Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Fix a hanging issue when processing empty data. #841

Merged
merged 2 commits into from
Sep 25, 2020
Merged

Fix a hanging issue when processing empty data. #841

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
c199aaa
Fix a hanging issue when processing empty data.
firestarman Sep 23, 2020
fa7d545
Add tests for processing empty data.
firestarman Sep 24, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 10 additions & 6 deletions integration_tests/src/main/python/udf_cudf_test.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,10 +36,13 @@
'spark.rapids.sql.python.gpu.enabled': 'true'
}

small_data = [(1, 1.0), (1, 2.0), (2, 3.0), (2, 5.0), (2, 10.0)]

def _create_df(spark):
elements = list(map(lambda i: (i, i/1.0), range(1, 5000)))
return spark.createDataFrame(elements * 2, ("id", "v"))
large_data = list(map(lambda i: (i, i/1.0), range(1, 5000))) * 2


def _create_df(spark, data=large_data):
return spark.createDataFrame(data, ("id", "v"))


# since this test requires to run different functions on CPU and GPU(need cudf),
Expand Down Expand Up @@ -76,13 +79,14 @@ def _plus_one_gpu_func(v: pd.Series) -> pd.Series:


@cudf_udf
def test_with_column(enable_cudf_udf):
@pytest.mark.parametrize('data', [small_data, large_data], ids=['small data', 'large data'])
def test_with_column(enable_cudf_udf, data):
def cpu_run(spark):
df = _create_df(spark)
df = _create_df(spark, data)
return df.withColumn("v1", _plus_one_cpu_func(df.v)).collect()

def gpu_run(spark):
df = _create_df(spark)
df = _create_df(spark, data)
return df.withColumn("v1", _plus_one_gpu_func(df.v)).collect()

_assert_cpu_gpu(cpu_run, gpu_run, gpu_conf=_conf)
Expand Down
15 changes: 13 additions & 2 deletions .../src/main/scala/org/apache/spark/sql/rapids/execution/python/GpuArrowEvalPythonExec.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -187,6 +187,14 @@ class BatchQueue extends AutoCloseable with Arm {
}
}

def finish(): Unit = synchronized {
if (!isSet) {
// Wake up anyone waiting for the first batch.
isSet = true
notifyAll()
}
}

def remove(): ColumnarBatch = synchronized {
if (queue.isEmpty) {
null
Expand Down Expand Up @@ -369,7 +377,8 @@ class GpuArrowPythonRunner(
schema: StructType,
timeZoneId: String,
conf: Map[String, String],
batchSize: Long)
batchSize: Long,
onDataWriteFinished: () => Unit)
extends BasePythonRunner[ColumnarBatch, ColumnarBatch](funcs, evalType, argOffsets)
with GpuPythonArrowOutput {

Expand Down Expand Up @@ -431,6 +440,7 @@ class GpuArrowPythonRunner(
} {
writer.close()
dataOut.flush()
if (onDataWriteFinished != null) onDataWriteFinished()
}
}
}
Expand Down Expand Up @@ -587,7 +597,8 @@ case class GpuArrowEvalPythonExec(
schema,
sessionLocalTimeZone,
pythonRunnerConf,
batchSize){
batchSize,
() => queue.finish()){
override def minReadTargetBatchSize: Int = targetReadBatchSize
}.compute(projectedIterator,
context.partitionId(),
Expand Down