912,  25919, 5004,  -76,   1813,  25852, 5174,  -81,
+    1716,  25780, 5347,  -87,   1622,  25704, 5522,  -92,   1531,  25621, 5701,  -98,   1442,
+    25533, 5882,  -103,  1357,  25440, 6066,  -109,  1274,  25342, 6253,  -115,  1193,  25239,
+    6442,  -121,  1115,  25131, 6635,  -127,  1040,  25018, 6830,  -133,  967,   24899, 7027,
+    -140,  897,   24776, 7227,  -146,  829,   24648, 7430,  -153,  764,   24516, 7635,  -159,
+    701,   24379, 7842,  -166,  641,   24237, 8052,  -174,  583,   24091, 8264,  -181,  526,
+    23940, 8478,  -187,  472,   23785, 8695,  -194,  420,   23626, 8914,  -202,  371,   23462,
+    9135,  -209,  324,   23295, 9358,  -215,  279,   23123, 9583,  -222,  236,   22948, 9809,
+    -230,  194,   22769, 10038, -237,  154,   22586, 10269, -243,  117,   22399, 10501, -250,
+    81,    22208, 10735, -258,  47,    22015, 10970, -265,  15,    21818, 11206, -271,  -16,
+    21618, 11444, -277,  -44,   21415, 11684, -283,  -71,   21208, 11924, -290,  -97,   20999,
+    12166, -296,  -121,  20786, 12409, -302,  -143,  20571, 12653, -306,  -163,  20354, 12898,
+    -311,  -183,  20134, 13143, -316,  -201,  19911, 13389, -321,  -218,  19686, 13635, -325,
+    -234,  19459, 13882, -328,  -248,  19230, 14130, -332,  -261,  18998, 14377, -335,  -273,
+    18765, 14625, -337,  -284,  18531, 14873, -339,  -294,  18295, 15121, -341,  -302,  18057,
+    15369, -341,  -310,  17817, 15617, -341,  -317,  17577, 15864, -340,  -323,  17335, 16111,
+    -340,  -328,  17092, 16357, -338,  -332,  16848, 16603, -336,  -336,  16603, 16848, -332,
+    -338,  16357, 17092, -328,  -340,  16111, 17335, -323,  -340,  15864, 17577, -317,  -341,
+    15617, 17817, -310,  -341,  15369, 18057, -302,  -341,  15121, 18295, -294,  -339,  14873,
+    18531, -284,  -337,  14625, 18765, -273,  -335,  14377, 18998, -261,  -332,  14130, 19230,
+    -248,  -328,  13882, 19459, -234,  -325,  13635, 19686, -218,  -321,  13389, 19911, -201,
+    -316,  13143, 20134, -183,  -311,  12898, 20354, -163,  -306,  12653, 20571, -143,  -302,
+    12409, 20786, -121,  -296,  12166, 20999, -97,   -290,  11924, 21208, -71,   -283,  11684,
+    21415, -44,   -277,  11444, 21618, -16,   -271,  11206, 21818, 15,    -265,  10970, 22015,
+    47,    -258,  10735, 22208, 81,    -250,  10501, 22399, 117,   -243,  10269, 22586, 154,
+    -237,  10038, 22769, 194,   -230,  9809,  22948, 236,   -222,  9583,  23123, 279,   -215,
+    9358,  23295, 324,   -209,  9135,  23462, 371,   -202,  8914,  23626, 420,   -194,  8695,
+    23785, 472,   -187,  8478,  23940, 526,   -181,  8264,  24091, 583,   -174,  8052,  24237,
+    641,   -166,  7842,  24379, 701,   -159,  7635,  24516, 764,   -153,  7430,  24648, 829,
+    -146,  7227,  24776, 897,   -140,  7027,  24899, 967,   -133,  6830,  25018, 1040,  -127,
+    6635,  25131, 1115,  -121,  6442,  25239, 1193,  -115,  6253,  25342, 1274,  -109,  6066,
+    25440, 1357,  -103,  5882,  25533, 1442,  -98,   5701,  25621, 1531,  -92,   5522,  25704,
+    1622,  -87,   5347,  25780, 1716,  -81,   5174,  25852, 1813,  -76,   5004,  25919, 1912,
+    -72,   4837,  25980, 2015,  -67,   4673,  26035, 2120,  -63,   4512,  26085, 2227,  -58,
+    4354,  26130, 2338,  -54,   4199,  26169, 2451,  -50,   4046,  26202, 2568,  -46,   3897,
+    26230, 2688,  -42,   3751,  26253, 2811,  -38,   3608,  26270, 2936,  -34,   3467,  26281,
+    3064,  -32,   3329,  26287, 3195};

 std::vector<s16> Interpolate(InterpolationState& state, std::vector<s16> input, double ratio) {
    if (input.size() < 2)
@@ -27,43 +147,51 @@ std::vector<s16> Interpolate(InterpolationState& state, std::vector<s16> input,

    if (ratio <= 0) {
        LOG_CRITICAL(Audio, "Nonsensical interpolation ratio {}", ratio);
-        ratio = 1.0;
+        return input;
    }

-    if (ratio != state.current_ratio) {
-        const double cutoff_frequency = std::min(0.5 / ratio, 0.5 * ratio);
-        state.nyquist = CascadingFilter::LowPass(std::clamp(cutoff_frequency, 0.0, 0.4), 3);
-        state.current_ratio = ratio;
-    }
-    state.nyquist.Process(input);
+    const s32 step{static_cast<s32>(ratio * 0x8000)};
+    const std::array<s16, 512>& lut = [step] {
+        if (step > 0xaaaa) {
+            return curve_lut0;
+        }
+        if (step <= 0x8000) {
+            return curve_lut1;
+        }
+        return curve_lut2;
+    }();

-    constexpr std::size_t taps = InterpolationState::lanczos_taps;
-    const std::size_t num_frames = input.size() / 2;
+    const std::size_t num_frames{input.size() / 2};

    std::vector<s16> output;
-    output.reserve(static_cast<std::size_t>(input.size() / ratio + 4));
+    output.reserve(static_cast<std::size_t>(input.size() / ratio + InterpolationState::taps));

-    double& pos = state.position;
-    auto& h = state.history;
-    for (std::size_t i = 0; i < num_frames; ++i) {
-        std::rotate(h.begin(), h.end() - 1, h.end());
-        h[0][0] = input[i * 2 + 0];
-        h[0][1] = input[i * 2 + 1];
+    for (std::size_t frame{}; frame < num_frames; ++frame) {
+        const std::size_t lut_index{(state.fraction >> 8) * InterpolationState::taps};

-        while (pos <= 1.0) {
-            double l = 0.0;
-            double r = 0.0;
-            for (std::size_t j = 0; j < h.size(); j++) {
-                const double lanczos_calc = Lanczos(taps, pos + j - taps + 1);
-                l += lanczos_calc * h[j][0];
-                r += lanczos_calc * h[j][1];
-            }
-            output.emplace_back(static_cast<s16>(std::clamp(l, -32768.0, 32767.0)));
-            output.emplace_back(static_cast<s16>(std::clamp(r, -32768.0, 32767.0)));
+        std::rotate(state.history.begin(), state.history.end() - 1, state.history.end());
+        state.history[0][0] = input[frame * 2 + 0];
+        state.history[0][1] = input[frame * 2 + 1];

-            pos += ratio;
+        while (state.position <= 1.0) {
+            const s32 left{state.history[0][0] * lut[lut_index + 0] +
+                           state.history[1][0] * lut[lut_index + 1] +
+                           state.history[2][0] * lut[lut_index + 2] +
+                           state.history[3][0] * lut[lut_index + 3]};
+            const s32 right{state.history[0][1] * lut[lut_index + 0] +
+                            state.history[1][1] * lut[lut_index + 1] +
+                            state.history[2][1] * lut[lut_index + 2] +
+                            state.history[3][1] * lut[lut_index + 3]};
+            const s32 new_offset{state.fraction + step};
+
+            state.fraction = new_offset & 0x7fff;
+
+            output.emplace_back(static_cast<s16>(std::clamp(left >> 15, SHRT_MIN, SHRT_MAX)));
+            output.emplace_back(static_cast<s16>(std::clamp(right >> 15, SHRT_MIN, SHRT_MAX)));
+
+            state.position += ratio;
        }
-        pos -= 1.0;
+        state.position -= 1.0;
    }

    return output;
@@ -6,19 +6,17 @@

 #include <array>
 #include <vector>
-#include "audio_core/algorithm/filter.h"
+
 #include "common/common_types.h"

 namespace AudioCore {

 struct InterpolationState {
-    static constexpr std::size_t lanczos_taps = 4;
-    static constexpr std::size_t history_size = lanczos_taps * 2 - 1;
-
-    double current_ratio = 0.0;
-    CascadingFilter nyquist;
-    std::array<std::array<s16, 2>, history_size> history = {};
-    double position = 0;
+    static constexpr std::size_t taps{4};
+    static constexpr std::size_t history_size{taps * 2 - 1};
+    std::array<std::array<s16, 2>, history_size> history{};
+    double position{};
+    s32 fraction{};
 };

 /// Interpolates input signal to produce output signal.
@@ -200,7 +200,8 @@ private:
    DownloadResult DownloadInternal(const std::string& resolved_path, u32 timeout_seconds,
                                    const std::string& content_type_name) {
        if (client == nullptr) {
-            client = std::make_unique<httplib::SSLClient>(BOXCAT_HOSTNAME, PORT, timeout_seconds);
+            client = std::make_unique<httplib::SSLClient>(BOXCAT_HOSTNAME, PORT);
+            client->set_timeout_sec(timeout_seconds);
        }

        httplib::Headers headers{
@@ -448,8 +449,8 @@ std::optional<std::vector<u8>> Boxcat::GetLaunchParameter(TitleIDVersion title)

 Boxcat::StatusResult Boxcat::GetStatus(std::optional<std::string>& global,
                                       std::map<std::string, EventStatus>& games) {
-    httplib::SSLClient client{BOXCAT_HOSTNAME, static_cast<int>(PORT),
-                              static_cast<int>(TIMEOUT_SECONDS)};
+    httplib::SSLClient client{BOXCAT_HOSTNAME, static_cast<int>(PORT)};
+    client.set_timeout_sec(static_cast<int>(TIMEOUT_SECONDS));

    httplib::Headers headers{
        {std::string("Game-Assets-API-Version"), std::string(BOXCAT_API_VERSION)},
@@ -489,7 +489,7 @@ void Maxwell3D::FlushMMEInlineDraw() {

    const bool is_indexed = mme_draw.current_mode == MMEDrawMode::Indexed;
    if (ShouldExecute()) {
-        rasterizer.DrawMultiBatch(is_indexed);
+        rasterizer.Draw(is_indexed, true);
    }

    // TODO(bunnei): Below, we reset vertex count so that we can use these registers to determine if
@@ -654,7 +654,7 @@ void Maxwell3D::DrawArrays() {

    const bool is_indexed{regs.index_array.count && !regs.vertex_buffer.count};
    if (ShouldExecute()) {
-        rasterizer.DrawBatch(is_indexed);
+        rasterizer.Draw(is_indexed, false);
    }

    // TODO(bunnei): Below, we reset vertex count so that we can use these registers to determine if
@@ -9,6 +9,7 @@
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/vm_manager.h"
 #include "core/memory.h"
+#include "video_core/gpu.h"
 #include "video_core/memory_manager.h"
 #include "video_core/rasterizer_interface.h"

@@ -84,7 +85,9 @@ GPUVAddr MemoryManager::UnmapBuffer(GPUVAddr gpu_addr, u64 size) {
    const auto cpu_addr = GpuToCpuAddress(gpu_addr);
    ASSERT(cpu_addr);

-    rasterizer.FlushAndInvalidateRegion(cache_addr, aligned_size);
+    // Flush and invalidate through the GPU interface, to be asynchronous if possible.
+    system.GPU().FlushAndInvalidateRegion(cache_addr, aligned_size);
+
    UnmapRange(gpu_addr, aligned_size);
    ASSERT(system.CurrentProcess()
               ->VMManager()
@@ -242,6 +245,8 @@ void MemoryManager::ReadBlock(GPUVAddr src_addr, void* dest_buffer, const std::s
        switch (page_table.attributes[page_index]) {
        case Common::PageType::Memory: {
            const u8* src_ptr{page_table.pointers[page_index] + page_offset};
+            // Flush must happen on the rasterizer interface, such that memory is always synchronous
+            // when it is read (even when in asynchronous GPU mode). Fixes Dead Cells title menu.
            rasterizer.FlushRegion(ToCacheAddr(src_ptr), copy_amount);
            std::memcpy(dest_buffer, src_ptr, copy_amount);
            break;
@@ -292,6 +297,8 @@ void MemoryManager::WriteBlock(GPUVAddr dest_addr, const void* src_buffer, const
        switch (page_table.attributes[page_index]) {
        case Common::PageType::Memory: {
            u8* dest_ptr{page_table.pointers[page_index] + page_offset};
+            // Invalidate must happen on the rasterizer interface, such that memory is always
+            // synchronous when it is written (even when in asynchronous GPU mode).
            rasterizer.InvalidateRegion(ToCacheAddr(dest_ptr), copy_amount);
            std::memcpy(dest_ptr, src_buffer, copy_amount);
            break;
@@ -339,6 +346,8 @@ void MemoryManager::CopyBlock(GPUVAddr dest_addr, GPUVAddr src_addr, const std::

        switch (page_table.attributes[page_index]) {
        case Common::PageType::Memory: {
+            // Flush must happen on the rasterizer interface, such that memory is always synchronous
+            // when it is copied (even when in asynchronous GPU mode).
            const u8* src_ptr{page_table.pointers[page_index] + page_offset};
            rasterizer.FlushRegion(ToCacheAddr(src_ptr), copy_amount);
            WriteBlock(dest_addr, src_ptr, copy_amount);
@@ -35,11 +35,8 @@ class RasterizerInterface {
 public:
    virtual ~RasterizerInterface() {}

-    /// Draw the current batch of vertex arrays
-    virtual bool DrawBatch(bool is_indexed) = 0;
-
-    /// Draw the current batch of multiple instances of vertex arrays
-    virtual bool DrawMultiBatch(bool is_indexed) = 0;
+    /// Dispatches a draw invocation
+    virtual void Draw(bool is_indexed, bool is_instanced) = 0;

    /// Clear the current framebuffer
    virtual void Clear() = 0;
@@ -617,7 +617,7 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {

    // Setup shaders and their used resources.
    texture_cache.GuardSamplers(true);
-    const auto primitive_mode = MaxwellToGL::PrimitiveTopology(gpu.regs.draw.topology);
+    const GLenum primitive_mode = MaxwellToGL::PrimitiveTopology(gpu.regs.draw.topology);
    SetupShaders(primitive_mode);
    texture_cache.GuardSamplers(false);

@@ -650,31 +650,41 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
    const GLsizei num_instances =
        static_cast<GLsizei>(is_instanced ? gpu.mme_draw.instance_count : 1);
    if (is_indexed) {
-        const GLenum index_format = MaxwellToGL::IndexFormat(gpu.regs.index_array.format);
        const GLint base_vertex = static_cast<GLint>(gpu.regs.vb_element_base);
        const GLsizei num_vertices = static_cast<GLsizei>(gpu.regs.index_array.count);
-        glDrawElementsInstancedBaseVertexBaseInstance(
-            primitive_mode, num_vertices, index_format,
-            reinterpret_cast<const void*>(index_buffer_offset), num_instances, base_vertex,
-            base_instance);
+        const GLvoid* offset = reinterpret_cast<const GLvoid*>(index_buffer_offset);
+        const GLenum format = MaxwellToGL::IndexFormat(gpu.regs.index_array.format);
+        if (num_instances == 1 && base_instance == 0 && base_vertex == 0) {
+            glDrawElements(primitive_mode, num_vertices, format, offset);
+        } else if (num_instances == 1 && base_instance == 0) {
+            glDrawElementsBaseVertex(primitive_mode, num_vertices, format, offset, base_vertex);
+        } else if (base_vertex == 0 && base_instance == 0) {
+            glDrawElementsInstanced(primitive_mode, num_vertices, format, offset, num_instances);
+        } else if (base_vertex == 0) {
+            glDrawElementsInstancedBaseInstance(primitive_mode, num_vertices, format, offset,
+                                                num_instances, base_instance);
+        } else if (base_instance == 0) {
+            glDrawElementsInstancedBaseVertex(primitive_mode, num_vertices, format, offset,
+                                              num_instances, base_vertex);
+        } else {
+            glDrawElementsInstancedBaseVertexBaseInstance(primitive_mode, num_vertices, format,
+                                                          offset, num_instances, base_vertex,
+                                                          base_instance);
+        }
    } else {
        const GLint base_vertex = static_cast<GLint>(gpu.regs.vertex_buffer.first);
        const GLsizei num_vertices = static_cast<GLsizei>(gpu.regs.vertex_buffer.count);
-        glDrawArraysInstancedBaseInstance(primitive_mode, base_vertex, num_vertices, num_instances,
-                                          base_instance);
+        if (num_instances == 1 && base_instance == 0) {
+            glDrawArrays(primitive_mode, base_vertex, num_vertices);
+        } else if (base_instance == 0) {
+            glDrawArraysInstanced(primitive_mode, base_vertex, num_vertices, num_instances);
+        } else {
+            glDrawArraysInstancedBaseInstance(primitive_mode, base_vertex, num_vertices,
+                                              num_instances, base_instance);
+        }
    }
 }

-bool RasterizerOpenGL::DrawBatch(bool is_indexed) {
-    Draw(is_indexed, false);
-    return true;
-}
-
-bool RasterizerOpenGL::DrawMultiBatch(bool is_indexed) {
-    Draw(is_indexed, true);
-    return true;
-}
-
 void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) {
    if (device.HasBrokenCompute()) {
        return;
@@ -58,8 +58,7 @@ public:
                              ScreenInfo& info);
    ~RasterizerOpenGL() override;

-    bool DrawBatch(bool is_indexed) override;
-    bool DrawMultiBatch(bool is_indexed) override;
+    void Draw(bool is_indexed, bool is_instanced) override;
    void Clear() override;
    void DispatchCompute(GPUVAddr code_addr) override;
    void ResetCounter(VideoCore::QueryType type) override;
@@ -110,9 +109,6 @@ private:
    void SetupGlobalMemory(u32 binding, const GLShader::GlobalMemoryEntry& entry, GPUVAddr gpu_addr,
                           std::size_t size);

-    /// Syncs all the state, shaders, render targets and textures setting before a draw call.
-    void Draw(bool is_indexed, bool is_instanced);
-
    /// Configures the current textures to use for the draw command.
    void SetupDrawTextures(std::size_t stage_index, const Shader& shader);

@@ -164,7 +164,7 @@ struct FormatTuple {
    {vk::Format::eUndefined, {}},                                // ASTC_2D_5X4
    {vk::Format::eUndefined, {}},                                // BGRA8_SRGB
    {vk::Format::eBc1RgbaSrgbBlock, {}},                         // DXT1_SRGB
-    {vk::Format::eUndefined, {}},                                // DXT23_SRGB
+    {vk::Format::eBc2SrgbBlock, {}},                             // DXT23_SRGB
    {vk::Format::eBc3SrgbBlock, {}},                             // DXT45_SRGB
    {vk::Format::eBc7SrgbBlock, {}},                             // BC7U_SRGB
    {vk::Format::eR4G4B4A4UnormPack16, Attachable},              // R4G4B4A4U
@@ -363,6 +363,8 @@ vk::Format VertexFormat(Maxwell::VertexAttribute::Type type, Maxwell::VertexAttr
            return vk::Format::eR8G8B8A8Uint;
        case Maxwell::VertexAttribute::Size::Size_32:
            return vk::Format::eR32Uint;
+        case Maxwell::VertexAttribute::Size::Size_32_32_32_32:
+            return vk::Format::eR32G32B32A32Uint;
        default:
            break;
        }
@@ -107,6 +107,8 @@ bool VKDevice::Create(const vk::DispatchLoaderDynamic& dldi, vk::Instance instan
    features.occlusionQueryPrecise = true;
    features.fragmentStoresAndAtomics = true;
    features.shaderImageGatherExtended = true;
+    features.shaderStorageImageReadWithoutFormat =
+        is_shader_storage_img_read_without_format_supported;
    features.shaderStorageImageWriteWithoutFormat = true;
    features.textureCompressionASTC_LDR = is_optimal_astc_supported;

@@ -465,6 +467,8 @@ void VKDevice::SetupFamilies(const vk::DispatchLoaderDynamic& dldi, vk::SurfaceK

 void VKDevice::SetupFeatures(const vk::DispatchLoaderDynamic& dldi) {
    const auto supported_features{physical.getFeatures(dldi)};
+    is_shader_storage_img_read_without_format_supported =
+        supported_features.shaderStorageImageReadWithoutFormat;
    is_optimal_astc_supported = IsOptimalAstcSupported(supported_features, dldi);
 }

@@ -538,6 +542,7 @@ std::unordered_map<vk::Format, vk::FormatProperties> VKDevice::GetFormatProperti
                                        vk::Format::eBc6HUfloatBlock,
                                        vk::Format::eBc6HSfloatBlock,
                                        vk::Format::eBc1RgbaSrgbBlock,
+                                        vk::Format::eBc2SrgbBlock,
                                        vk::Format::eBc3SrgbBlock,
                                        vk::Format::eBc7SrgbBlock,
                                        vk::Format::eAstc4x4SrgbBlock,
@@ -122,6 +122,11 @@ public:
        return properties.limits.maxPushConstantsSize;
    }

+    /// Returns true if Shader storage Image Read Without Format supported.
+    bool IsShaderStorageImageReadWithoutFormatSupported() const {
+        return is_shader_storage_img_read_without_format_supported;
+    }
+
    /// Returns true if ASTC is natively supported.
    bool IsOptimalAstcSupported() const {
        return is_optimal_astc_supported;
@@ -227,6 +232,8 @@ private:
    bool ext_depth_range_unrestricted{};       ///< Support for VK_EXT_depth_range_unrestricted.
    bool ext_shader_viewport_index_layer{};    ///< Support for VK_EXT_shader_viewport_index_layer.
    bool nv_device_diagnostic_checkpoints{};   ///< Support for VK_NV_device_diagnostic_checkpoints.
+    bool is_shader_storage_img_read_without_format_supported{}; ///< Support for shader storage
+                                                                ///< image read without format

    // Telemetry parameters
    std::string vendor_name;                      ///< Device's driver name.
@@ -295,16 +295,6 @@ RasterizerVulkan::RasterizerVulkan(Core::System& system, Core::Frontend::EmuWind

 RasterizerVulkan::~RasterizerVulkan() = default;

-bool RasterizerVulkan::DrawBatch(bool is_indexed) {
-    Draw(is_indexed, false);
-    return true;
-}
-
-bool RasterizerVulkan::DrawMultiBatch(bool is_indexed) {
-    Draw(is_indexed, true);
-    return true;
-}
-
 void RasterizerVulkan::Draw(bool is_indexed, bool is_instanced) {
    MICROPROFILE_SCOPE(Vulkan_Drawing);

@@ -105,8 +105,7 @@ public:
                              VKScheduler& scheduler);
    ~RasterizerVulkan() override;

-    bool DrawBatch(bool is_indexed) override;
-    bool DrawMultiBatch(bool is_indexed) override;
+    void Draw(bool is_indexed, bool is_instanced) override;
    void Clear() override;
    void DispatchCompute(GPUVAddr code_addr) override;
    void ResetCounter(VideoCore::QueryType type) override;
@@ -143,8 +142,6 @@ private:

    static constexpr std::size_t ZETA_TEXCEPTION_INDEX = 8;

-    void Draw(bool is_indexed, bool is_instanced);
-
    void FlushWork();

    Texceptions UpdateAttachments();
@@ -23,7 +23,14 @@ static std::optional<vk::BorderColor> TryConvertBorderColor(std::array<float, 4>
    } else if (color == std::array<float, 4>{1, 1, 1, 1}) {
        return vk::BorderColor::eFloatOpaqueWhite;
    } else {
-        return {};
+        if (color[0] + color[1] + color[2] > 1.35f) {
+            // If color elements are brighter than roughly 0.5 average, use white border
+            return vk::BorderColor::eFloatOpaqueWhite;
+        }
+        if (color[3] > 0.5f) {
+            return vk::BorderColor::eFloatOpaqueBlack;
+        }
+        return vk::BorderColor::eFloatTransparentBlack;
    }
 }

@@ -37,8 +44,6 @@ UniqueSampler VKSamplerCache::CreateSampler(const Tegra::Texture::TSCEntry& tsc)

    const auto border_color{tsc.GetBorderColor()};
    const auto vk_border_color{TryConvertBorderColor(border_color)};
-    UNIMPLEMENTED_IF_MSG(!vk_border_color, "Unimplemented border color {} {} {} {}",
-                         border_color[0], border_color[1], border_color[2], border_color[3]);

    constexpr bool unnormalized_coords{false};

@@ -292,6 +292,10 @@ public:
            }
        }

+        if (device.IsShaderStorageImageReadWithoutFormatSupported()) {
+            AddCapability(spv::Capability::StorageImageReadWithoutFormat);
+        }
+
        if (device.IsFloat16Supported()) {
            AddCapability(spv::Capability::Float16);
        }
@@ -1755,8 +1759,16 @@ private:
    }

    Expression ImageLoad(Operation operation) {
-        UNIMPLEMENTED();
-        return {};
+        if (!device.IsShaderStorageImageReadWithoutFormatSupported()) {
+            return {v_float_zero, Type::Float};
+        }
+
+        const auto& meta{std::get<MetaImage>(operation.GetMeta())};
+
+        const Id coords = GetCoordinates(operation, Type::Int);
+        const Id texel = OpImageRead(t_uint4, GetImage(operation), coords);
+
+        return {OpCompositeExtract(t_uint, texel, meta.element), Type::Uint};
    }

    Expression ImageStore(Operation operation) {
@@ -522,68 +522,53 @@ Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
                               Node array, Node depth_compare, u32 bias_offset,
                               std::vector<Node> aoffi,
                               std::optional<Tegra::Shader::Register> bindless_reg) {
-    const auto is_array = static_cast<bool>(array);
-    const auto is_shadow = static_cast<bool>(depth_compare);
+    const bool is_array = array != nullptr;
+    const bool is_shadow = depth_compare != nullptr;
    const bool is_bindless = bindless_reg.has_value();

-    UNIMPLEMENTED_IF_MSG((texture_type == TextureType::Texture3D && (is_array || is_shadow)) ||
-                             (texture_type == TextureType::TextureCube && is_array && is_shadow),
-                         "This method is not supported.");
+    UNIMPLEMENTED_IF(texture_type == TextureType::TextureCube && is_array && is_shadow);
+    ASSERT_MSG(texture_type != TextureType::Texture3D || !is_array || !is_shadow,
+               "Illegal texture type");

    const SamplerInfo info{texture_type, is_array, is_shadow, false};
-    Node index_var{};
+    Node index_var;
    const Sampler* sampler = is_bindless ? GetBindlessSampler(*bindless_reg, index_var, info)
                                         : GetSampler(instr.sampler, info);
-    Node4 values;
-    if (sampler == nullptr) {
-        for (u32 element = 0; element < values.size(); ++element) {
-            values[element] = Immediate(0);
-        }
-        return values;
+    if (!sampler) {
+        return {Immediate(0), Immediate(0), Immediate(0), Immediate(0)};
    }

    const bool lod_needed = process_mode == TextureProcessMode::LZ ||
                            process_mode == TextureProcessMode::LL ||
                            process_mode == TextureProcessMode::LLA;
-
-    // LOD selection (either via bias or explicit textureLod) not supported in GL for
-    // sampler2DArrayShadow and samplerCubeArrayShadow.
-    const bool gl_lod_supported =
-        !((texture_type == Tegra::Shader::TextureType::Texture2D && is_array && is_shadow) ||
-          (texture_type == Tegra::Shader::TextureType::TextureCube && is_array && is_shadow));
-
-    const OperationCode read_method =
-        (lod_needed && gl_lod_supported) ? OperationCode::TextureLod : OperationCode::Texture;
-
-    UNIMPLEMENTED_IF(process_mode != TextureProcessMode::None && !gl_lod_supported);
+    const OperationCode opcode = lod_needed ? OperationCode::TextureLod : OperationCode::Texture;

    Node bias;
    Node lod;
-    if (process_mode != TextureProcessMode::None && gl_lod_supported) {
-        switch (process_mode) {
-        case TextureProcessMode::LZ:
-            lod = Immediate(0.0f);
-            break;
-        case TextureProcessMode::LB:
-            // If present, lod or bias are always stored in the register
-            // indexed by the gpr20 field with an offset depending on the
-            // usage of the other registers
-            bias = GetRegister(instr.gpr20.Value() + bias_offset);
-            break;
-        case TextureProcessMode::LL:
-            lod = GetRegister(instr.gpr20.Value() + bias_offset);
-            break;
-        default:
-            UNIMPLEMENTED_MSG("Unimplemented process mode={}", static_cast<u32>(process_mode));
-            break;
-        }
+    switch (process_mode) {
+    case TextureProcessMode::None:
+        break;
+    case TextureProcessMode::LZ:
+        lod = Immediate(0.0f);
+        break;
+    case TextureProcessMode::LB:
+        // If present, lod or bias are always stored in the register indexed by the gpr20 field with
+        // an offset depending on the usage of the other registers.
+        bias = GetRegister(instr.gpr20.Value() + bias_offset);
+        break;
+    case TextureProcessMode::LL:
+        lod = GetRegister(instr.gpr20.Value() + bias_offset);
+        break;
+    default:
+        UNIMPLEMENTED_MSG("Unimplemented process mode={}", static_cast<u32>(process_mode));
+        break;
    }

+    Node4 values;
    for (u32 element = 0; element < values.size(); ++element) {
-        auto copy_coords = coords;
        MetaTexture meta{*sampler, array, depth_compare, aoffi,    {}, {}, bias,
                         lod,      {},    element,       index_var};
-        values[element] = Operation(read_method, meta, std::move(copy_coords));
+        values[element] = Operation(opcode, meta, coords);
    }

    return values;
@@ -721,7 +721,6 @@ private:
    std::pair<TSurface, TView> GetSurface(const GPUVAddr gpu_addr, const CacheAddr cache_addr,
                                          const SurfaceParams& params, bool preserve_contents,
                                          bool is_render) {
-
        // Step 1
        // Check Level 1 Cache for a fast structural match. If candidate surface
        // matches at certain level we are pretty much done.
@@ -733,14 +732,18 @@ private:
                return RecycleSurface(overlaps, params, gpu_addr, preserve_contents,
                                      topological_result);
            }
+
            const auto struct_result = current_surface->MatchesStructure(params);
-            if (struct_result != MatchStructureResult::None &&
-                (params.target != SurfaceTarget::Texture3D ||
-                 current_surface->MatchTarget(params.target))) {
-                if (struct_result == MatchStructureResult::FullMatch) {
-                    return ManageStructuralMatch(current_surface, params, is_render);
-                } else {
-                    return RebuildSurface(current_surface, params, is_render);
+            if (struct_result != MatchStructureResult::None) {
+                const auto& old_params = current_surface->GetSurfaceParams();
+                const bool not_3d = params.target != SurfaceTarget::Texture3D &&
+                                    old_params.target != SurfaceTarget::Texture3D;
+                if (not_3d || current_surface->MatchTarget(params.target)) {
+                    if (struct_result == MatchStructureResult::FullMatch) {
+                        return ManageStructuralMatch(current_surface, params, is_render);
+                    } else {
+                        return RebuildSurface(current_surface, params, is_render);
+                    }
                }
            }
        }
@@ -73,14 +73,12 @@ struct Client::Impl {
                if (!parsedUrl.GetPort(&port)) {
                    port = HTTP_PORT;
                }
-                cli = std::make_unique<httplib::Client>(parsedUrl.m_Host.c_str(), port,
-                                                        TIMEOUT_SECONDS);
+                cli = std::make_unique<httplib::Client>(parsedUrl.m_Host.c_str(), port);
            } else if (parsedUrl.m_Scheme == "https") {
                if (!parsedUrl.GetPort(&port)) {
                    port = HTTPS_PORT;
                }
-                cli = std::make_unique<httplib::SSLClient>(parsedUrl.m_Host.c_str(), port,
-                                                           TIMEOUT_SECONDS);
+                cli = std::make_unique<httplib::SSLClient>(parsedUrl.m_Host.c_str(), port);
            } else {
                LOG_ERROR(WebService, "Bad URL scheme {}", parsedUrl.m_Scheme);
                return Common::WebResult{Common::WebResult::Code::InvalidURL, "Bad URL scheme"};
@@ -90,6 +88,7 @@ struct Client::Impl {
            LOG_ERROR(WebService, "Invalid URL {}", host + path);
            return Common::WebResult{Common::WebResult::Code::InvalidURL, "Invalid URL"};
        }
+        cli->set_timeout_sec(TIMEOUT_SECONDS);

        httplib::Headers params;
        if (!jwt.empty()) {
Author	SHA1	Message	Date
bunnei	1989e1b9ac	audio_core: interpolate: Improvements to fix audio crackling. - Fixes audio crackling in Crash Team Racing Nitro-Fueled, Super Mario Odyssey, and others. - Addresses followup issues from #3310.	2020-02-22 22:26:16 -05:00
bunnei	d4da52bbd9	Merge pull request #3444 from bunnei/linux-audio-fix audio_core: interpolate: Fix include for climits (Linux build break).	2020-02-22 03:08:05 -05:00
bunnei	f5cf67140b	audio_core: interpolate: Fix include for climits (Linux build break).	2020-02-22 02:29:41 -05:00
bunnei	19bce3685a	Merge pull request #3310 from FearlessTobi/fast-resample audio_core: Switch to a faster interpolation technique	2020-02-22 01:54:40 -05:00
bunnei	27d57e0c4a	Merge pull request #3442 from ReinUsesLisp/fix-3d-assert shader/texture: Fix illegal 3D texture assert	2020-02-21 22:08:57 -05:00
ReinUsesLisp	7dc488a375	shader/texture: Fix illegal 3D texture assert Fix typo in the illegal 3D texture assert logic. We care about catching arrayed 3D textures or 3D shadow textures, not regular 3D textures.	2020-02-21 15:57:27 -03:00
Rodrigo Locatti	4a6a1aeab4	Merge pull request #3433 from namkazt/patch-1 renderer_vulkan: Add the rest of case for TryConvertBorderColor	2020-02-21 15:56:09 -03:00
Rodrigo Locatti	ef27b4b7b5	Merge pull request #3434 from namkazt/patch-2 vk_shader: Implement ImageLoad	2020-02-21 15:55:05 -03:00
Rodrigo Locatti	6b2719c0bb	Merge pull request #3435 from namkazt/patch-3 vulkan: add DXT23_SRGB	2020-02-21 15:48:19 -03:00
bunnei	dc7ebc2d01	Merge pull request #3423 from ReinUsesLisp/no-match-3d texture_cache: Avoid matches in 3D textures	2020-02-21 12:16:51 -05:00
Nguyen Dac Nam	c0c4da27d9	vk_device: remove left over from other branch	2020-02-21 08:56:18 +07:00
bunnei	fe8e5d8ae4	Merge pull request #3438 from bunnei/gpu-mem-manager-fix video_core: memory_manager: Flush/invalidate asynchronously when possible.	2020-02-20 20:04:05 -05:00
Nguyen Dac Nam	ecf275887b	clang-format	2020-02-20 09:39:30 +07:00
Nguyen Dac Nam	fbbad95845	shader_decompiler: only add StorageImageReadWithoutFormat when available	2020-02-20 09:28:13 +07:00
bunnei	2342c0d50e	Merge pull request #3432 from brianclinkenbeard/update-httplib Update httplib to 0.5.5	2020-02-19 21:15:06 -05:00
bunnei	bf0c929d4c	Merge pull request #3415 from ReinUsesLisp/texture-code shader/texture: Allow 2D shadow arrays and simplify code	2020-02-19 20:06:14 -05:00
bunnei	d65fa7d65c	video_core: memory_manager: Flush/invalidate asynchronously on Unmap. - Minor perf improvement.	2020-02-19 20:03:52 -05:00
Brian Clinkenbeard	d31156931d	fix issue with windows getnameinfo()	2020-02-19 16:16:49 -08:00
bunnei	b2bc7682b4	Merge pull request #3414 from ReinUsesLisp/maxwell-3d-draw maxwell_3d: Unify draw methods	2020-02-19 16:13:50 -05:00
bunnei	c8261a1a57	Merge pull request #3411 from ReinUsesLisp/specific-funcs gl_rasterizer: Use the least generic OpenGL draw function possible	2020-02-19 15:37:41 -05:00
bunnei	fd4c5463e8	Merge pull request #3437 from namkazt/patch-5 shader_conversion: add conversion I2F for Short	2020-02-19 11:27:28 -05:00
Nguyen Dac Nam	88cb05e6e7	shader_decompiler: add check in case of device not support ShaderStorageImageReadWithoutFormat	2020-02-19 12:57:22 +07:00
Nguyen Dac Nam	e61c7e9310	vk_device: setup shaderStorageImageReadWithoutFormat	2020-02-19 12:56:36 +07:00
Nguyen Dac Nam	47106ab152	vk_device: add check for shaderStorageImageReadWithoutFormat	2020-02-19 12:55:56 +07:00
Nguyen Dac Nam	2ef8af93aa	vk_shader: add Capability StorageImageReadWithoutFormat	2020-02-19 10:16:51 +07:00
Brian Clinkenbeard	ad4e5c15fb	httplib compatibility	2020-02-18 18:04:33 -08:00
Nguyen Dac Nam	f6f0762e81	vk_shader: Implement function ImageLoad (Used by Kirby Start Allies) Please enter the commit message for your changes. Lines starting	2020-02-19 08:39:01 +07:00
Brian Clinkenbeard	7f6c686d55	update httplib to latest commit	2020-02-18 17:11:40 -08:00
Nguyen Dac Nam	ec206f7f95	fixups mistake auto commit.	2020-02-19 01:24:32 +07:00
Nguyen Dac Nam	eaf60ca5d8	Update code structure Co-Authored-By: Mat M. <mathew1800@gmail.com>	2020-02-19 01:23:08 +07:00
Nguyen Dac Nam	9295966d26	add vertex UnsignedInt size RGBA	2020-02-18 21:52:51 +07:00
Nguyen Dac Nam	9fc42fffd9	add eBc2SrgbBlock to formats	2020-02-18 21:44:09 +07:00
Nguyen Dac Nam	493f0ad904	vulkan: add DXT23_SRGB	2020-02-18 21:39:50 +07:00
Nguyen Dac Nam	ba84f0988f	renderer_vulkan: Add the rest of case for TryConvertBorderColor	2020-02-18 16:52:54 +07:00
Brian Clinkenbeard	9e42025e5b	update httplib README	2020-02-17 22:54:09 -08:00
Brian Clinkenbeard	76b55c3624	0.4.2 works too	2020-02-17 22:53:25 -08:00
Brian Clinkenbeard	293d4d553a	update httplib to 0.2.6	2020-02-17 20:13:24 -08:00
ReinUsesLisp	bfda5ff3f6	texture_cache: Avoid matches in 3D textures Code before this commit was trying to match 3D textures with another target. Fix that.	2020-02-16 04:15:42 -03:00
ReinUsesLisp	6910ade146	shader/texture: Allow 2D shadow arrays and simplify code Shadow sampler 2D arrays are supported on OpenGL, so there's no reason to forbid these. Enable textureLod usage on these. Minor style changes.	2020-02-15 02:36:28 -03:00
ReinUsesLisp	91aa58e410	maxwell_3d: Unify draw methods Pass instanced state of a draw invocation as an argument instead of having two separate virtual methods.	2020-02-14 18:09:40 -03:00
ReinUsesLisp	336a4f8e99	gl_rasterizer: Use the least generic OpenGL draw function possible This may help some implementations.	2020-02-13 21:55:21 -03:00
FearlessTobi	e3cad7d49e	audio_core: Switch to a faster interpolation technique	2020-01-24 00:38:22 +01:00