CUDA Python Reference

CUDA Built-in Target deprecation notice

The CUDA target built-in to Numba is deprecated, with further development moved to the NVIDIA numba-cuda package. Please see Built-in CUDA target deprecation and maintenance status.

• CUDA Host API
  • Device Management
    • Device detection and enquiry
      • is_available()
      • detect()
    • Context management
      • Context
        • Context.get_memory_info()
        • Context.pop()
        • Context.push()
        • Context.reset()
      • current_context()
      • require_context()
      • synchronize()
      • close()
    • Device management
      • numba.cuda.gpus
      • numba.cuda.gpus.current
      • _DeviceContextManager
      • select_device()
      • get_current_device()
      • list_devices()
      • numba.cuda.cudadrv.driver.Device
        • numba.cuda.cudadrv.driver.Device.compute_capability
        • numba.cuda.cudadrv.driver.Device.id
        • numba.cuda.cudadrv.driver.Device.name
        • numba.cuda.cudadrv.driver.Device.uuid
        • numba.cuda.cudadrv.driver.Device.reset()
        • numba.cuda.cudadrv.driver.Device.supports_float16
  • Compilation
    • compile()
    • compile_for_current_device()
    • compile_ptx()
    • compile_ptx_for_current_device()
  • Measurement
    • Profiling
      • profile_start()
      • profile_stop()
      • profiling()
    • Events
      • event()
      • event_elapsed_time()
      • Event
        • Event.query()
        • Event.record()
        • Event.synchronize()
        • Event.wait()
  • Stream Management
    • Stream
      • Stream.add_callback()
      • Stream.async_done()
      • Stream.auto_synchronize()
      • Stream.synchronize()
    • stream()
    • default_stream()
    • legacy_default_stream()
    • per_thread_default_stream()
    • external_stream()
  • Runtime
    • Runtime
      • Runtime.get_version()
      • Runtime.is_supported_version()
      • Runtime.supported_versions
    • is_supported_version()
• CUDA Kernel API
  • Kernel declaration
    • jit()
  • Dispatcher objects
    • CUDADispatcher
      • CUDADispatcher.extensions
      • CUDADispatcher.forall()
      • CUDADispatcher.get_const_mem_size()
      • CUDADispatcher.get_local_mem_per_thread()
      • CUDADispatcher.get_max_threads_per_block()
      • CUDADispatcher.get_regs_per_thread()
      • CUDADispatcher.get_shared_mem_per_block()
      • CUDADispatcher.inspect_asm()
      • CUDADispatcher.inspect_llvm()
      • CUDADispatcher.inspect_sass()
      • CUDADispatcher.inspect_types()
      • CUDADispatcher.specialize()
      • CUDADispatcher.specialized
  • Intrinsic Attributes and Functions
    • Thread Indexing
      • numba.cuda.threadIdx
      • numba.cuda.blockIdx
      • numba.cuda.blockDim
      • numba.cuda.gridDim
      • numba.cuda.laneid
      • numba.cuda.warpsize
      • numba.cuda.grid()
      • numba.cuda.gridsize()
    • Memory Management
      • numba.cuda.shared.array()
      • numba.cuda.local.array()
      • numba.cuda.const.array_like()
    • Synchronization and Atomic Operations
      • numba.cuda.atomic.add()
      • numba.cuda.atomic.sub()
      • numba.cuda.atomic.and_()
      • numba.cuda.atomic.or_()
      • numba.cuda.atomic.xor()
      • numba.cuda.atomic.exch()
      • numba.cuda.atomic.inc()
      • numba.cuda.atomic.dec()
      • numba.cuda.atomic.max()
      • numba.cuda.atomic.cas()
      • numba.cuda.syncthreads()
      • numba.cuda.syncthreads_count()
      • numba.cuda.syncthreads_and()
      • numba.cuda.syncthreads_or()
    • Cooperative Groups
      • numba.cuda.cg.this_grid()
      • numba.cuda.cg.GridGroup
        • numba.cuda.cg.GridGroup.sync()
    • Memory Fences
      • numba.cuda.threadfence()
      • numba.cuda.threadfence_block()
      • numba.cuda.threadfence_system()
    • Warp Intrinsics
      • numba.cuda.syncwarp()
      • numba.cuda.all_sync()
      • numba.cuda.any_sync()
      • numba.cuda.eq_sync()
      • numba.cuda.ballot_sync()
      • numba.cuda.shfl_sync()
      • numba.cuda.shfl_up_sync()
      • numba.cuda.shfl_down_sync()
      • numba.cuda.shfl_xor_sync()
      • numba.cuda.match_any_sync()
      • numba.cuda.match_all_sync()
      • numba.cuda.activemask()
      • numba.cuda.lanemask_lt()
    • Integer Intrinsics
      • numba.cuda.popc()
      • numba.cuda.brev()
      • numba.cuda.clz()
      • numba.cuda.ffs()
    • Floating Point Intrinsics
      • numba.cuda.fma()
      • numba.cuda.cbrt()
    • 16-bit Floating Point Intrinsics
      • numba.cuda.is_float16_supported()
      • numba.cuda.fp16.hfma()
      • numba.cuda.fp16.hadd()
      • numba.cuda.fp16.hsub()
      • numba.cuda.fp16.hmul()
      • numba.cuda.fp16.hdiv()
      • numba.cuda.fp16.hneg()
      • numba.cuda.fp16.habs()
      • numba.cuda.fp16.hsin()
      • numba.cuda.fp16.hcos()
      • numba.cuda.fp16.hlog()
      • numba.cuda.fp16.hlog10()
      • numba.cuda.fp16.hlog2()
      • numba.cuda.fp16.hexp()
      • numba.cuda.fp16.hexp10()
      • numba.cuda.fp16.hexp2()
      • numba.cuda.fp16.hfloor()
      • numba.cuda.fp16.hceil()
      • numba.cuda.fp16.hsqrt()
      • numba.cuda.fp16.hrsqrt()
      • numba.cuda.fp16.hrcp()
      • numba.cuda.fp16.hrint()
      • numba.cuda.fp16.htrunc()
      • numba.cuda.fp16.heq()
      • numba.cuda.fp16.hne()
      • numba.cuda.fp16.hgt()
      • numba.cuda.fp16.hge()
      • numba.cuda.fp16.hlt()
      • numba.cuda.fp16.hle()
      • numba.cuda.fp16.hmax()
      • numba.cuda.fp16.hmin()
    • Control Flow Instructions
      • numba.cuda.selp()
    • Timer Intrinsics
      • numba.cuda.nanosleep()
• CUDA-Specific Types
  • Vector Types
• Memory Management
  • to_device()
  • device_array()
  • device_array_like()
  • pinned_array()
  • pinned_array_like()
  • mapped_array()
  • mapped_array_like()
  • managed_array()
  • pinned()
  • mapped()
  • Device Objects
    • DeviceNDArray
      • DeviceNDArray.copy_to_device()
      • DeviceNDArray.copy_to_host()
      • DeviceNDArray.is_c_contiguous()
      • DeviceNDArray.is_f_contiguous()
      • DeviceNDArray.ravel()
      • DeviceNDArray.reshape()
      • DeviceNDArray.split()
    • DeviceRecord
      • DeviceRecord.copy_to_device()
      • DeviceRecord.copy_to_host()
    • MappedNDArray
      • MappedNDArray.copy_to_device()
      • MappedNDArray.copy_to_host()
      • MappedNDArray.split()
• Libdevice functions
  • Wrapped functions
    • abs()
    • acos()
    • acosf()
    • acosh()
    • acoshf()
    • asin()
    • asinf()
    • asinh()
    • asinhf()
    • atan()
    • atan2()
    • atan2f()
    • atanf()
    • atanh()
    • atanhf()
    • brev()
    • brevll()
    • byte_perm()
    • cbrt()
    • cbrtf()
    • ceil()
    • ceilf()
    • clz()
    • clzll()
    • copysign()
    • copysignf()
    • cos()
    • cosf()
    • cosh()
    • coshf()
    • cospi()
    • cospif()
    • dadd_rd()
    • dadd_rn()
    • dadd_ru()
    • dadd_rz()
    • ddiv_rd()
    • ddiv_rn()
    • ddiv_ru()
    • ddiv_rz()
    • dmul_rd()
    • dmul_rn()
    • dmul_ru()
    • dmul_rz()
    • double2float_rd()
    • double2float_rn()
    • double2float_ru()
    • double2float_rz()
    • double2hiint()
    • double2int_rd()
    • double2int_rn()
    • double2int_ru()
    • double2int_rz()
    • double2ll_rd()
    • double2ll_rn()
    • double2ll_ru()
    • double2ll_rz()
    • double2loint()
    • double2uint_rd()
    • double2uint_rn()
    • double2uint_ru()
    • double2uint_rz()
    • double2ull_rd()
    • double2ull_rn()
    • double2ull_ru()
    • double2ull_rz()
    • double_as_longlong()
    • drcp_rd()
    • drcp_rn()
    • drcp_ru()
    • drcp_rz()
    • dsqrt_rd()
    • dsqrt_rn()
    • dsqrt_ru()
    • dsqrt_rz()
    • erf()
    • erfc()
    • erfcf()
    • erfcinv()
    • erfcinvf()
    • erfcx()
    • erfcxf()
    • erff()
    • erfinv()
    • erfinvf()
    • exp()
    • exp10()
    • exp10f()
    • exp2()
    • exp2f()
    • expf()
    • expm1()
    • expm1f()
    • fabs()
    • fabsf()
    • fadd_rd()
    • fadd_rn()
    • fadd_ru()
    • fadd_rz()
    • fast_cosf()
    • fast_exp10f()
    • fast_expf()
    • fast_fdividef()
    • fast_log10f()
    • fast_log2f()
    • fast_logf()
    • fast_powf()
    • fast_sincosf()
    • fast_sinf()
    • fast_tanf()
    • fdim()
    • fdimf()
    • fdiv_rd()
    • fdiv_rn()
    • fdiv_ru()
    • fdiv_rz()
    • ffs()
    • ffsll()
    • finitef()
    • float2half_rn()
    • float2int_rd()
    • float2int_rn()
    • float2int_ru()
    • float2int_rz()
    • float2ll_rd()
    • float2ll_rn()
    • float2ll_ru()
    • float2ll_rz()
    • float2uint_rd()
    • float2uint_rn()
    • float2uint_ru()
    • float2uint_rz()
    • float2ull_rd()
    • float2ull_rn()
    • float2ull_ru()
    • float2ull_rz()
    • float_as_int()
    • floor()
    • floorf()
    • fma()
    • fma_rd()
    • fma_rn()
    • fma_ru()
    • fma_rz()
    • fmaf()
    • fmaf_rd()
    • fmaf_rn()
    • fmaf_ru()
    • fmaf_rz()
    • fmax()
    • fmaxf()
    • fmin()
    • fminf()
    • fmod()
    • fmodf()
    • fmul_rd()
    • fmul_rn()
    • fmul_ru()
    • fmul_rz()
    • frcp_rd()
    • frcp_rn()
    • frcp_ru()
    • frcp_rz()
    • frexp()
    • frexpf()
    • frsqrt_rn()
    • fsqrt_rd()
    • fsqrt_rn()
    • fsqrt_ru()
    • fsqrt_rz()
    • fsub_rd()
    • fsub_rn()
    • fsub_ru()
    • fsub_rz()
    • hadd()
    • half2float()
    • hiloint2double()
    • hypot()
    • hypotf()
    • ilogb()
    • ilogbf()
    • int2double_rn()
    • int2float_rd()
    • int2float_rn()
    • int2float_ru()
    • int2float_rz()
    • int_as_float()
    • isfinited()
    • isinfd()
    • isinff()
    • isnand()
    • isnanf()
    • j0()
    • j0f()
    • j1()
    • j1f()
    • jn()
    • jnf()
    • ldexp()
    • ldexpf()
    • lgamma()
    • lgammaf()
    • ll2double_rd()
    • ll2double_rn()
    • ll2double_ru()
    • ll2double_rz()
    • ll2float_rd()
    • ll2float_rn()
    • ll2float_ru()
    • ll2float_rz()
    • llabs()
    • llmax()
    • llmin()
    • llrint()
    • llrintf()
    • llround()
    • llroundf()
    • log()
    • log10()
    • log10f()
    • log1p()
    • log1pf()
    • log2()
    • log2f()
    • logb()
    • logbf()
    • logf()
    • longlong_as_double()
    • max()
    • min()
    • modf()
    • modff()
    • mul24()
    • mul64hi()
    • mulhi()
    • nearbyint()
    • nearbyintf()
    • nextafter()
    • nextafterf()
    • normcdf()
    • normcdff()
    • normcdfinv()
    • normcdfinvf()
    • popc()
    • popcll()
    • pow()
    • powf()
    • powi()
    • powif()
    • rcbrt()
    • rcbrtf()
    • remainder()
    • remainderf()
    • remquo()
    • remquof()
    • rhadd()
    • rint()
    • rintf()
    • round()
    • roundf()
    • rsqrt()
    • rsqrtf()
    • sad()
    • saturatef()
    • scalbn()
    • scalbnf()
    • signbitd()
    • signbitf()
    • sin()
    • sincos()
    • sincosf()
    • sincospi()
    • sincospif()
    • sinf()
    • sinh()
    • sinhf()
    • sinpi()
    • sinpif()
    • sqrt()
    • sqrtf()
    • tan()
    • tanf()
    • tanh()
    • tanhf()
    • tgamma()
    • tgammaf()
    • trunc()
    • truncf()
    • uhadd()
    • uint2double_rn()
    • uint2float_rd()
    • uint2float_rn()
    • uint2float_ru()
    • uint2float_rz()
    • ull2double_rd()
    • ull2double_rn()
    • ull2double_ru()
    • ull2double_rz()
    • ull2float_rd()
    • ull2float_rn()
    • ull2float_ru()
    • ull2float_rz()
    • ullmax()
    • ullmin()
    • umax()
    • umin()
    • umul24()
    • umul64hi()
    • umulhi()
    • urhadd()
    • usad()
    • y0()
    • y0f()
    • y1()
    • y1f()
    • yn()
    • ynf()