examples: Add application CUDA memory pool example

An example to show application managed CUDA memory pool usage

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/7427>

subprojects/gst-plugins-bad/tests/examples/cuda/cudamemorypool.c

@@ -0,0 +1,216 @@
+/*
+ * GStreamer
+ * Copyright (C) 2024 Seungha Yang <seungha@centricular.com>
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Library General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Library General Public License for more details.
+ *
+ * You should have received a copy of the GNU Library General Public
+ * License along with this library; if not, write to the
+ * Free Software Foundation, Inc., 51 Franklin St, Fifth Floor,
+ * Boston, MA 02110-1301, USA.
+ */
+
+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
+
+#include <gst/gst.h>
+#include <gst/cuda/gstcuda.h>
+#include <gst/video/video.h>
+
+typedef struct
+{
+  GstCudaMemoryPool *mem_pool;
+} NeedPoolCbData;
+
+static GstCudaMemoryPool *
+on_need_pool (GstCudaAllocator * allocator, GstCudaContext * context,
+    NeedPoolCbData * data)
+{
+  g_assert (data->mem_pool);
+
+  gst_println ("Need pool callback");
+
+  return gst_cuda_memory_pool_ref (data->mem_pool);
+}
+
+gint
+main (gint argc, gchar ** argv)
+{
+  GstCudaContext *context;
+  GstCudaStream *stream;
+  GstBufferPool *pool;
+  GstStructure *config;
+  gboolean stream_ordered_alloc = FALSE;
+  CUmemoryPool handle;
+  CUresult ret;
+  guint64 threshold;
+  guint64 reserved = 0;
+  guint64 current = 0;
+  GstBuffer *buffer;
+  GstCaps *caps;
+  GstVideoInfo info;
+  GstMapInfo map_info;
+  guint i;
+  NeedPoolCbData data;
+
+  gst_init (NULL, NULL);
+
+  if (!gst_cuda_load_library ()) {
+    gst_println ("Couldn't load cuda library");
+    return 0;
+  }
+
+  /* Install need-pool callback to provide application's CUDA memory pool.
+   * This callback will be called when GstCudaBufferPool is configured */
+  gst_cuda_register_allocator_need_pool_callback (
+      (GstCudaMemoryAllocatorNeedPoolCallback) on_need_pool, &data, NULL);
+
+  context = gst_cuda_context_new (0);
+  if (!context) {
+    gst_println ("Couldn't create cuda context");
+    return 0;
+  }
+
+  g_object_get (context, "stream-ordered-alloc", &stream_ordered_alloc, NULL);
+  if (!stream_ordered_alloc) {
+    gst_println ("Stream ordered allocation is not supported");
+    gst_object_unref (context);
+    return 0;
+  }
+
+  stream = gst_cuda_stream_new (context);
+  if (!stream) {
+    gst_printerrln ("Couldn't create cuda stream");
+    return 0;
+  }
+
+  /* Since default prop is enough in this example case, pass NULL prop */
+  data.mem_pool = gst_cuda_memory_pool_new (context, NULL);
+  if (!data.mem_pool) {
+    gst_printerrln ("Couldn't create memory pool");
+    return 0;
+  }
+
+  handle = gst_cuda_memory_pool_get_handle (data.mem_pool);
+
+  gst_cuda_context_push (context);
+  /* Configure pool attributes. In this example, release threshold will be
+   * increased (default is zero) so that allocated memory can be retained */
+  threshold = 1024 * 1024 * 20;
+
+  ret = CuMemPoolSetAttribute (handle, CU_MEMPOOL_ATTR_RELEASE_THRESHOLD,
+      (void *) &threshold);
+  if (!gst_cuda_result (ret)) {
+    gst_printerrln ("Couldn't increase release threshold");
+    return 0;
+  }
+
+  ret = CuMemPoolGetAttribute (handle, CU_MEMPOOL_ATTR_RESERVED_MEM_CURRENT,
+      (void *) &reserved);
+  if (!gst_cuda_result (ret)) {
+    gst_printerrln ("Couldn't get reserved size");
+    return 0;
+  }
+
+  ret = CuMemPoolGetAttribute (handle, CU_MEMPOOL_ATTR_USED_MEM_CURRENT,
+      (void *) &current);
+  if (!gst_cuda_result (ret)) {
+    gst_printerrln ("Couldn't get current size");
+    return 0;
+  }
+  gst_println ("Initial pool configuration, release threshold: %"
+      G_GUINT64_FORMAT ", reserved: %" G_GUINT64_FORMAT ", current: %"
+      G_GUINT64_FORMAT, threshold, reserved, current);
+  gst_cuda_context_pop (NULL);
+
+  gst_video_info_set_format (&info, GST_VIDEO_FORMAT_RGBA, 640, 480);
+  caps = gst_video_info_to_caps (&info);
+
+  for (i = 0; i < 2; i++) {
+    /* Creates cuda buffer pool */
+    pool = gst_cuda_buffer_pool_new (context);
+    if (!pool) {
+      gst_printerrln ("Couldn't create buffer pool");
+      return 0;
+    }
+
+    config = gst_buffer_pool_get_config (pool);
+    gst_buffer_pool_config_set_params (config, caps, info.size, 0, 0);
+
+    /* Sets CUDA specific buffer pool options. For stream ordered allocation,
+     * CUDA stream object must be configured in the config */
+    gst_buffer_pool_config_set_cuda_stream (config, stream);
+    /* NOTE: stream ordered allocation is enabled by default */
+    gst_buffer_pool_config_set_cuda_stream_ordered_alloc (config, TRUE);
+    gst_buffer_pool_set_config (pool, config);
+
+    gst_buffer_pool_set_active (pool, TRUE);
+
+    gst_buffer_pool_acquire_buffer (pool, &buffer, NULL);
+
+    gst_cuda_context_push (context);
+    ret = CuMemPoolGetAttribute (handle, CU_MEMPOOL_ATTR_RESERVED_MEM_CURRENT,
+        (void *) &reserved);
+    if (!gst_cuda_result (ret)) {
+      gst_printerrln ("Couldn't get reserved size");
+      return 0;
+    }
+
+    ret = CuMemPoolGetAttribute (handle, CU_MEMPOOL_ATTR_USED_MEM_CURRENT,
+        (void *) &current);
+    if (!gst_cuda_result (ret)) {
+      gst_printerrln ("Couldn't get current size");
+      return 0;
+    }
+
+    gst_buffer_map (buffer, &map_info, GST_MAP_READ | GST_MAP_CUDA);
+
+    gst_println ("[%d] After allocation, address %p, reserved: %"
+        G_GUINT64_FORMAT ", current: %" G_GUINT64_FORMAT, i,
+        map_info.data, reserved, current);
+    gst_buffer_unmap (buffer, &map_info);
+    gst_cuda_context_pop (NULL);
+
+    /* Release pool and check pool status */
+    gst_buffer_unref (buffer);
+    gst_buffer_pool_set_active (pool, FALSE);
+    gst_object_unref (pool);
+
+    gst_cuda_context_push (context);
+    ret = CuMemPoolGetAttribute (handle, CU_MEMPOOL_ATTR_RESERVED_MEM_CURRENT,
+        (void *) &reserved);
+    if (!gst_cuda_result (ret)) {
+      gst_printerrln ("Couldn't get reserved size");
+      return 0;
+    }
+
+    ret = CuMemPoolGetAttribute (handle, CU_MEMPOOL_ATTR_USED_MEM_CURRENT,
+        (void *) &current);
+    if (!gst_cuda_result (ret)) {
+      gst_printerrln ("Couldn't get current size");
+      return 0;
+    }
+
+    gst_println ("[%d] After buffer pool release, reserved: %" G_GUINT64_FORMAT
+        ", current: %" G_GUINT64_FORMAT, i, reserved, current);
+    gst_cuda_context_pop (NULL);
+  }
+
+  gst_caps_unref (caps);
+  gst_cuda_memory_pool_unref (data.mem_pool);
+  gst_cuda_stream_unref (stream);
+  gst_object_unref (context);
+
+  gst_deinit ();
+
+  return 0;
+}

subprojects/gst-plugins-bad/tests/examples/cuda/meson.build

@@ -78,3 +78,9 @@ executable('cuda-template',
   dependencies: [gst_dep, gstvideo_dep, gstcuda_dep, gl_header_dep] + cuda_deps,
   c_args : gst_plugins_bad_args + ['-DGST_USE_UNSTABLE_API'],
   install: false)
+
+executable('cudamemorypool', 'cudamemorypool.c',
+  include_directories : [configinc, cuda_stubinc],
+  dependencies: [gst_dep, gstvideo_dep, gstcuda_dep],
+  c_args : gst_plugins_bad_args + ['-DGST_USE_UNSTABLE_API'],
+  install: false)