[Archive] Preserve original PR-2653 graph processing head

fern/pages/lucene_api/lucene-api-com-nvidia-cuvs-lucene-acceleratedhnswparams.md

@@ -18,11 +18,11 @@ public class AcceleratedHNSWParams
 public int getWriterThreads()
 ```
 
-Get the cuVS writer threads parameter
+Get the maximum thread count for cuVS writes and post-build graph processing.
 
 **Returns**
 
-cuVS writer threads parameter
+maximum writer and graph-processing thread count
 
 _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWParams.java:149`_
 
@@ -218,7 +218,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWP
 public Builder withWriterThreads(int writerThreads)
 ```
 
-Set the number of cuVS writer threads while building the index
+Set the maximum number of threads used for cuVS writes and post-build graph processing.
 Valid range - Minimum: \{@value MIN_WRITER_THREADS\}, Maximum: \{@value MAX_WRITER_THREADS\}
 Default value - \{@value DEFAULT_WRITER_THREADS\}
 

fern/pages/lucene_api/lucene-api-com-nvidia-cuvs-lucene-acceleratedhnswutils.md

@@ -21,7 +21,7 @@ public static GPUBuiltHnswGraph createSingleVectorHnswGraph(int size, int dimens
 Creates a dummy HNSW graph for a single vector.
 The graph will have 1 level with 1 node and no neighbors.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:54`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:56`_
 
 ### createMultiLayerHnswGraph
 
@@ -35,7 +35,7 @@ M = ceil(cagraGraphDegree / 2), where cagraGraphDegree is the CAGRA adjacency li
 Each layer contains 1/M nodes from the previous layer
 Creates layers until the highest layer has ≤ M nodes
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:81`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:83`_
 
 ### writeGraph
 
@@ -62,7 +62,7 @@ a 2D array of offsets
 | --- | --- |
 | `IOException` | I/O Exceptions |
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:237`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:263`_
 
 ### writeMeta
 
@@ -91,7 +91,7 @@ Writes the meta information for the index.
 | --- | --- |
 | `IOException` | I/O Exceptions |
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:302`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:443`_
 
 ### printInfoStream
 
@@ -107,7 +107,7 @@ A utility method to print info/debugging messages using InfoStream.
 | --- | --- |
 | `msg` | the debugging message to print |
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:384`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:525`_
 
 ### writeEmpty
 
@@ -129,7 +129,7 @@ Writes an empty meta information for the field.
 | --- | --- |
 | `IOException` | I/O Exceptions |
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:396`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:537`_
 
 ### quantizeFloatVectorsToBinary
 
@@ -152,7 +152,7 @@ Bits are packed: 8 dimensions per byte.
 
 A list of byte binary representation for the input vectors
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:409`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:550`_
 
 ### quantizeFloatVectorsToScalar
 
@@ -172,6 +172,6 @@ Scalar quantization.
 
 A list of byte scalar representation for the input vectors
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:451`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:592`_
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:31`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWUtils.java:33`_

fern/pages/lucene_api/lucene-api-com-nvidia-cuvs-lucene-gpubuilthnswgraph.md

@@ -31,7 +31,27 @@ Multi-layer constructor that supports arbitrary number of layers.
 | `layerNodes` | the nodes on the layer |
 | `layerAdjacencies` | adjacency list |
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:41`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:51`_
+
+### create
+
+```java
+static GPUBuiltHnswGraph create( int size, int dimensions, List<int[]> layerNodes, List<CuVSMatrix> layerAdjacencies, int requestedWorkers) throws IOException
+```
+
+Builds a graph while bounding adjacency materialization to the requested writer threads.
+
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:66`_
+
+### fitsParallelGraphCopyBudget
+
+```java
+static boolean fitsParallelGraphCopyBudget(long rows, long columns)
+```
+
+Returns whether an INT32 graph can be copied without exceeding the host-memory budget.
+
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:162`_
 
 ### getNodesOnLevel
 
@@ -41,7 +61,7 @@ public NodesIterator getNodesOnLevel(int level)
 
 Get all nodes on a given level as node 0th ordinals.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:89`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:234`_
 
 ### getNeighbors
 
@@ -62,7 +82,7 @@ Get the neighbors for the node and the level it resides.
 
 an instance of NeighborArray
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:107`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:252`_
 
 ### seek
 
@@ -72,7 +92,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGrap
 
 Move the pointer to exactly the given level's target.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:132`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:277`_
 
 ### nextNeighbor
 
@@ -82,7 +102,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGrap
 
 Iterates over the neighbor list.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:142`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:287`_
 
 ### entryNode
 
@@ -92,7 +112,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGrap
 
 Returns graph's entry point on the top level.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:173`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:318`_
 
 ### maxConn
 
@@ -102,7 +122,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGrap
 
 returns M, the maximum number of connections for a node.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:192`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:337`_
 
 ### neighborCount
 
@@ -112,7 +132,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGrap
 
 Returns the neighbor count.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:207`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:352`_
 
 ### size
 
@@ -122,7 +142,7 @@ public int size()
 
 Returns the number of nodes in the graph.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:282`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:427`_
 
 ### numLevels
 
@@ -136,7 +156,7 @@ Returns the number of levels in the HNSW graph.
 
 the number of levels
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:291`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:436`_
 
 ### dimensions
 
@@ -150,6 +170,6 @@ Gets the vector dimension.
 
 the vector dimension
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:300`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:445`_
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:21`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/GPUBuiltHnswGraph.java:25`_

fern/pages/lucene_api/lucene-api-com-nvidia-cuvs-lucene-lucene99acceleratedhnswvectorswriter.md

@@ -57,7 +57,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/Lucene99Accelera
 
 Build the indexes and writes it to the disk.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/Lucene99AcceleratedHNSWVectorsWriter.java:203`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/Lucene99AcceleratedHNSWVectorsWriter.java:205`_
 
 ### mergeOneField
 
@@ -67,7 +67,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/Lucene99Accelera
 
 Write field for merging.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/Lucene99AcceleratedHNSWVectorsWriter.java:291`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/Lucene99AcceleratedHNSWVectorsWriter.java:293`_
 
 ### finish
 
@@ -77,7 +77,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/Lucene99Accelera
 
 Called once at the end before close.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/Lucene99AcceleratedHNSWVectorsWriter.java:300`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/Lucene99AcceleratedHNSWVectorsWriter.java:302`_
 
 ### close
 
@@ -87,7 +87,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/Lucene99Accelera
 
 Closes the resources.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/Lucene99AcceleratedHNSWVectorsWriter.java:320`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/Lucene99AcceleratedHNSWVectorsWriter.java:322`_
 
 ### ramBytesUsed
 
@@ -97,6 +97,6 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/Lucene99Accelera
 
 Returns the memory usage of this object in bytes.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/Lucene99AcceleratedHNSWVectorsWriter.java:330`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/Lucene99AcceleratedHNSWVectorsWriter.java:332`_
 
 _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/Lucene99AcceleratedHNSWVectorsWriter.java:52`_

fern/pages/lucene_api/lucene-api-com-nvidia-cuvs-lucene-luceneacceleratedhnswbinaryquantizedvectorswriter.md

@@ -57,7 +57,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAccelerate
 
 Build the indexes and writes it to the disk.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWBinaryQuantizedVectorsWriter.java:215`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWBinaryQuantizedVectorsWriter.java:217`_
 
 ### mergeOneField
 
@@ -67,7 +67,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAccelerate
 
 Write field for merging.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWBinaryQuantizedVectorsWriter.java:302`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWBinaryQuantizedVectorsWriter.java:304`_
 
 ### finish
 
@@ -77,7 +77,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAccelerate
 
 Called once at the end before close.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWBinaryQuantizedVectorsWriter.java:333`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWBinaryQuantizedVectorsWriter.java:335`_
 
 ### close
 
@@ -87,7 +87,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAccelerate
 
 Closes the resources.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWBinaryQuantizedVectorsWriter.java:353`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWBinaryQuantizedVectorsWriter.java:355`_
 
 ### ramBytesUsed
 
@@ -97,6 +97,6 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAccelerate
 
 Returns the memory usage of this object in bytes.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWBinaryQuantizedVectorsWriter.java:362`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWBinaryQuantizedVectorsWriter.java:364`_
 
 _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWBinaryQuantizedVectorsWriter.java:57`_

fern/pages/lucene_api/lucene-api-com-nvidia-cuvs-lucene-luceneacceleratedhnswscalarquantizedvectorswriter.md

@@ -57,7 +57,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAccelerate
 
 Build the indexes and writes it to the disk.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWScalarQuantizedVectorsWriter.java:241`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWScalarQuantizedVectorsWriter.java:243`_
 
 ### mergeOneField
 
@@ -67,7 +67,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAccelerate
 
 Write field for merging.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWScalarQuantizedVectorsWriter.java:326`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWScalarQuantizedVectorsWriter.java:328`_
 
 ### finish
 
@@ -77,7 +77,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAccelerate
 
 Called once at the end before close.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWScalarQuantizedVectorsWriter.java:357`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWScalarQuantizedVectorsWriter.java:359`_
 
 ### close
 
@@ -87,7 +87,7 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAccelerate
 
 Closes the resources.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWScalarQuantizedVectorsWriter.java:377`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWScalarQuantizedVectorsWriter.java:379`_
 
 ### ramBytesUsed
 
@@ -97,6 +97,6 @@ _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAccelerate
 
 Returns the memory usage of this object in bytes.
 
-_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWScalarQuantizedVectorsWriter.java:386`_
+_Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWScalarQuantizedVectorsWriter.java:388`_
 
 _Source: `java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/LuceneAcceleratedHNSWScalarQuantizedVectorsWriter.java:56`_

fern/pages/user_guide/lucene.md

@@ -231,7 +231,9 @@ Both `AcceleratedHNSWParams` and `GPUSearchParams` default to a `HEURISTIC` stra
 - For the accelerated HNSW codecs, set `maxConn` and `beamWidth`, the HNSW parameters you would tune on the CPU. cuVS derives graph degrees and the build algorithm from them. These two values also configure the CPU fallback writer, so one setting covers both paths.
 - For the GPU search codec, set `buildQuality`. Higher values spend more build time for a higher-quality graph. This codec also passes `graphDegree` into the heuristic, so leave it at its default unless you intend to cap the graph.
 
-Increasing `writerThreads` raises index build concurrency. The accelerated HNSW codecs default to a single writer thread, while the GPU search codec defaults to 32.
+`writerThreads` controls cuVS writer concurrency for both parameter types. For the accelerated HNSW codecs, it also bounds the CPU workers used to materialize and serialize the finished graph; the effective worker count cannot exceed the available processors or work items, and the default of one keeps this post-processing serial. The GPU search codec continues to use the setting only for native writer concurrency and defaults to 32.
+
+With more than one effective worker, parallel device-graph materialization is eligible for adjacency layers with at least 65,536 nodes. It uses one complete temporary native-host copy of an eligible device adjacency when that copy is at most 4 GiB; device adjacencies above that limit use serial materialization, although level-zero serialization can still run in parallel.
 
 Switching either class to the `CUSTOM` strategy exposes the underlying CAGRA parameters directly, including `graphDegree`, `intermediateGraphDegree`, the graph build algorithm, and its parameters. Use `CUSTOM` only when you have measurements that justify specific values; the defaults derived by cuVS are a better starting point. For background on the parameters themselves, see the [CAGRA indexing guide](/user-guide/api-guides/indexing-guide/cagra) and the [tuning guide](/getting-started/introduction/tuning-indexes).
 

java/cuvs-lucene/src/main/java/com/nvidia/cuvs/lucene/AcceleratedHNSWParams.java

@@ -95,7 +95,7 @@ public static enum Strategy {
   /**
    * Constructs an instance of {@link AcceleratedHNSWParams} with specific parameter values.
    *
-   * @param writerThreads Number of cuVS writer threads to use.
+   * @param writerThreads Maximum threads to use for cuVS writes and post-build graph processing.
    * @param intermediateGraphDegree The intermediate graph degree while building the CAGRA index.
    * @param graphdegree The graph degree to use while building the CAGRA index.
    * @param hnswLayers The number of HNSW layers to build in the HNSW index.
@@ -143,9 +143,9 @@ private AcceleratedHNSWParams(
   }
 
   /**
-   * Get the cuVS writer threads parameter
+   * Get the maximum thread count for cuVS writes and post-build graph processing.
    *
-   * @return cuVS writer threads parameter
+   * @return maximum writer and graph-processing thread count
    */
   public int getWriterThreads() {
     return writerThreads;
@@ -326,7 +326,7 @@ public static class Builder {
     private HnswHeuristicType hnswHeuristicType = DEFAULT_HNSW_HEURISTIC_TYPE;
 
     /**
-     * Set the number of cuVS writer threads while building the index
+     * Set the maximum number of threads used for cuVS writes and post-build graph processing.
      * Valid range - Minimum: {@value MIN_WRITER_THREADS}, Maximum: {@value MAX_WRITER_THREADS}
      * Default value - {@value DEFAULT_WRITER_THREADS}
      *