diff --git a/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp b/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
index 306d59d0867cd..70db7b4918515 100644
--- a/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
+++ b/llvm/lib/Target/AMDGPU/SIInsertWaitcnts.cpp
@@ -531,6 +531,7 @@ class SIInsertWaitcnts {
   // instruction.
   WaitEventType getVmemWaitEventType(const MachineInstr &Inst) const {
     switch (Inst.getOpcode()) {
+    // FIXME: GLOBAL_INV needs to be tracked with xcnt too.
     case AMDGPU::GLOBAL_INV:
       return VMEM_READ_ACCESS; // tracked using loadcnt
     case AMDGPU::GLOBAL_WB:
@@ -633,8 +634,11 @@ class WaitcntBrackets {
                              const MachineOperand &Op) const;
 
   bool counterOutOfOrder(InstCounterType T) const;
-  void simplifyWaitcnt(AMDGPU::Waitcnt &Wait) const;
+  void simplifyWaitcnt(AMDGPU::Waitcnt &Wait);
   void simplifyWaitcnt(InstCounterType T, unsigned &Count) const;
+  bool hasRedundantXCntWithKmCnt(const AMDGPU::Waitcnt &Wait);
+  bool canOptimizeXCntWithLoadCnt(const AMDGPU::Waitcnt &Wait);
+  void simplifyXcnt(AMDGPU::Waitcnt &CheckWait, AMDGPU::Waitcnt &UpdateWait);
 
   void determineWait(InstCounterType T, RegInterval Interval,
                      AMDGPU::Waitcnt &Wait) const;
@@ -646,7 +650,6 @@ class WaitcntBrackets {
 
   void applyWaitcnt(const AMDGPU::Waitcnt &Wait);
   void applyWaitcnt(InstCounterType T, unsigned Count);
-  void applyXcnt(const AMDGPU::Waitcnt &Wait);
   void updateByEvent(WaitEventType E, MachineInstr &MI);
 
   unsigned hasPendingEvent() const { return PendingEvents; }
@@ -1192,7 +1195,7 @@ void WaitcntBrackets::print(raw_ostream &OS) const {
 
 /// Simplify the waitcnt, in the sense of removing redundant counts, and return
 /// whether a waitcnt instruction is needed at all.
-void WaitcntBrackets::simplifyWaitcnt(AMDGPU::Waitcnt &Wait) const {
+void WaitcntBrackets::simplifyWaitcnt(AMDGPU::Waitcnt &Wait) {
   simplifyWaitcnt(LOAD_CNT, Wait.LoadCnt);
   simplifyWaitcnt(EXP_CNT, Wait.ExpCnt);
   simplifyWaitcnt(DS_CNT, Wait.DsCnt);
@@ -1200,7 +1203,7 @@ void WaitcntBrackets::simplifyWaitcnt(AMDGPU::Waitcnt &Wait) const {
   simplifyWaitcnt(SAMPLE_CNT, Wait.SampleCnt);
   simplifyWaitcnt(BVH_CNT, Wait.BvhCnt);
   simplifyWaitcnt(KM_CNT, Wait.KmCnt);
-  simplifyWaitcnt(X_CNT, Wait.XCnt);
+  simplifyXcnt(Wait, Wait);
 }
 
 void WaitcntBrackets::simplifyWaitcnt(InstCounterType T,
@@ -1270,7 +1273,7 @@ void WaitcntBrackets::applyWaitcnt(const AMDGPU::Waitcnt &Wait) {
   applyWaitcnt(SAMPLE_CNT, Wait.SampleCnt);
   applyWaitcnt(BVH_CNT, Wait.BvhCnt);
   applyWaitcnt(KM_CNT, Wait.KmCnt);
-  applyXcnt(Wait);
+  applyWaitcnt(X_CNT, Wait.XCnt);
 }
 
 void WaitcntBrackets::applyWaitcnt(InstCounterType T, unsigned Count) {
@@ -1287,35 +1290,42 @@ void WaitcntBrackets::applyWaitcnt(InstCounterType T, unsigned Count) {
   }
 }
 
-void WaitcntBrackets::applyXcnt(const AMDGPU::Waitcnt &Wait) {
-  // On entry to a block with multiple predescessors, there may
-  // be pending SMEM and VMEM events active at the same time.
-  // In such cases, only clear one active event at a time.
-
+bool WaitcntBrackets::hasRedundantXCntWithKmCnt(const AMDGPU::Waitcnt &Wait) {
   // Wait on XCNT is redundant if we are already waiting for a load to complete.
   // SMEM can return out of order, so only omit XCNT wait if we are waiting till
   // zero.
-  if (Wait.KmCnt == 0 && hasPendingEvent(SMEM_GROUP)) {
-    if (!hasMixedPendingEvents(X_CNT))
-      applyWaitcnt(X_CNT, 0);
-    else
-      PendingEvents &= ~(1 << SMEM_GROUP);
-    return;
-  }
+  return Wait.KmCnt == 0 && hasPendingEvent(SMEM_GROUP);
+}
 
+bool WaitcntBrackets::canOptimizeXCntWithLoadCnt(const AMDGPU::Waitcnt &Wait) {
   // If we have pending store we cannot optimize XCnt because we do not wait for
   // stores. VMEM loads retun in order, so if we only have loads XCnt is
   // decremented to the same number as LOADCnt.
-  if (Wait.LoadCnt != ~0u && hasPendingEvent(VMEM_GROUP) &&
-      !hasPendingEvent(STORE_CNT)) {
-    if (!hasMixedPendingEvents(X_CNT))
-      applyWaitcnt(X_CNT, std::min(Wait.XCnt, Wait.LoadCnt));
-    else if (Wait.LoadCnt == 0)
+  return Wait.LoadCnt != ~0u && hasPendingEvent(VMEM_GROUP) &&
+         !hasPendingEvent(STORE_CNT);
+}
+
+void WaitcntBrackets::simplifyXcnt(AMDGPU::Waitcnt &CheckWait,
+                                   AMDGPU::Waitcnt &UpdateWait) {
+  // Try to simplify xcnt further by checking for joint kmcnt and loadcnt
+  // optimizations. On entry to a block with multiple predescessors, there may
+  // be pending SMEM and VMEM events active at the same time.
+  // In such cases, only clear one active event at a time.
+  // TODO: Revisit xcnt optimizations for gfx1250.
+  if (hasRedundantXCntWithKmCnt(CheckWait)) {
+    if (!hasMixedPendingEvents(X_CNT)) {
+      applyWaitcnt(X_CNT, 0);
+    } else {
+      PendingEvents &= ~(1 << SMEM_GROUP);
+    }
+  } else if (canOptimizeXCntWithLoadCnt(CheckWait)) {
+    if (!hasMixedPendingEvents(X_CNT)) {
+      applyWaitcnt(X_CNT, std::min(CheckWait.XCnt, CheckWait.LoadCnt));
+    } else if (CheckWait.LoadCnt == 0) {
       PendingEvents &= ~(1 << VMEM_GROUP);
-    return;
+    }
   }
-
-  applyWaitcnt(X_CNT, Wait.XCnt);
+  simplifyWaitcnt(X_CNT, UpdateWait.XCnt);
 }
 
 // Where there are multiple types of event in the bracket of a counter,
@@ -1650,6 +1660,8 @@ bool WaitcntGeneratorGFX12Plus::applyPreexistingWaitcnt(
     }
   }
 
+  // Save the pre combine waitcnt in order to make xcnt checks.
+  AMDGPU::Waitcnt PreCombine = Wait;
   if (CombinedLoadDsCntInstr) {
     // Only keep an S_WAIT_LOADCNT_DSCNT if both counters actually need
     // to be waited for. Otherwise, let the instruction be deleted so
@@ -1740,6 +1752,13 @@ bool WaitcntGeneratorGFX12Plus::applyPreexistingWaitcnt(
   }
 
   for (auto CT : inst_counter_types(NUM_EXTENDED_INST_CNTS)) {
+    if ((CT == KM_CNT && ScoreBrackets.hasRedundantXCntWithKmCnt(PreCombine)) ||
+        (CT == LOAD_CNT &&
+         ScoreBrackets.canOptimizeXCntWithLoadCnt(PreCombine))) {
+      // Xcnt may need to be updated depending on a pre-existing KM/LOAD_CNT
+      // due to taking the backedge of a block.
+      ScoreBrackets.simplifyXcnt(PreCombine, Wait);
+    }
     if (!WaitInstrs[CT])
       continue;
 
@@ -2086,6 +2105,14 @@ bool SIInsertWaitcnts::generateWaitcntInstBefore(MachineInstr &MI,
   // Verify that the wait is actually needed.
   ScoreBrackets.simplifyWaitcnt(Wait);
 
+  // Since the translation for VMEM addresses occur in-order, we can apply the
+  // XCnt if the current instruction is of VMEM type and has a memory
+  // dependency with another VMEM instruction in flight.
+  if (Wait.XCnt != ~0u && isVmemAccess(MI)) {
+    ScoreBrackets.applyWaitcnt(X_CNT, Wait.XCnt);
+    Wait.XCnt = ~0u;
+  }
+
   // When forcing emit, we need to skip terminators because that would break the
   // terminators of the MBB if we emit a waitcnt between terminators.
   if (ForceEmitZeroFlag && !MI.isTerminator())
@@ -2154,21 +2181,6 @@ bool SIInsertWaitcnts::generateWaitcnt(AMDGPU::Waitcnt Wait,
                       << "Update Instr: " << *It);
   }
 
-  // XCnt may be already consumed by a load wait.
-  if (Wait.XCnt != ~0u) {
-    if (Wait.KmCnt == 0 && !ScoreBrackets.hasPendingEvent(SMEM_GROUP))
-      Wait.XCnt = ~0u;
-
-    if (Wait.LoadCnt == 0 && !ScoreBrackets.hasPendingEvent(VMEM_GROUP))
-      Wait.XCnt = ~0u;
-
-    // Since the translation for VMEM addresses occur in-order, we can skip the
-    // XCnt if the current instruction is of VMEM type and has a memory
-    // dependency with another VMEM instruction in flight.
-    if (isVmemAccess(*It))
-      Wait.XCnt = ~0u;
-  }
-
   if (WCG->createNewWaitcnt(Block, It, Wait))
     Modified = true;
 
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/fp64-atomics-gfx90a.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/fp64-atomics-gfx90a.ll
index 7e297f46a780e..b5d593a9c15ed 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/fp64-atomics-gfx90a.ll
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/fp64-atomics-gfx90a.ll
@@ -1501,7 +1501,6 @@ define amdgpu_kernel void @global_atomic_fadd_f64_noret_pat(ptr addrspace(1) %pt
 ; GFX1250-NEXT:    v_mul_f64_e32 v[0:1], 4.0, v[0:1]
 ; GFX1250-NEXT:    global_wb scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
@@ -1574,7 +1573,6 @@ define amdgpu_kernel void @global_atomic_fadd_f64_noret_pat_agent(ptr addrspace(
 ; GFX1250-NEXT:    v_mul_f64_e32 v[0:1], 4.0, v[0:1]
 ; GFX1250-NEXT:    global_wb scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
@@ -1649,7 +1647,6 @@ define amdgpu_kernel void @global_atomic_fadd_f64_noret_pat_system(ptr addrspace
 ; GFX1250-NEXT:    v_mul_f64_e32 v[0:1], 4.0, v[0:1]
 ; GFX1250-NEXT:    global_wb scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
@@ -1722,7 +1719,6 @@ define amdgpu_kernel void @global_atomic_fadd_f64_noret_pat_flush(ptr addrspace(
 ; GFX1250-NEXT:    v_mul_f64_e32 v[0:1], 4.0, v[0:1]
 ; GFX1250-NEXT:    global_wb scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
@@ -1913,7 +1909,6 @@ define amdgpu_kernel void @global_atomic_fadd_f64_noret_pat_agent_safe(ptr addrs
 ; GFX1250-NEXT:    v_mul_f64_e32 v[0:1], 4.0, v[0:1]
 ; GFX1250-NEXT:    global_wb scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
@@ -1959,7 +1954,6 @@ define amdgpu_kernel void @flat_atomic_fadd_f64_noret_pat(ptr %ptr) #1 {
 ; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1250-NEXT:    global_wb scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    flat_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt_dscnt 0x0
@@ -2002,7 +1996,6 @@ define amdgpu_kernel void @flat_atomic_fadd_f64_noret_pat_agent(ptr %ptr) #1 {
 ; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1250-NEXT:    global_wb scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    flat_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt_dscnt 0x0
@@ -2047,7 +2040,6 @@ define amdgpu_kernel void @flat_atomic_fadd_f64_noret_pat_system(ptr %ptr) #1 {
 ; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1250-NEXT:    global_wb scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    flat_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
@@ -2210,7 +2202,6 @@ define amdgpu_kernel void @flat_atomic_fadd_f64_noret_pat_agent_safe(ptr %ptr) {
 ; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1250-NEXT:    global_wb scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    flat_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt_dscnt 0x0
diff --git a/llvm/test/CodeGen/AMDGPU/bf16.ll b/llvm/test/CodeGen/AMDGPU/bf16.ll
index 393e9fecbb308..28d7e6916e519 100644
--- a/llvm/test/CodeGen/AMDGPU/bf16.ll
+++ b/llvm/test/CodeGen/AMDGPU/bf16.ll
@@ -2520,6 +2520,7 @@ define void @test_load_store_f32_to_bf16(ptr addrspace(1) %in, ptr addrspace(1)
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_load_b32 v0, v[0:1], off
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_cvt_pk_bf16_f32 v0, v0, s0
 ; GFX1250-NEXT:    global_store_b16 v[2:3], v0, off
 ; GFX1250-NEXT:    s_set_pc_i64 s[30:31]
@@ -2783,6 +2784,7 @@ define void @test_load_store_bf16_to_f32(ptr addrspace(1) %in, ptr addrspace(1)
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_load_u16 v0, v[0:1], off
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 16, v0
 ; GFX1250-NEXT:    global_store_b32 v[2:3], v0, off
 ; GFX1250-NEXT:    s_set_pc_i64 s[30:31]
@@ -2872,6 +2874,7 @@ define void @test_load_store_bf16_to_f64(ptr addrspace(1) %in, ptr addrspace(1)
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_load_u16 v0, v[0:1], off
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 16, v0
 ; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX1250-NEXT:    v_cvt_f64_f32_e32 v[0:1], v0
@@ -6850,6 +6853,7 @@ define <2 x float> @global_extload_v2bf16_to_v2f32(ptr addrspace(1) %ptr) {
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_load_b32 v1, v[0:1], off
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 16, v1
 ; GFX1250-NEXT:    v_and_b32_e32 v1, 0xffff0000, v1
 ; GFX1250-NEXT:    s_set_pc_i64 s[30:31]
@@ -6943,6 +6947,7 @@ define <3 x float> @global_extload_v3bf16_to_v3f32(ptr addrspace(1) %ptr) {
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_load_b64 v[2:3], v[0:1], off
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 16, v2
 ; GFX1250-NEXT:    v_and_b32_e32 v1, 0xffff0000, v2
 ; GFX1250-NEXT:    v_lshlrev_b32_e32 v2, 16, v3
@@ -7033,6 +7038,7 @@ define <4 x float> @global_extload_v4bf16_to_v4f32(ptr addrspace(1) %ptr) {
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_load_b64 v[2:3], v[0:1], off
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 16, v2
 ; GFX1250-NEXT:    v_and_b32_e32 v1, 0xffff0000, v2
 ; GFX1250-NEXT:    v_lshlrev_b32_e32 v2, 16, v3
@@ -7134,6 +7140,7 @@ define <5 x float> @global_extload_v5bf16_to_v5f32(ptr addrspace(1) %ptr) {
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_load_b128 v[2:5], v[0:1], off
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 16, v2
 ; GFX1250-NEXT:    v_and_b32_e32 v1, 0xffff0000, v2
 ; GFX1250-NEXT:    v_lshlrev_b32_e32 v2, 16, v3
@@ -7251,6 +7258,7 @@ define <6 x float> @global_extload_v6bf16_to_v6f32(ptr addrspace(1) %ptr) {
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_load_b96 v[4:6], v[0:1], off
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_dual_lshlrev_b32 v0, 16, v4 :: v_dual_lshlrev_b32 v2, 16, v5
 ; GFX1250-NEXT:    v_and_b32_e32 v1, 0xffff0000, v4
 ; GFX1250-NEXT:    v_and_b32_e32 v3, 0xffff0000, v5
@@ -7367,6 +7375,7 @@ define <8 x float> @global_extload_v8bf16_to_v8f32(ptr addrspace(1) %ptr) {
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_load_b128 v[4:7], v[0:1], off
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_dual_lshlrev_b32 v0, 16, v4 :: v_dual_lshlrev_b32 v2, 16, v5
 ; GFX1250-NEXT:    v_and_b32_e32 v1, 0xffff0000, v4
 ; GFX1250-NEXT:    v_and_b32_e32 v3, 0xffff0000, v5
@@ -8001,6 +8010,7 @@ define <2 x double> @global_extload_v2bf16_to_v2f64(ptr addrspace(1) %ptr) {
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_load_b32 v0, v[0:1], off
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_lshlrev_b32_e32 v1, 16, v0
 ; GFX1250-NEXT:    v_and_b32_e32 v2, 0xffff0000, v0
 ; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2)
@@ -8241,6 +8251,7 @@ define <4 x double> @global_extload_v4bf16_to_v4f64(ptr addrspace(1) %ptr) {
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_load_b64 v[2:3], v[0:1], off
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_dual_lshlrev_b32 v0, 16, v2 :: v_dual_lshlrev_b32 v4, 16, v3
 ; GFX1250-NEXT:    v_and_b32_e32 v2, 0xffff0000, v2
 ; GFX1250-NEXT:    v_and_b32_e32 v6, 0xffff0000, v3
@@ -8377,6 +8388,7 @@ define <5 x double> @global_extload_v5bf16_to_v5f64(ptr addrspace(1) %ptr) {
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_load_b128 v[2:5], v[0:1], off
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_dual_lshlrev_b32 v0, 16, v2 :: v_dual_lshlrev_b32 v5, 16, v3
 ; GFX1250-NEXT:    v_and_b32_e32 v2, 0xffff0000, v2
 ; GFX1250-NEXT:    v_and_b32_e32 v6, 0xffff0000, v3
@@ -8522,6 +8534,7 @@ define <6 x double> @global_extload_v6bf16_to_v6f64(ptr addrspace(1) %ptr) {
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_load_b96 v[4:6], v[0:1], off
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 16, v4
 ; GFX1250-NEXT:    v_and_b32_e32 v2, 0xffff0000, v4
 ; GFX1250-NEXT:    v_lshlrev_b32_e32 v4, 16, v5
@@ -8693,6 +8706,7 @@ define <8 x double> @global_extload_v8bf16_to_v8f64(ptr addrspace(1) %ptr) {
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_load_b128 v[8:11], v[0:1], off
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_dual_lshlrev_b32 v0, 16, v8 :: v_dual_lshlrev_b32 v4, 16, v9
 ; GFX1250-NEXT:    v_and_b32_e32 v2, 0xffff0000, v8
 ; GFX1250-NEXT:    v_and_b32_e32 v6, 0xffff0000, v9
diff --git a/llvm/test/CodeGen/AMDGPU/branch-relaxation-gfx1250.ll b/llvm/test/CodeGen/AMDGPU/branch-relaxation-gfx1250.ll
index f465e3c505c02..31307b245bafe 100644
--- a/llvm/test/CodeGen/AMDGPU/branch-relaxation-gfx1250.ll
+++ b/llvm/test/CodeGen/AMDGPU/branch-relaxation-gfx1250.ll
@@ -152,7 +152,6 @@ define amdgpu_kernel void @min_long_forward_vbranch(ptr addrspace(1) %arg) #0 {
 ; GCN-NEXT:    s_wait_kmcnt 0x0
 ; GCN-NEXT:    global_load_b32 v2, v0, s[0:1] scale_offset scope:SCOPE_SYS
 ; GCN-NEXT:    s_wait_loadcnt 0x0
-; GCN-NEXT:    s_wait_xcnt 0x0
 ; GCN-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GCN-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GCN-NEXT:    v_add_nc_u64_e32 v[0:1], s[0:1], v[0:1]
diff --git a/llvm/test/CodeGen/AMDGPU/flat-load-saddr-to-vaddr.ll b/llvm/test/CodeGen/AMDGPU/flat-load-saddr-to-vaddr.ll
index e8efa859ce13f..213233e802a96 100644
--- a/llvm/test/CodeGen/AMDGPU/flat-load-saddr-to-vaddr.ll
+++ b/llvm/test/CodeGen/AMDGPU/flat-load-saddr-to-vaddr.ll
@@ -27,7 +27,6 @@ define amdgpu_kernel void @test_move_load_address_to_vgpr(ptr addrspace(1) nocap
 ; GCN-NEXT:    s_wait_dscnt 0x0
 ; GCN-NEXT:    flat_load_b32 v3, v[0:1] scope:SCOPE_SYS
 ; GCN-NEXT:    s_wait_loadcnt 0x0
-; GCN-NEXT:    s_wait_xcnt 0x0
 ; GCN-NEXT:    v_add_nc_u64_e32 v[0:1], 4, v[0:1]
 ; GCN-NEXT:    v_add_co_u32 v2, s0, v2, 1
 ; GCN-NEXT:    s_and_b32 vcc_lo, exec_lo, s0
diff --git a/llvm/test/CodeGen/AMDGPU/flat-saddr-load.ll b/llvm/test/CodeGen/AMDGPU/flat-saddr-load.ll
index 31344c78990b8..f5feeb2f49171 100644
--- a/llvm/test/CodeGen/AMDGPU/flat-saddr-load.ll
+++ b/llvm/test/CodeGen/AMDGPU/flat-saddr-load.ll
@@ -2107,7 +2107,7 @@ define amdgpu_ps void @flat_addr_64bit_lsr_iv(ptr inreg %arg) {
 ; GFX1250-SDAG-NEXT:    s_mov_b64 s[0:1], 0
 ; GFX1250-SDAG-NEXT:  .LBB116_1: ; %bb3
 ; GFX1250-SDAG-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
+; GFX1250-SDAG-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-SDAG-NEXT:    s_add_nc_u64 s[4:5], s[2:3], s[0:1]
 ; GFX1250-SDAG-NEXT:    s_add_nc_u64 s[0:1], s[0:1], 4
 ; GFX1250-SDAG-NEXT:    s_wait_dscnt 0x0
@@ -2126,7 +2126,6 @@ define amdgpu_ps void @flat_addr_64bit_lsr_iv(ptr inreg %arg) {
 ; GFX1250-GISEL-NEXT:  .LBB116_1: ; %bb3
 ; GFX1250-GISEL-NEXT:    ; =>This Inner Loop Header: Depth=1
 ; GFX1250-GISEL-NEXT:    s_wait_dscnt 0x0
-; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1)
 ; GFX1250-GISEL-NEXT:    v_add_co_u32 v4, vcc_lo, v0, v2
 ; GFX1250-GISEL-NEXT:    v_add_co_ci_u32_e64 v5, null, v1, v3, vcc_lo
@@ -2162,7 +2161,7 @@ define amdgpu_ps void @flat_addr_64bit_lsr_iv_multiload(ptr inreg %arg, ptr inre
 ; GFX1250-SDAG-NEXT:    s_mov_b64 s[0:1], 0
 ; GFX1250-SDAG-NEXT:  .LBB117_1: ; %bb3
 ; GFX1250-SDAG-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
+; GFX1250-SDAG-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-SDAG-NEXT:    s_add_nc_u64 s[4:5], s[2:3], s[0:1]
 ; GFX1250-SDAG-NEXT:    s_add_nc_u64 s[0:1], s[0:1], 4
 ; GFX1250-SDAG-NEXT:    s_wait_dscnt 0x0
@@ -2183,7 +2182,6 @@ define amdgpu_ps void @flat_addr_64bit_lsr_iv_multiload(ptr inreg %arg, ptr inre
 ; GFX1250-GISEL-NEXT:    v_mov_b64_e32 v[2:3], s[0:1]
 ; GFX1250-GISEL-NEXT:  .LBB117_1: ; %bb3
 ; GFX1250-GISEL-NEXT:    ; =>This Inner Loop Header: Depth=1
-; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1)
 ; GFX1250-GISEL-NEXT:    v_add_co_u32 v4, vcc_lo, v0, v2
 ; GFX1250-GISEL-NEXT:    v_add_co_ci_u32_e64 v5, null, v1, v3, vcc_lo
diff --git a/llvm/test/CodeGen/AMDGPU/fmin3.ll b/llvm/test/CodeGen/AMDGPU/fmin3.ll
index 6a6f232c55e24..2756472652bc9 100644
--- a/llvm/test/CodeGen/AMDGPU/fmin3.ll
+++ b/llvm/test/CodeGen/AMDGPU/fmin3.ll
@@ -1233,7 +1233,6 @@ define amdgpu_kernel void @test_fmin3_olt_0_f64(ptr addrspace(1) %out, ptr addrs
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-NEXT:    buffer_load_b64 v[2:3], off, s[16:19], null scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x1
 ; GFX1250-NEXT:    s_mov_b32 s4, s14
 ; GFX1250-NEXT:    s_mov_b32 s5, s15
 ; GFX1250-NEXT:    s_mov_b32 s0, s8
@@ -1443,7 +1442,6 @@ define amdgpu_kernel void @test_fmin3_olt_1_f64(ptr addrspace(1) %out, ptr addrs
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-NEXT:    buffer_load_b64 v[2:3], off, s[16:19], null scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x1
 ; GFX1250-NEXT:    s_mov_b32 s4, s14
 ; GFX1250-NEXT:    s_mov_b32 s5, s15
 ; GFX1250-NEXT:    s_mov_b32 s0, s8
diff --git a/llvm/test/CodeGen/AMDGPU/fp-atomics-gfx942.ll b/llvm/test/CodeGen/AMDGPU/fp-atomics-gfx942.ll
index ba761bedb905c..9e5a4428b011f 100644
--- a/llvm/test/CodeGen/AMDGPU/fp-atomics-gfx942.ll
+++ b/llvm/test/CodeGen/AMDGPU/fp-atomics-gfx942.ll
@@ -38,7 +38,6 @@ define amdgpu_kernel void @flat_atomic_fadd_f32_noret_pat(ptr %ptr) {
 ; GFX1250-NEXT:    v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, 4.0
 ; GFX1250-NEXT:    global_wb scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    flat_atomic_add_f32 v0, v1, s[0:1] scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt_dscnt 0x0
@@ -80,7 +79,6 @@ define amdgpu_kernel void @flat_atomic_fadd_f32_noret_pat_ieee(ptr %ptr) #0 {
 ; GFX1250-NEXT:    v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, 4.0
 ; GFX1250-NEXT:    global_wb scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    flat_atomic_add_f32 v0, v1, s[0:1] scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt_dscnt 0x0
diff --git a/llvm/test/CodeGen/AMDGPU/fp64-atomics-gfx90a.ll b/llvm/test/CodeGen/AMDGPU/fp64-atomics-gfx90a.ll
index 6484c2f82ff94..831af7b6c10ba 100644
--- a/llvm/test/CodeGen/AMDGPU/fp64-atomics-gfx90a.ll
+++ b/llvm/test/CodeGen/AMDGPU/fp64-atomics-gfx90a.ll
@@ -1473,7 +1473,6 @@ define amdgpu_kernel void @global_atomic_fadd_f64_noret_pat(ptr addrspace(1) %pt
 ; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1250-NEXT:    global_wb scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
@@ -1516,7 +1515,6 @@ define amdgpu_kernel void @global_atomic_fadd_f64_noret_pat_agent(ptr addrspace(
 ; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1250-NEXT:    global_wb scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
@@ -1561,7 +1559,6 @@ define amdgpu_kernel void @global_atomic_fadd_f64_noret_pat_system(ptr addrspace
 ; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1250-NEXT:    global_wb scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
@@ -1604,7 +1601,6 @@ define amdgpu_kernel void @global_atomic_fadd_f64_noret_pat_flush(ptr addrspace(
 ; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1250-NEXT:    global_wb scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
@@ -1776,7 +1772,6 @@ define amdgpu_kernel void @global_atomic_fadd_f64_noret_pat_agent_safe(ptr addrs
 ; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1250-NEXT:    global_wb scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
@@ -1821,7 +1816,6 @@ define amdgpu_kernel void @flat_atomic_fadd_f64_noret_pat(ptr %ptr) #1 {
 ; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1250-NEXT:    global_wb scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    flat_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt_dscnt 0x0
@@ -1864,7 +1858,6 @@ define amdgpu_kernel void @flat_atomic_fadd_f64_noret_pat_agent(ptr %ptr) #1 {
 ; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1250-NEXT:    global_wb scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    flat_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt_dscnt 0x0
@@ -1909,7 +1902,6 @@ define amdgpu_kernel void @flat_atomic_fadd_f64_noret_pat_system(ptr %ptr) #1 {
 ; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1250-NEXT:    global_wb scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    flat_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
@@ -2083,7 +2075,6 @@ define amdgpu_kernel void @flat_atomic_fadd_f64_noret_pat_agent_safe(ptr %ptr) {
 ; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1250-NEXT:    global_wb scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    flat_atomic_add_f64 v2, v[0:1], s[0:1] scope:SCOPE_DEV
 ; GFX1250-NEXT:    s_wait_storecnt_dscnt 0x0
diff --git a/llvm/test/CodeGen/AMDGPU/global-atomicrmw-fadd.ll b/llvm/test/CodeGen/AMDGPU/global-atomicrmw-fadd.ll
index 1978e68fdae9c..867cc6e2f8d58 100644
--- a/llvm/test/CodeGen/AMDGPU/global-atomicrmw-fadd.ll
+++ b/llvm/test/CodeGen/AMDGPU/global-atomicrmw-fadd.ll
@@ -8814,7 +8814,7 @@ define half @global_agent_atomic_fadd_ret_f16__amdgpu_no_fine_grained_memory(ptr
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v5, v7
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-TRUE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_cbranch_execnz .LBB44_1
 ; GFX1250-TRUE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -8857,7 +8857,7 @@ define half @global_agent_atomic_fadd_ret_f16__amdgpu_no_fine_grained_memory(ptr
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v5, v7
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-FAKE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_cbranch_execnz .LBB44_1
 ; GFX1250-FAKE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -9322,7 +9322,7 @@ define half @global_agent_atomic_fadd_ret_f16__offset12b_pos__amdgpu_no_fine_gra
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v5, v7
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-TRUE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_cbranch_execnz .LBB45_1
 ; GFX1250-TRUE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -9365,7 +9365,7 @@ define half @global_agent_atomic_fadd_ret_f16__offset12b_pos__amdgpu_no_fine_gra
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v5, v7
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-FAKE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_cbranch_execnz .LBB45_1
 ; GFX1250-FAKE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -9844,7 +9844,7 @@ define half @global_agent_atomic_fadd_ret_f16__offset12b_neg__amdgpu_no_fine_gra
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v5, v7
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-TRUE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_cbranch_execnz .LBB46_1
 ; GFX1250-TRUE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -9888,7 +9888,7 @@ define half @global_agent_atomic_fadd_ret_f16__offset12b_neg__amdgpu_no_fine_gra
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v5, v7
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-FAKE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_cbranch_execnz .LBB46_1
 ; GFX1250-FAKE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -10365,7 +10365,6 @@ define void @global_agent_atomic_fadd_noret_f16__amdgpu_no_fine_grained_memory(p
 ; GFX1250-TRUE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v4, v5
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-TRUE16-NEXT:    v_mov_b32_e32 v5, v4
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -10407,7 +10406,6 @@ define void @global_agent_atomic_fadd_noret_f16__amdgpu_no_fine_grained_memory(p
 ; GFX1250-FAKE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v4, v5
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-FAKE16-NEXT:    v_mov_b32_e32 v5, v4
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -10857,7 +10855,6 @@ define void @global_agent_atomic_fadd_noret_f16__offset12b_pos__amdgpu_no_fine_g
 ; GFX1250-TRUE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v4, v5
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-TRUE16-NEXT:    v_mov_b32_e32 v5, v4
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -10899,7 +10896,6 @@ define void @global_agent_atomic_fadd_noret_f16__offset12b_pos__amdgpu_no_fine_g
 ; GFX1250-FAKE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v4, v5
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-FAKE16-NEXT:    v_mov_b32_e32 v5, v4
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -11363,7 +11359,6 @@ define void @global_agent_atomic_fadd_noret_f16__offset12b_neg__amdgpu_no_fine_g
 ; GFX1250-TRUE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v4, v5
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-TRUE16-NEXT:    v_mov_b32_e32 v5, v4
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -11406,7 +11401,6 @@ define void @global_agent_atomic_fadd_noret_f16__offset12b_neg__amdgpu_no_fine_g
 ; GFX1250-FAKE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v4, v5
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-FAKE16-NEXT:    v_mov_b32_e32 v5, v4
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -11861,7 +11855,7 @@ define half @global_agent_atomic_fadd_ret_f16__offset12b_pos__align4__amdgpu_no_
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v3, v5
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-TRUE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_cbranch_execnz .LBB50_1
 ; GFX1250-TRUE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -11893,7 +11887,7 @@ define half @global_agent_atomic_fadd_ret_f16__offset12b_pos__align4__amdgpu_no_
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v3, v5
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-FAKE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_cbranch_execnz .LBB50_1
 ; GFX1250-FAKE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -12245,7 +12239,6 @@ define void @global_agent_atomic_fadd_noret_f16__offset12b__align4_pos__amdgpu_n
 ; GFX1250-TRUE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v3, v5
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-TRUE16-NEXT:    v_mov_b32_e32 v5, v3
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -12276,7 +12269,6 @@ define void @global_agent_atomic_fadd_noret_f16__offset12b__align4_pos__amdgpu_n
 ; GFX1250-FAKE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v3, v5
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-FAKE16-NEXT:    v_mov_b32_e32 v5, v3
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -12631,7 +12623,7 @@ define half @global_system_atomic_fadd_ret_f16__offset12b_pos__amdgpu_no_fine_gr
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_SYS
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v5, v7
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-TRUE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_cbranch_execnz .LBB52_1
 ; GFX1250-TRUE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -12674,7 +12666,7 @@ define half @global_system_atomic_fadd_ret_f16__offset12b_pos__amdgpu_no_fine_gr
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_SYS
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v5, v7
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-FAKE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_cbranch_execnz .LBB52_1
 ; GFX1250-FAKE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -13154,7 +13146,6 @@ define void @global_system_atomic_fadd_noret_f16__offset12b_pos__amdgpu_no_fine_
 ; GFX1250-TRUE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_SYS
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v4, v5
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-TRUE16-NEXT:    v_mov_b32_e32 v5, v4
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -13196,7 +13187,6 @@ define void @global_system_atomic_fadd_noret_f16__offset12b_pos__amdgpu_no_fine_
 ; GFX1250-FAKE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_SYS
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v4, v5
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-FAKE16-NEXT:    v_mov_b32_e32 v5, v4
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -13676,7 +13666,7 @@ define bfloat @global_agent_atomic_fadd_ret_bf16__amdgpu_no_fine_grained_memory(
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v5, v7
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-TRUE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_cbranch_execnz .LBB54_1
 ; GFX1250-TRUE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -13722,7 +13712,7 @@ define bfloat @global_agent_atomic_fadd_ret_bf16__amdgpu_no_fine_grained_memory(
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v5, v7
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-FAKE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_cbranch_execnz .LBB54_1
 ; GFX1250-FAKE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -14273,7 +14263,7 @@ define bfloat @global_agent_atomic_fadd_ret_bf16__offset12b_pos__amdgpu_no_fine_
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v5, v7
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-TRUE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_cbranch_execnz .LBB55_1
 ; GFX1250-TRUE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -14319,7 +14309,7 @@ define bfloat @global_agent_atomic_fadd_ret_bf16__offset12b_pos__amdgpu_no_fine_
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v5, v7
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-FAKE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_cbranch_execnz .LBB55_1
 ; GFX1250-FAKE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -14888,7 +14878,7 @@ define bfloat @global_agent_atomic_fadd_ret_bf16__offset12b_neg__amdgpu_no_fine_
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v5, v7
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-TRUE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_cbranch_execnz .LBB56_1
 ; GFX1250-TRUE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -14936,7 +14926,7 @@ define bfloat @global_agent_atomic_fadd_ret_bf16__offset12b_neg__amdgpu_no_fine_
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v5, v7
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-FAKE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_cbranch_execnz .LBB56_1
 ; GFX1250-FAKE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -15502,7 +15492,6 @@ define void @global_agent_atomic_fadd_noret_bf16__amdgpu_no_fine_grained_memory(
 ; GFX1250-TRUE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v4, v5
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-TRUE16-NEXT:    v_mov_b32_e32 v5, v4
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -15547,7 +15536,6 @@ define void @global_agent_atomic_fadd_noret_bf16__amdgpu_no_fine_grained_memory(
 ; GFX1250-FAKE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v4, v5
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-FAKE16-NEXT:    v_mov_b32_e32 v5, v4
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -16081,7 +16069,6 @@ define void @global_agent_atomic_fadd_noret_bf16__offset12b_pos__amdgpu_no_fine_
 ; GFX1250-TRUE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v4, v5
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-TRUE16-NEXT:    v_mov_b32_e32 v5, v4
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -16126,7 +16113,6 @@ define void @global_agent_atomic_fadd_noret_bf16__offset12b_pos__amdgpu_no_fine_
 ; GFX1250-FAKE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v2, v3
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-FAKE16-NEXT:    v_mov_b32_e32 v3, v2
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -16678,7 +16664,6 @@ define void @global_agent_atomic_fadd_noret_bf16__offset12b_neg__amdgpu_no_fine_
 ; GFX1250-TRUE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v4, v5
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-TRUE16-NEXT:    v_mov_b32_e32 v5, v4
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -16725,7 +16710,6 @@ define void @global_agent_atomic_fadd_noret_bf16__offset12b_neg__amdgpu_no_fine_
 ; GFX1250-FAKE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v2, v3
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-FAKE16-NEXT:    v_mov_b32_e32 v3, v2
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -17269,7 +17253,7 @@ define bfloat @global_agent_atomic_fadd_ret_bf16__offset12b_pos__align4__amdgpu_
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v3, v5
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-TRUE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_cbranch_execnz .LBB60_1
 ; GFX1250-TRUE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -17305,7 +17289,7 @@ define bfloat @global_agent_atomic_fadd_ret_bf16__offset12b_pos__align4__amdgpu_
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v3, v5
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-FAKE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_cbranch_execnz .LBB60_1
 ; GFX1250-FAKE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -17753,7 +17737,6 @@ define void @global_agent_atomic_fadd_noret_bf16__offset12b__align4_pos__amdgpu_
 ; GFX1250-TRUE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v3, v5
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-TRUE16-NEXT:    v_mov_b32_e32 v5, v3
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -17788,7 +17771,6 @@ define void @global_agent_atomic_fadd_noret_bf16__offset12b__align4_pos__amdgpu_
 ; GFX1250-FAKE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_DEV
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v2, v3
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-FAKE16-NEXT:    v_mov_b32_e32 v3, v2
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -18238,7 +18220,7 @@ define bfloat @global_system_atomic_fadd_ret_bf16__offset12b_pos__amdgpu_no_fine
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_SYS
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v5, v7
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-TRUE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_cbranch_execnz .LBB62_1
 ; GFX1250-TRUE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -18284,7 +18266,7 @@ define bfloat @global_system_atomic_fadd_ret_bf16__offset12b_pos__amdgpu_no_fine
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_SYS
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v5, v7
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
+; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
 ; GFX1250-FAKE16-NEXT:    s_and_not1_b32 exec_lo, exec_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_cbranch_execnz .LBB62_1
 ; GFX1250-FAKE16-NEXT:  ; %bb.2: ; %atomicrmw.end
@@ -18854,7 +18836,6 @@ define void @global_system_atomic_fadd_noret_bf16__offset12b_pos__amdgpu_no_fine
 ; GFX1250-TRUE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-TRUE16-NEXT:    global_inv scope:SCOPE_SYS
 ; GFX1250-TRUE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v4, v5
-; GFX1250-TRUE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-TRUE16-NEXT:    v_mov_b32_e32 v5, v4
 ; GFX1250-TRUE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-TRUE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
@@ -18899,7 +18880,6 @@ define void @global_system_atomic_fadd_noret_bf16__offset12b_pos__amdgpu_no_fine
 ; GFX1250-FAKE16-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-FAKE16-NEXT:    global_inv scope:SCOPE_SYS
 ; GFX1250-FAKE16-NEXT:    v_cmp_eq_u32_e32 vcc_lo, v2, v3
-; GFX1250-FAKE16-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-FAKE16-NEXT:    v_mov_b32_e32 v3, v2
 ; GFX1250-FAKE16-NEXT:    s_or_b32 s0, vcc_lo, s0
 ; GFX1250-FAKE16-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.is.private.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.is.private.ll
index 3324018ca7237..5b36d4cefa2e3 100644
--- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.is.private.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.is.private.ll
@@ -66,7 +66,6 @@ define amdgpu_kernel void @is_private_vgpr(ptr addrspace(1) %ptr.ptr) {
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_load_b64 v[0:1], v0, s[0:1] scale_offset scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_xor_b32_e32 v0, src_flat_scratch_base_hi, v1
 ; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX1250-NEXT:    v_cmp_gt_u32_e32 vcc_lo, 0x4000000, v0
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.is.shared.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.is.shared.ll
index 355d0026091d9..7aecae901becf 100644
--- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.is.shared.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.is.shared.ll
@@ -101,7 +101,6 @@ define amdgpu_kernel void @is_local_vgpr(ptr addrspace(1) %ptr.ptr) {
 ; GFX1250-NEXT:    global_load_b64 v[0:1], v0, s[2:3] scale_offset scope:SCOPE_SYS
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v1
-; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_cndmask_b32_e64 v0, 0, 1, vcc_lo
 ; GFX1250-NEXT:    global_store_b32 v[0:1], v0, off
 ; GFX1250-NEXT:    s_endpgm
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.load.tr.gfx1250.w32.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.load.tr.gfx1250.w32.ll
index d9f2fc55709a6..f7ed5341141d4 100644
--- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.load.tr.gfx1250.w32.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.load.tr.gfx1250.w32.ll
@@ -330,6 +330,7 @@ define { i32, <3 x i32> } @global_load_tr6_b96_vaddr_no_align2_requirement(ptr a
 ; GFX1250-NEXT:    s_wait_kmcnt 0x0
 ; GFX1250-NEXT:    global_load_tr6_b96 v[2:4], v[0:1], off offset:32
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, v2
 ; GFX1250-NEXT:    v_dual_mov_b32 v2, v3 :: v_dual_mov_b32 v3, v4
 ; GFX1250-NEXT:    s_set_pc_i64 s[30:31]
@@ -348,6 +349,7 @@ define { i32, <3 x i32> } @global_load_tr6_b96_saddr_no_align2_requirement(ptr a
 ; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX1250-NEXT:    global_load_tr6_b96 v[2:4], v0, s[0:1] offset:32
 ; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-NEXT:    v_dual_mov_b32 v0, 0 :: v_dual_mov_b32 v1, v2
 ; GFX1250-NEXT:    v_dual_mov_b32 v2, v3 :: v_dual_mov_b32 v3, v4
 ; GFX1250-NEXT:    s_set_pc_i64 s[30:31]