import triton
import triton.language as tl

from torch._inductor.runtime import triton_helpers, triton_heuristics
from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math
from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, DeviceProperties
triton_helpers.set_driver_to_gpu()

@triton_heuristics.persistent_reduction(
    size_hints={'x': 8, 'r0_': 1024},
    reduction_hint=ReductionHint.INNER,
    filename=__file__,
    triton_meta={'signature': {'in_ptr0': '*bf16', 'in_ptr1': '*bf16', 'in_ptr2': '*bf16', 'out_ptr1': '*bf16', 'xnumel': 'i32', 'r0_numel': 'i32', 'XBLOCK': 'constexpr'}, 'device': DeviceProperties(type='cuda', index=0, multi_processor_count=48, cc=121, major=12, regs_per_multiprocessor=65536, max_threads_per_multi_processor=1536, max_threads_per_block=1024, warp_size=32), 'constants': {}, 'native_matmul': False, 'enable_fp_fusion': True, 'launch_pdl': False, 'disable_ftz': False, 'configs': [{(0,): [['tt.divisibility', 16]], (1,): [['tt.divisibility', 16]], (2,): [['tt.divisibility', 16]], (3,): [['tt.divisibility', 16]], (5,): [['tt.divisibility', 16]]}]},
    inductor_meta={'grid_type': 'Grid1D', 'autotune_hints': set(), 'kernel_name': 'triton_per_fused__to_copy__unsafe_view_add_mean_mul_pow_rsqrt_8', 'mutated_arg_names': [], 'optimize_mem': True, 'no_x_dim': None, 'atomic_add_found': False, 'num_load': 3, 'num_store': 1, 'num_reduction': 1, 'backend_hash': 'BE9F3F68E84A48F2C239366BB10BBF3F0E1498DBECF7F747524E7B7961E82579', 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': False, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False, 'deterministic': False, 'force_filter_reduction_configs': False, 'mix_order_reduction_allow_multi_stages': False, 'are_deterministic_algorithms_enabled': False, 'tiling_scores': {'x': 0, 'r0_': 43008}}
)
@triton.jit
def triton_per_fused__to_copy__unsafe_view_add_mean_mul_pow_rsqrt_8(in_ptr0, in_ptr1, in_ptr2, out_ptr1, xnumel, r0_numel, XBLOCK : tl.constexpr):
    xnumel = 5
    r0_numel = 1024
    R0_BLOCK: tl.constexpr = 1024
    rnumel = r0_numel
    RBLOCK: tl.constexpr = R0_BLOCK
    xoffset = tl.program_id(0) * XBLOCK
    xindex = xoffset + tl.arange(0, XBLOCK)[:, None]
    xmask = xindex < xnumel
    r0_index = tl.arange(0, R0_BLOCK)[None, :]
    r0_offset = 0
    r0_mask = tl.full([R0_BLOCK], True, tl.int1)[None, :]
    roffset = r0_offset
    rindex = r0_index
    r0_1 = r0_index
    x0 = xindex
    tmp0 = tl.load(in_ptr0 + (r0_1 + 1024*x0), xmask, other=0.0).to(tl.float32)
    tmp1 = tl.load(in_ptr1 + (r0_1 + 1024*x0), xmask, other=0.0).to(tl.float32)
    tmp16 = tl.load(in_ptr2 + (r0_1), None, eviction_policy='evict_last').to(tl.float32)
    tmp2 = tmp0 + tmp1
    tmp3 = tmp2.to(tl.float32)
    tmp4 = tmp3 * tmp3
    tmp5 = tl.broadcast_to(tmp4, [XBLOCK, R0_BLOCK])
    tmp7 = tl.where(xmask, tmp5, 0)
    tmp8 = tl.sum(tmp7, 1)[:, None].to(tl.float32)
    tmp9 = tl.full([1, 1], 1024.0, tl.float32)
    tmp10 = (tmp8 / tmp9)
    tmp11 = tl.full([1, 1], 1e-05, tl.float32)
    tmp12 = tmp10 + tmp11
    tmp13 = libdevice.rsqrt(tmp12)
    tmp14 = tmp3 * tmp13
    tmp15 = tmp14.to(tl.float32)
    tmp17 = tmp15 * tmp16
    tl.store(out_ptr1 + (r0_1 + 1024*x0), tmp17, xmask)