atomic_ref atomic_ref compare_exchange_strong compare_exchange_weak difference_type exchange fetch_add fetch_and fetch_dec_mod fetch_inc_mod fetch_max fetch_min fetch_or fetch_sub fetch_xor is_lock_free load load operator T operator&= operator++ operator++ operator+= operator-- operator-- operator-= operator= operator^= operator|= store store value_type

fetch_max() template<typename T > template<STDGPU_DETAIL_OVERLOAD_IF(std::is_integral_v< T >||std::is_floating_point_v< T >) > STDGPU_DEVICE_ONLY T stdgpu::atomic_ref< T >::fetch_max ( const T  arg, const memory_order  order = memory_order_seq_cst  ) noexcept Atomically computes and stores the maximum of the stored value and the given argument. Parameters [in] arg The other argument of maximum [in] order The memory order Returns The old value