summaryrefslogtreecommitdiffstats
path: root/src/shader_recompiler/frontend
diff options
context:
space:
mode:
authorReinUsesLisp <reinuseslisp@airmail.cc>2021-04-04 10:17:17 +0200
committerameerj <52414509+ameerj@users.noreply.github.com>2021-07-23 03:51:26 +0200
commitda6cf2632cd4dc0d2b0278353fcaee0789b418c0 (patch)
tree90c2d6f6fa724365a4a23c888389e525e316a4fd /src/shader_recompiler/frontend
parentshader: Implement BAR and fix memory barriers (diff)
downloadyuzu-da6cf2632cd4dc0d2b0278353fcaee0789b418c0.tar
yuzu-da6cf2632cd4dc0d2b0278353fcaee0789b418c0.tar.gz
yuzu-da6cf2632cd4dc0d2b0278353fcaee0789b418c0.tar.bz2
yuzu-da6cf2632cd4dc0d2b0278353fcaee0789b418c0.tar.lz
yuzu-da6cf2632cd4dc0d2b0278353fcaee0789b418c0.tar.xz
yuzu-da6cf2632cd4dc0d2b0278353fcaee0789b418c0.tar.zst
yuzu-da6cf2632cd4dc0d2b0278353fcaee0789b418c0.zip
Diffstat (limited to 'src/shader_recompiler/frontend')
-rw-r--r--src/shader_recompiler/frontend/ir/ir_emitter.cpp20
-rw-r--r--src/shader_recompiler/frontend/ir/ir_emitter.h5
-rw-r--r--src/shader_recompiler/frontend/ir/opcodes.inc5
-rw-r--r--src/shader_recompiler/frontend/maxwell/translate/impl/move_special_register.cpp110
4 files changed, 105 insertions, 35 deletions
diff --git a/src/shader_recompiler/frontend/ir/ir_emitter.cpp b/src/shader_recompiler/frontend/ir/ir_emitter.cpp
index 246c3b9ef..ed1e0dd3b 100644
--- a/src/shader_recompiler/frontend/ir/ir_emitter.cpp
+++ b/src/shader_recompiler/frontend/ir/ir_emitter.cpp
@@ -1628,6 +1628,26 @@ U32 IREmitter::SubgroupBallot(const U1& value) {
return Inst<U32>(Opcode::SubgroupBallot, value);
}
+U32 IREmitter::SubgroupEqMask() {
+ return Inst<U32>(Opcode::SubgroupEqMask);
+}
+
+U32 IREmitter::SubgroupLtMask() {
+ return Inst<U32>(Opcode::SubgroupLtMask);
+}
+
+U32 IREmitter::SubgroupLeMask() {
+ return Inst<U32>(Opcode::SubgroupLeMask);
+}
+
+U32 IREmitter::SubgroupGtMask() {
+ return Inst<U32>(Opcode::SubgroupGtMask);
+}
+
+U32 IREmitter::SubgroupGeMask() {
+ return Inst<U32>(Opcode::SubgroupGeMask);
+}
+
U32 IREmitter::ShuffleIndex(const IR::U32& value, const IR::U32& index, const IR::U32& clamp,
const IR::U32& seg_mask) {
return Inst<U32>(Opcode::ShuffleIndex, value, index, clamp, seg_mask);
diff --git a/src/shader_recompiler/frontend/ir/ir_emitter.h b/src/shader_recompiler/frontend/ir/ir_emitter.h
index 1b00c548d..42756af43 100644
--- a/src/shader_recompiler/frontend/ir/ir_emitter.h
+++ b/src/shader_recompiler/frontend/ir/ir_emitter.h
@@ -281,6 +281,11 @@ public:
[[nodiscard]] U1 VoteAny(const U1& value);
[[nodiscard]] U1 VoteEqual(const U1& value);
[[nodiscard]] U32 SubgroupBallot(const U1& value);
+ [[nodiscard]] U32 SubgroupEqMask();
+ [[nodiscard]] U32 SubgroupLtMask();
+ [[nodiscard]] U32 SubgroupLeMask();
+ [[nodiscard]] U32 SubgroupGtMask();
+ [[nodiscard]] U32 SubgroupGeMask();
[[nodiscard]] U32 ShuffleIndex(const IR::U32& value, const IR::U32& index, const IR::U32& clamp,
const IR::U32& seg_mask);
[[nodiscard]] U32 ShuffleUp(const IR::U32& value, const IR::U32& index, const IR::U32& clamp,
diff --git a/src/shader_recompiler/frontend/ir/opcodes.inc b/src/shader_recompiler/frontend/ir/opcodes.inc
index dcd54bcf7..1697de965 100644
--- a/src/shader_recompiler/frontend/ir/opcodes.inc
+++ b/src/shader_recompiler/frontend/ir/opcodes.inc
@@ -417,6 +417,11 @@ OPCODE(VoteAll, U1, U1,
OPCODE(VoteAny, U1, U1, )
OPCODE(VoteEqual, U1, U1, )
OPCODE(SubgroupBallot, U32, U1, )
+OPCODE(SubgroupEqMask, U32, )
+OPCODE(SubgroupLtMask, U32, )
+OPCODE(SubgroupLeMask, U32, )
+OPCODE(SubgroupGtMask, U32, )
+OPCODE(SubgroupGeMask, U32, )
OPCODE(ShuffleIndex, U32, U32, U32, U32, U32, )
OPCODE(ShuffleUp, U32, U32, U32, U32, U32, )
OPCODE(ShuffleDown, U32, U32, U32, U32, U32, )
diff --git a/src/shader_recompiler/frontend/maxwell/translate/impl/move_special_register.cpp b/src/shader_recompiler/frontend/maxwell/translate/impl/move_special_register.cpp
index 7d9c42a83..be1f21e7b 100644
--- a/src/shader_recompiler/frontend/maxwell/translate/impl/move_special_register.cpp
+++ b/src/shader_recompiler/frontend/maxwell/translate/impl/move_special_register.cpp
@@ -10,6 +10,7 @@ namespace Shader::Maxwell {
namespace {
enum class SpecialRegister : u64 {
SR_LANEID = 0,
+ SR_CLOCK = 1,
SR_VIRTCFG = 2,
SR_VIRTID = 3,
SR_PM0 = 4,
@@ -20,6 +21,9 @@ enum class SpecialRegister : u64 {
SR_PM5 = 9,
SR_PM6 = 10,
SR_PM7 = 11,
+ SR12 = 12,
+ SR13 = 13,
+ SR14 = 14,
SR_ORDERING_TICKET = 15,
SR_PRIM_TYPE = 16,
SR_INVOCATION_ID = 17,
@@ -41,44 +45,70 @@ enum class SpecialRegister : u64 {
SR_TID_X = 33,
SR_TID_Y = 34,
SR_TID_Z = 35,
+ SR_CTA_PARAM = 36,
SR_CTAID_X = 37,
SR_CTAID_Y = 38,
SR_CTAID_Z = 39,
- SR_NTID = 49,
- SR_CirQueueIncrMinusOne = 50,
- SR_NLATC = 51,
- SR_SWINLO = 57,
- SR_SWINSZ = 58,
- SR_SMEMSZ = 59,
- SR_SMEMBANKS = 60,
- SR_LWINLO = 61,
- SR_LWINSZ = 62,
- SR_LMEMLOSZ = 63,
- SR_LMEMHIOFF = 64,
- SR_EQMASK = 65,
- SR_LTMASK = 66,
- SR_LEMASK = 67,
- SR_GTMASK = 68,
- SR_GEMASK = 69,
- SR_REGALLOC = 70,
- SR_GLOBALERRORSTATUS = 73,
- SR_WARPERRORSTATUS = 75,
- SR_PM_HI0 = 81,
- SR_PM_HI1 = 82,
- SR_PM_HI2 = 83,
- SR_PM_HI3 = 84,
- SR_PM_HI4 = 85,
- SR_PM_HI5 = 86,
- SR_PM_HI6 = 87,
- SR_PM_HI7 = 88,
- SR_CLOCKLO = 89,
- SR_CLOCKHI = 90,
- SR_GLOBALTIMERLO = 91,
- SR_GLOBALTIMERHI = 92,
- SR_HWTASKID = 105,
- SR_CIRCULARQUEUEENTRYINDEX = 106,
- SR_CIRCULARQUEUEENTRYADDRESSLOW = 107,
- SR_CIRCULARQUEUEENTRYADDRESSHIGH = 108,
+ SR_NTID = 40,
+ SR_CirQueueIncrMinusOne = 41,
+ SR_NLATC = 42,
+ SR43 = 43,
+ SR_SM_SPA_VERSION = 44,
+ SR_MULTIPASSSHADERINFO = 45,
+ SR_LWINHI = 46,
+ SR_SWINHI = 47,
+ SR_SWINLO = 48,
+ SR_SWINSZ = 49,
+ SR_SMEMSZ = 50,
+ SR_SMEMBANKS = 51,
+ SR_LWINLO = 52,
+ SR_LWINSZ = 53,
+ SR_LMEMLOSZ = 54,
+ SR_LMEMHIOFF = 55,
+ SR_EQMASK = 56,
+ SR_LTMASK = 57,
+ SR_LEMASK = 58,
+ SR_GTMASK = 59,
+ SR_GEMASK = 60,
+ SR_REGALLOC = 61,
+ SR_BARRIERALLOC = 62,
+ SR63 = 63,
+ SR_GLOBALERRORSTATUS = 64,
+ SR65 = 65,
+ SR_WARPERRORSTATUS = 66,
+ SR_WARPERRORSTATUSCLEAR = 67,
+ SR68 = 68,
+ SR69 = 69,
+ SR70 = 70,
+ SR71 = 71,
+ SR_PM_HI0 = 72,
+ SR_PM_HI1 = 73,
+ SR_PM_HI2 = 74,
+ SR_PM_HI3 = 75,
+ SR_PM_HI4 = 76,
+ SR_PM_HI5 = 77,
+ SR_PM_HI6 = 78,
+ SR_PM_HI7 = 79,
+ SR_CLOCKLO = 80,
+ SR_CLOCKHI = 81,
+ SR_GLOBALTIMERLO = 82,
+ SR_GLOBALTIMERHI = 83,
+ SR84 = 84,
+ SR85 = 85,
+ SR86 = 86,
+ SR87 = 87,
+ SR88 = 88,
+ SR89 = 89,
+ SR90 = 90,
+ SR91 = 91,
+ SR92 = 92,
+ SR93 = 93,
+ SR94 = 94,
+ SR95 = 95,
+ SR_HWTASKID = 96,
+ SR_CIRCULARQUEUEENTRYINDEX = 97,
+ SR_CIRCULARQUEUEENTRYADDRESSLOW = 98,
+ SR_CIRCULARQUEUEENTRYADDRESSHIGH = 99,
};
[[nodiscard]] IR::U32 Read(IR::IREmitter& ir, SpecialRegister special_register) {
@@ -103,6 +133,16 @@ enum class SpecialRegister : u64 {
return ir.Imm32(Common::BitCast<u32>(1.0f));
case SpecialRegister::SR_LANEID:
return ir.LaneId();
+ case SpecialRegister::SR_EQMASK:
+ return ir.SubgroupEqMask();
+ case SpecialRegister::SR_LTMASK:
+ return ir.SubgroupLtMask();
+ case SpecialRegister::SR_LEMASK:
+ return ir.SubgroupLeMask();
+ case SpecialRegister::SR_GTMASK:
+ return ir.SubgroupGtMask();
+ case SpecialRegister::SR_GEMASK:
+ return ir.SubgroupGeMask();
default:
throw NotImplementedException("S2R special register {}", special_register);
}