Add available_device_memory to fetch free amount of memory on a GPU (…

…#1567)

This PR adds `get_free_device_memory` that returns free GPU memory necessary for rapidsai/cudf#15628

Authors:
  - GALI PREM SAGAR (https://github.com/galipremsagar)

Approvers:
  - Mark Harris (https://github.com/harrism)
  - Bradley Dice (https://github.com/bdice)
  - Lawrence Mitchell (https://github.com/wence-)

URL: #1567

python/rmm/rmm/_lib/memory_resource.pxd

@@ -1,4 +1,4 @@
-# Copyright (c) 2020-2022, NVIDIA CORPORATION.
+# Copyright (c) 2020-2024, NVIDIA CORPORATION.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -14,6 +14,7 @@
 
 from libc.stdint cimport int8_t
 from libcpp.memory cimport shared_ptr
+from libcpp.pair cimport pair
 from libcpp.string cimport string
 from libcpp.vector cimport vector
 
@@ -32,6 +33,10 @@ cdef extern from "rmm/mr/device/device_memory_resource.hpp" \
             cuda_stream_view stream
         ) except +
 
+cdef extern from "rmm/cuda_device.hpp" namespace "rmm" nogil:
+    size_t percent_of_free_device_memory(int percent) except +
+    pair[size_t, size_t] available_device_memory() except +
+
 cdef class DeviceMemoryResource:
     cdef shared_ptr[device_memory_resource] c_obj
     cdef device_memory_resource* get_mr(self) noexcept nogil

python/rmm/rmm/_lib/memory_resource.pyx

@@ -1,4 +1,4 @@
-# Copyright (c) 2020-2022, NVIDIA CORPORATION.
+# Copyright (c) 2020-2024, NVIDIA CORPORATION.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -32,9 +32,16 @@ from libcpp.string cimport string
 from cuda.cudart import cudaError_t
 
 from rmm._cuda.gpu import CUDARuntimeError, getDevice, setDevice
+
 from rmm._cuda.stream cimport Stream
+
 from rmm._cuda.stream import DEFAULT_STREAM
+
 from rmm._lib.cuda_stream_view cimport cuda_stream_view
+from rmm._lib.memory_resource cimport (
+    available_device_memory as c_available_device_memory,
+    percent_of_free_device_memory as c_percent_of_free_device_memory,
+)
 from rmm._lib.per_device_resource cimport (
     cuda_device_id,
     set_per_device_resource as cpp_set_per_device_resource,
@@ -109,8 +116,6 @@ cdef extern from "rmm/mr/device/cuda_async_memory_resource.hpp" \
         win32
         win32_kmt
 
-cdef extern from "rmm/cuda_device.hpp" namespace "rmm" nogil:
-    size_t percent_of_free_device_memory(int percent) except +
 
 cdef extern from "rmm/mr/device/pool_memory_resource.hpp" \
         namespace "rmm::mr" nogil:
@@ -368,7 +373,7 @@ cdef class PoolMemoryResource(UpstreamResourceAdaptor):
         cdef size_t c_initial_pool_size
         cdef optional[size_t] c_maximum_pool_size
         c_initial_pool_size = (
-            percent_of_free_device_memory(50) if
+            c_percent_of_free_device_memory(50) if
             initial_pool_size is None
             else initial_pool_size
         )
@@ -1188,3 +1193,12 @@ def get_log_filenames():
         else None
         for i, each_mr in _per_device_mrs.items()
     }
+
+
+def available_device_memory():
+    """
+    Returns a tuple of free and total device memory memory.
+    """
+    cdef pair[size_t, size_t] res
+    res = c_available_device_memory()
+    return (res.first, res.second)

python/rmm/rmm/mr.py

@@ -1,4 +1,4 @@
-# Copyright (c) 2020-2021, NVIDIA CORPORATION.
+# Copyright (c) 2020-2024, NVIDIA CORPORATION.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -28,6 +28,7 @@
     UpstreamResourceAdaptor,
     _flush_logs,
     _initialize,
+    available_device_memory,
     disable_logging,
     enable_logging,
     get_current_device_resource,
@@ -57,6 +58,7 @@
     "UpstreamResourceAdaptor",
     "_flush_logs",
     "_initialize",
+    "available_device_memory",
     "set_per_device_resource",
     "enable_logging",
     "disable_logging",

python/rmm/rmm/tests/test_rmm.py

@@ -1002,3 +1002,16 @@ def test_invalid_logging_level(level):
         rmm.set_flush_level(level)
     with pytest.raises(TypeError):
         rmm.should_log(level)
+
+
+def test_available_device_memory():
+    from rmm.mr import available_device_memory
+
+    initial_memory = available_device_memory()
+    device_buffer = rmm.DeviceBuffer.to_device(  # noqa: F841
+        np.zeros(10000, dtype="u1")
+    )
+    final_memory = available_device_memory()
+    assert initial_memory[1] == final_memory[1]
+    assert initial_memory[0] > 0
+    assert final_memory[0] > 0