[AMDGPU] Update dlc usage for GFX11

[foad]

Summary

In GFX10 dlc controlled L1 cache bypass. In GFX11 it has been repurposed
to control MALL NOALLOC, and glc controls L1 as well as L0 cache bypass.

Update the documentation and SIMemoryLegalizer accordingly. Set dlc for
nontemporal and volatile accesses.

llvm/docs/AMDGPUUsage.rst

* Each L1 quadrant of a single SA accesses a different L2 channel. Each L1
   work-groups (which may be executing on different SAs) of an agent can be
   reordered relative to each other. A ``s_waitcnt vmcnt(0) & vscnt(0)`` is
   required to ensure synchronization between vector memory operations of
   different SAs. It ensures a previous vector memory operation has completed
   before executing a subsequent vector memory and so can be used to meet the
   requirements of acquire, release and sequential consistency.
 * The L2 cache can be kept coherent with other agents on some targets, or ranges
   of virtual addresses can be set up to bypass it to ensure system coherence.
-* On GFX10.3 a memory attached last level (MALL) cache exists for GPU memory.
+* On GFX10.3 and GFX11 a memory attached last level (MALL) cache exists for GPU memory.
   The MALL cache is fully coherent with GPU memory and has no impact on system
   coherence. All agents (GPU and CPU) access GPU memory through the MALL cache.
 
 Scalar memory operations are only used to access memory that is proven to not
 change during the execution of the kernel dispatch. This includes constant
 address space and global address space for program scope ``const`` variables.
 Therefore, the kernel machine code does not have to maintain the scalar cache to
 ensure it is coherent with the vector caches. The scalar and vector caches are
.. table:: AMDHSA Memory Model Code Sequences GFX10-GFX11
      ------------------------------------------------------------------------------------
      load         *none*       *none*         - global   - !volatile & !nontemporal
                                               - generic
                                               - private    1. buffer/global/flat_load
                                               - constant
                                                          - !volatile & nontemporal
 
                                                            1. buffer/global/flat_load
-                                                              slc=1
+                                                              slc=1 dlc=1
+
+                                                            - If GFX10, omit dlc=1.
 
                                                          - volatile
 
                                                            1. buffer/global/flat_load
                                                               glc=1 dlc=1
+
                                                            2. s_waitcnt vmcnt(0)
 
                                                             - Must happen before
                                                               any following volatile
                                                               global/generic
                                                               load/store.
                                                             - Ensures that
                                                               volatile
                                                               operations to
                                                               different
                                                               addresses will not
                                                               be reordered by
                                                               hardware.
 
      load         *none*       *none*         - local    1. ds_load
      store        *none*       *none*         - global   - !volatile & !nontemporal
                                               - generic
                                               - private    1. buffer/global/flat_store
                                               - constant
                                                          - !volatile & nontemporal
 
                                                            1. buffer/global/flat_store
-                                                              glc=1 slc=1
+                                                              glc=1 slc=1 dlc=1
+
+                                                            - If GFX10, omit dlc=1.
 
                                                          - volatile
 
                                                            1. buffer/global/flat_store
+                                                              dlc=1
+
+                                                            - If GFX10, omit dlc=1.
+
                                                            2. s_waitcnt vscnt(0)
 
                                                             - Must happen before
                                                               any following volatile
                                                               global/generic
                                                               load/store.
                                                             - Ensures that
                                                               volatile
load atomic  monotonic    - workgroup    - global   1. buffer/global/flat_load
                                                            - If CU wavefront execution
                                                              mode, omit glc=1.
 
      load atomic  monotonic    - singlethread - local    1. ds_load
                                - wavefront
                                - workgroup
      load atomic  monotonic    - agent        - global   1. buffer/global/flat_load
                                - system       - generic     glc=1 dlc=1
+
+                                                           - If GFX11, omit dlc=1.
+
      store atomic monotonic    - singlethread - global   1. buffer/global/flat_store
                                - wavefront    - generic
                                - workgroup
                                - agent
                                - system
      store atomic monotonic    - singlethread - local    1. ds_store
                                - wavefront
                                - workgroup
load atomic  acquire      - workgroup    - generic  1. flat_load glc=1
                                                              mode, omit.
                                                            - Ensures that
                                                              following
                                                              loads will not see
                                                              stale data.
 
      load atomic  acquire      - agent        - global   1. buffer/global_load
                                - system                     glc=1 dlc=1
+
+                                                           - If GFX11, omit dlc=1.
+
                                                          2. s_waitcnt vmcnt(0)
 
                                                            - Must happen before
                                                              following
                                                              buffer_gl*_inv.
                                                            - Ensures the load
                                                              has completed
                                                              before invalidating
                                                              the caches.
 
                                                          3. buffer_gl0_inv;
                                                             buffer_gl1_inv
 
                                                            - Must happen before
                                                              any following
                                                              global/generic
                                                              load/load
                                                              atomic/atomicrmw.
                                                            - Ensures that
                                                              following
                                                              loads will not see
                                                              stale global data.
 
      load atomic  acquire      - agent        - generic  1. flat_load glc=1 dlc=1
-                               - system                  2. s_waitcnt vmcnt(0) &
+                               - system
+                                                           - If GFX11, omit dlc=1.
+
+                                                         2. s_waitcnt vmcnt(0) &
                                                             lgkmcnt(0)
 
                                                            - If OpenCL omit
                                                              lgkmcnt(0).
                                                            - Must happen before
                                                              following
                                                              buffer_gl*_invl.
                                                            - Ensures the flat_load

llvm/lib/Target/AMDGPU/SIMemoryLegalizer.cpp

bool insertWait(MachineBasicBlock::iterator &MI,
                   Position Pos) const override;
 
   bool insertAcquire(MachineBasicBlock::iterator &MI,
                      SIAtomicScope Scope,
                      SIAtomicAddrSpace AddrSpace,
                      Position Pos) const override;
 };
 
+class SIGfx11CacheControl : public SIGfx10CacheControl {
+public:
+  SIGfx11CacheControl(const GCNSubtarget &ST) : SIGfx10CacheControl(ST) {}
+
+  bool enableLoadCacheBypass(const MachineBasicBlock::iterator &MI,
+                             SIAtomicScope Scope,
+                             SIAtomicAddrSpace AddrSpace) const override;
+
+  bool enableVolatileAndOrNonTemporal(MachineBasicBlock::iterator &MI,
+                                      SIAtomicAddrSpace AddrSpace, SIMemOp Op,
+                                      bool IsVolatile,
+                                      bool IsNonTemporal) const override;
+};
+
 class SIMemoryLegalizer final : public MachineFunctionPass {
 private:
 
   /// Cache Control.
   std::unique_ptr<SICacheControl> CC = nullptr;
 
   /// List of atomic pseudo instructions.
   std::list<MachineBasicBlock::iterator> AtomicPseudoMIs;
std::unique_ptr<SICacheControl> SICacheControl::create(const GCNSubtarget &ST) {
   if (ST.hasGFX940Insts())
     return std::make_unique<SIGfx940CacheControl>(ST);
   if (ST.hasGFX90AInsts())
     return std::make_unique<SIGfx90ACacheControl>(ST);
   if (Generation <= AMDGPUSubtarget::SOUTHERN_ISLANDS)
     return std::make_unique<SIGfx6CacheControl>(ST);
   if (Generation < AMDGPUSubtarget::GFX10)
     return std::make_unique<SIGfx7CacheControl>(ST);
+  if (Generation < AMDGPUSubtarget::GFX11)
     return std::make_unique<SIGfx10CacheControl>(ST);
+  return std::make_unique<SIGfx11CacheControl>(ST);
 }
 
 bool SIGfx6CacheControl::enableLoadCacheBypass(
     const MachineBasicBlock::iterator &MI,
     SIAtomicScope Scope,
     SIAtomicAddrSpace AddrSpace) const {
   assert(MI->mayLoad() && !MI->mayStore());
   bool Changed = false;
bool SIGfx10CacheControl::insertAcquire(MachineBasicBlock::iterator &MI,
   /// Other address spaces do not have a cache.
 
   if (Pos == Position::AFTER)
     --MI;
 
   return Changed;
 }
 
+bool SIGfx11CacheControl::enableLoadCacheBypass(
+    const MachineBasicBlock::iterator &MI, SIAtomicScope Scope,
+    SIAtomicAddrSpace AddrSpace) const {
+  assert(MI->mayLoad() && !MI->mayStore());
+  bool Changed = false;
+
+  if ((AddrSpace & SIAtomicAddrSpace::GLOBAL) != SIAtomicAddrSpace::NONE) {
+    switch (Scope) {
+    case SIAtomicScope::SYSTEM:
+    case SIAtomicScope::AGENT:
+      // Set the L0 and L1 cache policies to MISS_EVICT.
+      // Note: there is no L2 cache coherent bypass control at the ISA level.
+      Changed |= enableGLCBit(MI);
+      break;
+    case SIAtomicScope::WORKGROUP:
+      // In WGP mode the waves of a work-group can be executing on either CU of
+      // the WGP. Therefore need to bypass the L0 which is per CU. Otherwise in
+      // CU mode all waves of a work-group are on the same CU, and so the L0
+      // does not need to be bypassed.
+      if (!ST.isCuModeEnabled())
+        Changed |= enableGLCBit(MI);
+      break;
+    case SIAtomicScope::WAVEFRONT:
+    case SIAtomicScope::SINGLETHREAD:
+      // No cache to bypass.
+      break;
+    default:
+      llvm_unreachable("Unsupported synchronization scope");
+    }
+  }
+
+  /// The scratch address space does not need the global memory caches
+  /// to be bypassed as all memory operations by the same thread are
+  /// sequentially consistent, and no other thread can access scratch
+  /// memory.
+
+  /// Other address spaces do not have a cache.
+
+  return Changed;
+}
+
+bool SIGfx11CacheControl::enableVolatileAndOrNonTemporal(
+    MachineBasicBlock::iterator &MI, SIAtomicAddrSpace AddrSpace, SIMemOp Op,
+    bool IsVolatile, bool IsNonTemporal) const {
+
+  // Only handle load and store, not atomic read-modify-write insructions. The
+  // latter use glc to indicate if the atomic returns a result and so must not
+  // be used for cache control.
+  assert(MI->mayLoad() ^ MI->mayStore());
+
+  // Only update load and store, not LLVM IR atomic read-modify-write
+  // instructions. The latter are always marked as volatile so cannot sensibly
+  // handle it as do not want to pessimize all atomics. Also they do not support
+  // the nontemporal attribute.
+  assert(Op == SIMemOp::LOAD || Op == SIMemOp::STORE);
+
+  bool Changed = false;
+
+  if (IsVolatile) {
+    // Set L0 and L1 cache policy to be MISS_EVICT for load instructions
+    // and MISS_LRU for store instructions.
+    // Note: there is no L2 cache coherent bypass control at the ISA level.
+    if (Op == SIMemOp::LOAD)
+      Changed |= enableGLCBit(MI);
+
+    // Set MALL NOALLOC for load and store instructions.
+    Changed |= enableDLCBit(MI);
+
+    // Ensure operation has completed at system scope to cause all volatile
+    // operations to be visible outside the program in a global order. Do not
+    // request cross address space as only the global address space can be
+    // observable outside the program, so no need to cause a waitcnt for LDS
+    // address space operations.
+    Changed |= insertWait(MI, SIAtomicScope::SYSTEM, AddrSpace, Op, false,
+                          Position::AFTER);
+    return Changed;
+  }
+
+  if (IsNonTemporal) {
+    // For loads setting SLC configures L0 and L1 cache policy to HIT_EVICT
+    // and L2 cache policy to STREAM.
+    // For stores setting both GLC and SLC configures L0 and L1 cache policy
+    // to MISS_EVICT and the L2 cache policy to STREAM.
+    if (Op == SIMemOp::STORE)
+      Changed |= enableGLCBit(MI);
+    Changed |= enableSLCBit(MI);
+
+    // Set MALL NOALLOC for load and store instructions.
+    Changed |= enableDLCBit(MI);
+    return Changed;
+  }
+
+  return Changed;
+}
+
 bool SIMemoryLegalizer::removeAtomicPseudoMIs() {
   if (AtomicPseudoMIs.empty())
     return false;
 
   for (auto &MI : AtomicPseudoMIs)
     MI->eraseFromParent();
 
   AtomicPseudoMIs.clear();

llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-agent.ll

 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: flat_agent_monotonic_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_agent_monotonic_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-CU-NEXT:    s_endpgm
     i32* %in, i32* %out) {
 entry:
   %val = load atomic i32, i32* %in syncscope("agent") monotonic, align 4

 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: flat_agent_acquire_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-WGP-NEXT:    buffer_gl0_inv
 ; GFX11-WGP-NEXT:    buffer_gl1_inv
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_agent_acquire_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-CU-NEXT:    buffer_gl0_inv
 ; GFX11-CU-NEXT:    buffer_gl1_inv
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-CU-NEXT:    s_endpgm
     i32* %in, i32* %out) {

 ; GFX11-WGP-LABEL: flat_agent_seq_cst_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-WGP-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-WGP-NEXT:    buffer_gl0_inv
 ; GFX11-WGP-NEXT:    buffer_gl1_inv
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_agent_seq_cst_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-CU-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-CU-NEXT:    buffer_gl0_inv
 ; GFX11-CU-NEXT:    buffer_gl1_inv
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-CU-NEXT:    s_endpgm
     i32* %in, i32* %out) {

 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: flat_agent_one_as_monotonic_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_agent_one_as_monotonic_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-CU-NEXT:    s_endpgm
     i32* %in, i32* %out) {
 entry:
   %val = load atomic i32, i32* %in syncscope("agent-one-as") monotonic, align 4

 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: flat_agent_one_as_acquire_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    buffer_gl0_inv
 ; GFX11-WGP-NEXT:    buffer_gl1_inv
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_agent_one_as_acquire_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    buffer_gl0_inv
 ; GFX11-CU-NEXT:    buffer_gl1_inv
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-CU-NEXT:    s_endpgm

 ; GFX11-WGP-LABEL: flat_agent_one_as_seq_cst_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    buffer_gl0_inv
 ; GFX11-WGP-NEXT:    buffer_gl1_inv
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_agent_one_as_seq_cst_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    buffer_gl0_inv
 ; GFX11-CU-NEXT:    buffer_gl1_inv
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-CU-NEXT:    s_endpgm

llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-nontemporal.ll

 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: flat_nontemporal_load_0:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] slc
+; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] slc dlc
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_nontemporal_load_0:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] slc
+; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] slc dlc
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-CU-NEXT:    s_endpgm
     i32* %in, i32* %out) {
 entry:
   %val = load i32, i32* %in, align 4, !nontemporal !0

 ;
 ; GFX11-WGP-LABEL: flat_nontemporal_load_1:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_add_co_u32 v0, s0, s0, v0
 ; GFX11-WGP-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
-; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] slc
+; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] slc dlc
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_nontemporal_load_1:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_add_co_u32 v0, s0, s0, v0
 ; GFX11-CU-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
-; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] slc
+; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] slc dlc
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-CU-NEXT:    s_endpgm
     i32* %in, i32* %out) {
 entry:
   %tid = call i32 @llvm.amdgcn.workitem.id.x()

 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1]
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
-; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2 glc slc
+; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2 glc slc dlc
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_nontemporal_store_0:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1]
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
-; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2 glc slc
+; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2 glc slc dlc
 ; GFX11-CU-NEXT:    s_endpgm
     i32* %in, i32* %out) {
 entry:
   %val = load i32, i32* %in, align 4
   store i32 %val, i32* %out, !nontemporal !0
   ret void
 }
 

 ; GFX11-WGP-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX11-WGP-NEXT:    v_add_co_u32 v0, s0, s2, v0
 ; GFX11-WGP-NEXT:    flat_load_b32 v2, v[1:2]
 ; GFX11-WGP-NEXT:    v_add_co_ci_u32_e64 v1, s0, s3, 0, s0
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
-; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2 glc slc
+; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2 glc slc dlc
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_nontemporal_store_1:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX11-CU-NEXT:    v_add_co_u32 v0, s0, s2, v0
 ; GFX11-CU-NEXT:    flat_load_b32 v2, v[1:2]
 ; GFX11-CU-NEXT:    v_add_co_ci_u32_e64 v1, s0, s3, 0, s0
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
-; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2 glc slc
+; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2 glc slc dlc
 ; GFX11-CU-NEXT:    s_endpgm
     i32* %in, i32* %out) {
 entry:
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %val = load i32, i32* %in, align 4
   %out.gep = getelementptr inbounds i32, i32* %out, i32 %tid
   store i32 %val, i32* %out.gep, !nontemporal !0
   ret void
 }
 
 !0 = !{i32 1}
 declare i32 @llvm.amdgcn.workitem.id.x()

llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-system.ll

 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: flat_system_monotonic_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_system_monotonic_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-CU-NEXT:    s_endpgm
     i32* %in, i32* %out) {
 entry:
   %val = load atomic i32, i32* %in monotonic, align 4

 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: flat_system_acquire_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-WGP-NEXT:    buffer_gl0_inv
 ; GFX11-WGP-NEXT:    buffer_gl1_inv
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_system_acquire_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-CU-NEXT:    buffer_gl0_inv
 ; GFX11-CU-NEXT:    buffer_gl1_inv
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-CU-NEXT:    s_endpgm
     i32* %in, i32* %out) {

 ; GFX11-WGP-LABEL: flat_system_seq_cst_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-WGP-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-WGP-NEXT:    buffer_gl0_inv
 ; GFX11-WGP-NEXT:    buffer_gl1_inv
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_system_seq_cst_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-CU-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-CU-NEXT:    buffer_gl0_inv
 ; GFX11-CU-NEXT:    buffer_gl1_inv
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-CU-NEXT:    s_endpgm
     i32* %in, i32* %out) {

 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: flat_system_one_as_monotonic_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_system_one_as_monotonic_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-CU-NEXT:    s_endpgm
     i32* %in, i32* %out) {
 entry:
   %val = load atomic i32, i32* %in syncscope("one-as") monotonic, align 4

 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: flat_system_one_as_acquire_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    buffer_gl0_inv
 ; GFX11-WGP-NEXT:    buffer_gl1_inv
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_system_one_as_acquire_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    buffer_gl0_inv
 ; GFX11-CU-NEXT:    buffer_gl1_inv
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-CU-NEXT:    s_endpgm

 ; GFX11-WGP-LABEL: flat_system_one_as_seq_cst_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    buffer_gl0_inv
 ; GFX11-WGP-NEXT:    buffer_gl1_inv
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_system_one_as_seq_cst_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc dlc
+; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    buffer_gl0_inv
 ; GFX11-CU-NEXT:    buffer_gl1_inv
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2
 ; GFX11-CU-NEXT:    s_endpgm

llvm/test/CodeGen/AMDGPU/memory-legalizer-flat-volatile.ll

 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX11-WGP-NEXT:    flat_load_b32 v2, v[0:1]
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
-; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2
+; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2 dlc
 ; GFX11-WGP-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_nontemporal_store_0:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX11-CU-NEXT:    flat_load_b32 v2, v[0:1]
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
-; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2
+; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2 dlc
 ; GFX11-CU-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-CU-NEXT:    s_endpgm
     i32* %in, i32* %out) {
 entry:
   %val = load i32, i32* %in, align 4
   store volatile i32 %val, i32* %out
   ret void
 }

 ; GFX11-WGP-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX11-WGP-NEXT:    v_add_co_u32 v0, s0, s2, v0
 ; GFX11-WGP-NEXT:    flat_load_b32 v2, v[1:2]
 ; GFX11-WGP-NEXT:    v_add_co_ci_u32_e64 v1, s0, s3, 0, s0
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
-; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2
+; GFX11-WGP-NEXT:    flat_store_b32 v[0:1], v2 dlc
 ; GFX11-WGP-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: flat_nontemporal_store_1:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX11-CU-NEXT:    v_add_co_u32 v0, s0, s2, v0
 ; GFX11-CU-NEXT:    flat_load_b32 v2, v[1:2]
 ; GFX11-CU-NEXT:    v_add_co_ci_u32_e64 v1, s0, s3, 0, s0
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
-; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2
+; GFX11-CU-NEXT:    flat_store_b32 v[0:1], v2 dlc
 ; GFX11-CU-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-CU-NEXT:    s_endpgm
     i32* %in, i32* %out) {
 entry:
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %val = load i32, i32* %in, align 4
   %out.gep = getelementptr inbounds i32, i32* %out, i32 %tid
   store volatile i32 %val, i32* %out.gep

llvm/test/CodeGen/AMDGPU/memory-legalizer-global-agent.ll

 ; GFX940-TGSPLIT-NEXT:    global_store_dword v0, v1, s[2:3]
 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: global_agent_monotonic_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_agent_monotonic_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %val = load atomic i32, i32 addrspace(1)* %in syncscope("agent") monotonic, align 4
   store i32 %val, i32 addrspace(1)* %out
   ret void

 ; GFX940-TGSPLIT-NEXT:    global_store_dword v0, v1, s[2:3]
 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: global_agent_acquire_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    buffer_gl0_inv
 ; GFX11-WGP-NEXT:    buffer_gl1_inv
 ; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_agent_acquire_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    buffer_gl0_inv
 ; GFX11-CU-NEXT:    buffer_gl1_inv
 ; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %val = load atomic i32, i32 addrspace(1)* %in syncscope("agent") acquire, align 4

 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: global_agent_seq_cst_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-WGP-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    buffer_gl0_inv
 ; GFX11-WGP-NEXT:    buffer_gl1_inv
 ; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_agent_seq_cst_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-CU-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    buffer_gl0_inv
 ; GFX11-CU-NEXT:    buffer_gl1_inv
 ; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %val = load atomic i32, i32 addrspace(1)* %in syncscope("agent") seq_cst, align 4

 ; GFX940-TGSPLIT-NEXT:    global_store_dword v0, v1, s[2:3]
 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: global_agent_one_as_monotonic_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_agent_one_as_monotonic_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %val = load atomic i32, i32 addrspace(1)* %in syncscope("agent-one-as") monotonic, align 4
   store i32 %val, i32 addrspace(1)* %out
   ret void

 ; GFX940-TGSPLIT-NEXT:    global_store_dword v0, v1, s[2:3]
 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: global_agent_one_as_acquire_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    buffer_gl0_inv
 ; GFX11-WGP-NEXT:    buffer_gl1_inv
 ; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_agent_one_as_acquire_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    buffer_gl0_inv
 ; GFX11-CU-NEXT:    buffer_gl1_inv
 ; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %val = load atomic i32, i32 addrspace(1)* %in syncscope("agent-one-as") acquire, align 4

 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: global_agent_one_as_seq_cst_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-WGP-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    buffer_gl0_inv
 ; GFX11-WGP-NEXT:    buffer_gl1_inv
 ; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_agent_one_as_seq_cst_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-CU-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    buffer_gl0_inv
 ; GFX11-CU-NEXT:    buffer_gl1_inv
 ; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %val = load atomic i32, i32 addrspace(1)* %in syncscope("agent-one-as") seq_cst, align 4

llvm/test/CodeGen/AMDGPU/memory-legalizer-global-nontemporal.ll

 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: global_nontemporal_load_1:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-WGP-NEXT:    global_load_b32 v0, v0, s[0:1] slc
+; GFX11-WGP-NEXT:    global_load_b32 v0, v0, s[0:1] slc dlc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    global_store_b32 v1, v0, s[2:3]
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_nontemporal_load_1:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-CU-NEXT:    global_load_b32 v0, v0, s[0:1] slc
+; GFX11-CU-NEXT:    global_load_b32 v0, v0, s[0:1] slc dlc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    global_store_b32 v1, v0, s[2:3]
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %val.gep = getelementptr inbounds i32, i32 addrspace(1)* %in, i32 %tid
   %val = load i32, i32 addrspace(1)* %val.gep, align 4, !nontemporal !0

 ; GFX11-WGP-LABEL: global_nontemporal_store_0:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    s_load_b32 s0, s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s0
-; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3] glc slc
+; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3] glc slc dlc
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_nontemporal_store_0:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    s_load_b32 s0, s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s0
-; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3] glc slc
+; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3] glc slc dlc
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %val = load i32, i32 addrspace(1)* %in, align 4
   store i32 %val, i32 addrspace(1)* %out, !nontemporal !0
   ret void
 }
 

 ; GFX11-WGP-LABEL: global_nontemporal_store_1:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    s_load_b32 s0, s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s0
-; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3] glc slc
+; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3] glc slc dlc
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_nontemporal_store_1:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    s_load_b32 s0, s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s0
-; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3] glc slc
+; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3] glc slc dlc
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %val = load i32, i32 addrspace(1)* %in, align 4
   %out.gep = getelementptr inbounds i32, i32 addrspace(1)* %out, i32 %tid
   store i32 %val, i32 addrspace(1)* %out.gep, !nontemporal !0
   ret void
 }
 
 !0 = !{i32 1}
 declare i32 @llvm.amdgcn.workitem.id.x()

llvm/test/CodeGen/AMDGPU/memory-legalizer-global-system.ll

 ; GFX940-TGSPLIT-NEXT:    global_store_dword v0, v1, s[2:3]
 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: global_system_monotonic_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_system_monotonic_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %val = load atomic i32, i32 addrspace(1)* %in monotonic, align 4
   store i32 %val, i32 addrspace(1)* %out
   ret void

 ; GFX940-TGSPLIT-NEXT:    global_store_dword v0, v1, s[2:3]
 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: global_system_acquire_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    buffer_gl0_inv
 ; GFX11-WGP-NEXT:    buffer_gl1_inv
 ; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_system_acquire_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    buffer_gl0_inv
 ; GFX11-CU-NEXT:    buffer_gl1_inv
 ; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %val = load atomic i32, i32 addrspace(1)* %in acquire, align 4

 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: global_system_seq_cst_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-WGP-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    buffer_gl0_inv
 ; GFX11-WGP-NEXT:    buffer_gl1_inv
 ; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_system_seq_cst_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-CU-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    buffer_gl0_inv
 ; GFX11-CU-NEXT:    buffer_gl1_inv
 ; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %val = load atomic i32, i32 addrspace(1)* %in seq_cst, align 4

 ; GFX940-TGSPLIT-NEXT:    global_store_dword v0, v1, s[2:3]
 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: global_system_one_as_monotonic_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_system_one_as_monotonic_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %val = load atomic i32, i32 addrspace(1)* %in syncscope("one-as") monotonic, align 4
   store i32 %val, i32 addrspace(1)* %out
   ret void

 ; GFX940-TGSPLIT-NEXT:    global_store_dword v0, v1, s[2:3]
 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: global_system_one_as_acquire_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    buffer_gl0_inv
 ; GFX11-WGP-NEXT:    buffer_gl1_inv
 ; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_system_one_as_acquire_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    buffer_gl0_inv
 ; GFX11-CU-NEXT:    buffer_gl1_inv
 ; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %val = load atomic i32, i32 addrspace(1)* %in syncscope("one-as") acquire, align 4

 ; GFX940-TGSPLIT-NEXT:    s_endpgm
 ;
 ; GFX11-WGP-LABEL: global_system_one_as_seq_cst_load:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-WGP-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-WGP-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    buffer_gl0_inv
 ; GFX11-WGP-NEXT:    buffer_gl1_inv
 ; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_system_one_as_seq_cst_load:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX11-CU-NEXT:    s_waitcnt_vscnt null, 0x0
-; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc dlc
+; GFX11-CU-NEXT:    global_load_b32 v1, v0, s[0:1] glc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    buffer_gl0_inv
 ; GFX11-CU-NEXT:    buffer_gl1_inv
 ; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3]
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %val = load atomic i32, i32 addrspace(1)* %in syncscope("one-as") seq_cst, align 4

llvm/test/CodeGen/AMDGPU/memory-legalizer-global-volatile.ll

 ; GFX11-WGP-LABEL: global_volatile_store_0:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    s_load_b32 s0, s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s0
-; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3]
+; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3] dlc
 ; GFX11-WGP-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_volatile_store_0:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, 0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    s_load_b32 s0, s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s0
-; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3]
+; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3] dlc
 ; GFX11-CU-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %val = load i32, i32 addrspace(1)* %in, align 4
   store volatile i32 %val, i32 addrspace(1)* %out
   ret void
 }

 ; GFX11-WGP-LABEL: global_volatile_store_1:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    s_load_b32 s0, s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s0
-; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3]
+; GFX11-WGP-NEXT:    global_store_b32 v0, v1, s[2:3] dlc
 ; GFX11-WGP-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: global_volatile_store_1:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b128 s[0:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    s_load_b32 s0, s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s0
-; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3]
+; GFX11-CU-NEXT:    global_store_b32 v0, v1, s[2:3] dlc
 ; GFX11-CU-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(1)* %out) {
 entry:
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %val = load i32, i32 addrspace(1)* %in, align 4
   %out.gep = getelementptr inbounds i32, i32 addrspace(1)* %out, i32 %tid
   store volatile i32 %val, i32 addrspace(1)* %out.gep

llvm/test/CodeGen/AMDGPU/memory-legalizer-private-nontemporal.ll

 ;
 ; GFX11-WGP-LABEL: private_nontemporal_load_0:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_clause 0x1
 ; GFX11-WGP-NEXT:    s_load_b32 s2, s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_load_b64 s[0:1], s[0:1], 0x8
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-WGP-NEXT:    scratch_load_b32 v0, off, s2 slc
+; GFX11-WGP-NEXT:    scratch_load_b32 v0, off, s2 slc dlc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    global_store_b32 v1, v0, s[0:1]
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: private_nontemporal_load_0:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_clause 0x1
 ; GFX11-CU-NEXT:    s_load_b32 s2, s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_load_b64 s[0:1], s[0:1], 0x8
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-CU-NEXT:    scratch_load_b32 v0, off, s2 slc
+; GFX11-CU-NEXT:    scratch_load_b32 v0, off, s2 slc dlc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    global_store_b32 v1, v0, s[0:1]
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(5)* %in, i32 addrspace(1)* %out) {
 entry:
   %val = load i32, i32 addrspace(5)* %in, align 4, !nontemporal !0
   store i32 %val, i32 addrspace(1)* %out
   ret void

 ;
 ; GFX11-WGP-LABEL: private_nontemporal_load_1:
 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_load_b32 s2, s[0:1], 0x0
 ; GFX11-WGP-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-WGP-NEXT:    s_load_b64 s[0:1], s[0:1], 0x8
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-WGP-NEXT:    scratch_load_b32 v0, v0, s2 slc
+; GFX11-WGP-NEXT:    scratch_load_b32 v0, v0, s2 slc dlc
 ; GFX11-WGP-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-WGP-NEXT:    global_store_b32 v1, v0, s[0:1]
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: private_nontemporal_load_1:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_load_b32 s2, s[0:1], 0x0
 ; GFX11-CU-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-CU-NEXT:    s_load_b64 s[0:1], s[0:1], 0x8
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, 0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
-; GFX11-CU-NEXT:    scratch_load_b32 v0, v0, s2 slc
+; GFX11-CU-NEXT:    scratch_load_b32 v0, v0, s2 slc dlc
 ; GFX11-CU-NEXT:    s_waitcnt vmcnt(0)
 ; GFX11-CU-NEXT:    global_store_b32 v1, v0, s[0:1]
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(5)* %in, i32 addrspace(1)* %out) {
 entry:
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %val.gep = getelementptr inbounds i32, i32 addrspace(5)* %in, i32 %tid
   %val = load i32, i32 addrspace(5)* %val.gep, align 4, !nontemporal !0

 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_clause 0x1
 ; GFX11-WGP-NEXT:    s_load_b64 s[2:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_load_b32 s0, s[0:1], 0x8
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    s_load_b32 s1, s[2:3], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s1
-; GFX11-WGP-NEXT:    scratch_store_b32 off, v0, s0 glc slc
+; GFX11-WGP-NEXT:    scratch_store_b32 off, v0, s0 glc slc dlc
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: private_nontemporal_store_0:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_clause 0x1
 ; GFX11-CU-NEXT:    s_load_b64 s[2:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_load_b32 s0, s[0:1], 0x8
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    s_load_b32 s1, s[2:3], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s1
-; GFX11-CU-NEXT:    scratch_store_b32 off, v0, s0 glc slc
+; GFX11-CU-NEXT:    scratch_store_b32 off, v0, s0 glc slc dlc
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(5)* %out) {
 entry:
   %val = load i32, i32 addrspace(1)* %in, align 4
   store i32 %val, i32 addrspace(5)* %out, !nontemporal !0
   ret void
 }
 

 ; GFX11-WGP-NEXT:    s_clause 0x1
 ; GFX11-WGP-NEXT:    s_load_b64 s[2:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_load_b32 s0, s[0:1], 0x8
 ; GFX11-WGP-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    s_load_b32 s1, s[2:3], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-WGP-NEXT:    scratch_store_b32 v0, v1, s0 glc slc
+; GFX11-WGP-NEXT:    scratch_store_b32 v0, v1, s0 glc slc dlc
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: private_nontemporal_store_1:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_clause 0x1
 ; GFX11-CU-NEXT:    s_load_b64 s[2:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_load_b32 s0, s[0:1], 0x8
 ; GFX11-CU-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    s_load_b32 s1, s[2:3], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-CU-NEXT:    scratch_store_b32 v0, v1, s0 glc slc
+; GFX11-CU-NEXT:    scratch_store_b32 v0, v1, s0 glc slc dlc
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(5)* %out) {
 entry:
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %val = load i32, i32 addrspace(1)* %in, align 4
   %out.gep = getelementptr inbounds i32, i32 addrspace(5)* %out, i32 %tid
   store i32 %val, i32 addrspace(5)* %out.gep, !nontemporal !0
   ret void
 }
 
 !0 = !{i32 1}
 declare i32 @llvm.amdgcn.workitem.id.x()

llvm/test/CodeGen/AMDGPU/memory-legalizer-private-volatile.ll

 ; GFX11-WGP:       ; %bb.0: ; %entry
 ; GFX11-WGP-NEXT:    s_clause 0x1
 ; GFX11-WGP-NEXT:    s_load_b64 s[2:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_load_b32 s0, s[0:1], 0x8
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    s_load_b32 s1, s[2:3], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v0, s1
-; GFX11-WGP-NEXT:    scratch_store_b32 off, v0, s0
+; GFX11-WGP-NEXT:    scratch_store_b32 off, v0, s0 dlc
 ; GFX11-WGP-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: private_volatile_store_0:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_clause 0x1
 ; GFX11-CU-NEXT:    s_load_b64 s[2:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_load_b32 s0, s[0:1], 0x8
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    s_load_b32 s1, s[2:3], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v0, s1
-; GFX11-CU-NEXT:    scratch_store_b32 off, v0, s0
+; GFX11-CU-NEXT:    scratch_store_b32 off, v0, s0 dlc
 ; GFX11-CU-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(5)* %out) {
 entry:
   %val = load i32, i32 addrspace(1)* %in, align 4
   store volatile i32 %val, i32 addrspace(5)* %out
   ret void
 }

 ; GFX11-WGP-NEXT:    s_clause 0x1
 ; GFX11-WGP-NEXT:    s_load_b64 s[2:3], s[0:1], 0x0
 ; GFX11-WGP-NEXT:    s_load_b32 s0, s[0:1], 0x8
 ; GFX11-WGP-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    s_load_b32 s1, s[2:3], 0x0
 ; GFX11-WGP-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-WGP-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-WGP-NEXT:    scratch_store_b32 v0, v1, s0
+; GFX11-WGP-NEXT:    scratch_store_b32 v0, v1, s0 dlc
 ; GFX11-WGP-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-WGP-NEXT:    s_endpgm
 ;
 ; GFX11-CU-LABEL: private_volatile_store_1:
 ; GFX11-CU:       ; %bb.0: ; %entry
 ; GFX11-CU-NEXT:    s_clause 0x1
 ; GFX11-CU-NEXT:    s_load_b64 s[2:3], s[0:1], 0x0
 ; GFX11-CU-NEXT:    s_load_b32 s0, s[0:1], 0x8
 ; GFX11-CU-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    s_load_b32 s1, s[2:3], 0x0
 ; GFX11-CU-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX11-CU-NEXT:    v_mov_b32_e32 v1, s1
-; GFX11-CU-NEXT:    scratch_store_b32 v0, v1, s0
+; GFX11-CU-NEXT:    scratch_store_b32 v0, v1, s0 dlc
 ; GFX11-CU-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX11-CU-NEXT:    s_endpgm
     i32 addrspace(1)* %in, i32 addrspace(5)* %out) {
 entry:
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %val = load i32, i32 addrspace(1)* %in, align 4
   %out.gep = getelementptr inbounds i32, i32 addrspace(5)* %out, i32 %tid
   store volatile i32 %val, i32 addrspace(5)* %out.gep
   ret void
 }
 
 declare i32 @llvm.amdgcn.workitem.id.x()