include/ck_tile/core/arch/arch.hpp Source File

include/ck_tile/core/arch/arch.hpp Source File#

Composable Kernel: include/ck_tile/core/arch/arch.hpp Source File
 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2025, Advanced Micro Devices, Inc. All rights reserved.
  
 #pragma once
  
 // Address Space for AMDGCN
 // https://llvm.org/docs/AMDGPUUsage.html#address-space
  
 #include "ck_tile/core/config.hpp"
 #include "ck_tile/core/numeric/integer.hpp"
 #include "ck_tile/core/numeric/integral_constant.hpp"
  
 #define CK_TILE_S_CNT_MAX 0b1100'1111'0111'1111
 #define CK_TILE_VMCNT(cnt)                                              \
     ([]() { static_assert(!((cnt) >> 6), "VMCNT only has 6 bits"); }(), \
      ((cnt) & 0b1111) | (((cnt) & 0b110000) << 10))
 #define CK_TILE_EXPCNT(cnt) \
     ([]() { static_assert(!((cnt) >> 3), "EXP only has 3 bits"); }(), ((cnt) << 4))
 #define CK_TILE_LGKMCNT(cnt) \
     ([]() { static_assert(!((cnt) >> 4), "LGKM only has 4 bits"); }(), ((cnt) << 8))
  
 namespace ck_tile {
  
 template <typename, bool>
 struct safe_underlying_type;
  
 template <typename T>
 struct safe_underlying_type<T, true>
 {
     using type = std::underlying_type_t<T>;
 };
  
 template <typename T>
 struct safe_underlying_type<T, false>
 {
     using type = void;
 };
  
 template <typename T>
 using safe_underlying_type_t = typename safe_underlying_type<T, std::is_enum<T>::value>::type;
  
 enum struct address_space_enum : std::uint16_t
 {
     generic = 0,
     global,
     lds,
     sgpr,
     constant,
     vgpr
 };
  
 enum struct memory_operation_enum : std::uint16_t
 {
     set = 0,
     atomic_add,
     atomic_max,
     add
 };
  
 CK_TILE_HOST_DEVICE constexpr index_t get_warp_size()
 {
 #if defined(__GFX9__) || !defined(__HIP_DEVICE_COMPILE__)
     return 64;
 #else
     return 32;
 #endif
 }
  
 CK_TILE_DEVICE index_t get_grid_size() { return gridDim.x; }
  
 CK_TILE_DEVICE index_t get_block_size() { return blockDim.x; }
  
 // TODO: deprecate these
 CK_TILE_DEVICE index_t get_thread_local_1d_id() { return threadIdx.x; }
  
 CK_TILE_DEVICE index_t get_thread_global_1d_id() { return blockIdx.x * blockDim.x + threadIdx.x; }
  
 CK_TILE_DEVICE index_t get_block_1d_id() { return blockIdx.x; }
  
 // Use these instead
 CK_TILE_DEVICE index_t get_lane_id() { return __lane_id(); }
  
 CK_TILE_DEVICE index_t get_warp_id()
 {
     return __builtin_amdgcn_readfirstlane(threadIdx.x / get_warp_size());
 }
  
 CK_TILE_DEVICE index_t get_thread_id() { return threadIdx.x; }
  
 CK_TILE_DEVICE index_t get_block_id() { return blockIdx.x; }
  
 CK_TILE_DEVICE void block_sync_lds()
 {
 #if CK_TILE_EXPERIMENTAL_BLOCK_SYNC_LDS_WITHOUT_SYNC_VMEM
     // asm volatile("\
     // s_waitcnt lgkmcnt(0) \n \
     // s_barrier \
     // " ::);
  
     __builtin_amdgcn_s_waitcnt(0xc07f);
     __builtin_amdgcn_s_barrier();
 #else
     __syncthreads();
 #endif
 }
  
 CK_TILE_DEVICE void block_sync_load_raw(index_t cnt = 0)
 {
 #ifdef __gfx12__
     asm volatile("s_wait_loadcnt %0 \n"
                  "s_barrier_signal -1 \n"
                  "s_barrier_wait -1"
                  :
                  : "n"(cnt)
                  : "memory");
 #else
     asm volatile("s_waitcnt vmcnt(%0) \n"
                  "s_barrier"
                  :
                  : "n"(cnt)
                  : "memory");
 #endif
 }
  
 // https://llvm.org/docs/AMDGPU/gfx9_waitcnt.html
 struct waitcnt_arg
 {
     // bit numbers (hex) -------------------------> FE'DC'BA98'7'654'3210
     // [V]M [E]XP [L]GKM counters and [U]NUSED ---> VV'UU'LLLL'U'EEE'VVVV
     CK_TILE_DEVICE static constexpr index_t MAX = 0b11'00'1111'0'111'1111;
  
     CK_TILE_DEVICE static constexpr index_t kMaxVmCnt   = 0b111111;
     CK_TILE_DEVICE static constexpr index_t kMaxExpCnt  = 0b111;
     CK_TILE_DEVICE static constexpr index_t kMaxLgkmCnt = 0b1111;
  
     template <index_t cnt>
     CK_TILE_DEVICE static constexpr index_t from_vmcnt()
     {
         static_assert(cnt >= 0 && !(cnt >> 6), "valid range is [0..63]");
         return MAX & ((cnt & 0b1111) | ((cnt & 0b110000) << 10));
     }
  
     template <index_t cnt>
     CK_TILE_DEVICE static constexpr index_t from_expcnt()
     {
         static_assert(cnt >= 0 && !(cnt >> 3), "valid range is [0..7]");
         return MAX & (cnt << 4);
     }
  
     template <index_t cnt>
     CK_TILE_DEVICE static constexpr index_t from_lgkmcnt()
     {
         static_assert(cnt >= 0 && !(cnt >> 4), "valid range is [0..15]");
         return MAX & (cnt << 8);
     }
 };
  
 template <index_t vmcnt   = waitcnt_arg::kMaxVmCnt,
           index_t expcnt  = waitcnt_arg::kMaxExpCnt,
           index_t lgkmcnt = waitcnt_arg::kMaxLgkmCnt>
 CK_TILE_DEVICE void s_waitcnt()
 {
     __builtin_amdgcn_s_waitcnt(waitcnt_arg::from_vmcnt<vmcnt>() |
                                waitcnt_arg::from_expcnt<expcnt>() |
                                waitcnt_arg::from_lgkmcnt<lgkmcnt>());
 }
  
 template <index_t vmcnt   = waitcnt_arg::kMaxVmCnt,
           index_t expcnt  = waitcnt_arg::kMaxExpCnt,
           index_t lgkmcnt = waitcnt_arg::kMaxLgkmCnt>
 CK_TILE_DEVICE void s_waitcnt_barrier()
 {
     s_waitcnt<vmcnt, expcnt, lgkmcnt>();
     __builtin_amdgcn_s_barrier();
 }
  
 CK_TILE_DEVICE void block_sync_lds_direct_load()
 {
 #if 1
     // invoke clang builtins which *should* produce the same result as the inline asm below
     // difference: inline asm is being compiled to wait vmcnt(0) after the barrier
     s_waitcnt_barrier<0, waitcnt_arg::kMaxExpCnt, 0>();
 #else
     // same content as in old CK (#999)
     asm volatile("\
     s_waitcnt vmcnt(0) \n \
     s_waitcnt lgkmcnt(0) \n \
     s_barrier \
     " ::);
 #endif
 }
  
 CK_TILE_DEVICE void s_nop(index_t cnt = 0)
 {
 #if 1
     asm volatile("s_nop %0" : : "n"(cnt) :);
 #else
     __builtin_amdgcn_sched_barrier(cnt);
 #endif
 }
  
 #define CK_CONSTANT_ADDRESS_SPACE \
     __attribute__((address_space( \
         static_cast<safe_underlying_type_t<address_space_enum>>(address_space_enum::constant))))
  
 template <typename T>
 __device__ T* cast_pointer_to_generic_address_space(T CK_CONSTANT_ADDRESS_SPACE* p)
 {
     // cast a pointer in "Constant" address space (4) to "Generic" address space (0)
     // only c-style pointer cast seems be able to be compiled
 #pragma clang diagnostic push
 #pragma clang diagnostic ignored "-Wold-style-cast"
     return (T*)(p); // NOLINT(old-style-cast)
 #pragma clang diagnostic pop
 }
  
 template <typename T>
 __host__ __device__ T CK_CONSTANT_ADDRESS_SPACE* cast_pointer_to_constant_address_space(T* p)
 {
     // cast a pointer in "Generic" address space (0) to "Constant" address space (4)
     // only c-style pointer cast seems be able to be compiled;
 #pragma clang diagnostic push
 #pragma clang diagnostic ignored "-Wold-style-cast"
     return (T CK_CONSTANT_ADDRESS_SPACE*)p; // NOLINT(old-style-cast)
 #pragma clang diagnostic pop
 }
  
 CK_TILE_HOST_DEVICE constexpr index_t get_smem_capacity()
 {
 #if defined(__gfx950__)
     return 163840;
 #else
     return 65536;
 #endif
 }
  
 } // namespace ck_tile