From 882c84e6465e44463cbd70d3435f5e1793e77d2e Mon Sep 17 00:00:00 2001
From: Martin Bauer <martin.bauer@fau.de>
Date: Wed, 22 Mar 2017 17:09:53 +0100
Subject: [PATCH] More flexible ghost layer specification for CUDA kernels
---
gpucuda/indexing.py | 15 ++++++++-------
gpucuda/kernelcreation.py | 18 +++++++++++++-----
transformations.py | 2 ++
3 files changed, 23 insertions(+), 12 deletions(-)
diff --git a/gpucuda/indexing.py b/gpucuda/indexing.py
index 7cc46c133..6db21f806 100644
--- a/gpucuda/indexing.py
+++ b/gpucuda/indexing.py
@@ -73,8 +73,8 @@ class BlockIndexing(AbstractIndexing):
blockSize = self.limitBlockSizeToDeviceMaximum(blockSize)
self._blockSize = blockSize
- self._coordinates = [blockIndex * bs + threadIndex + ghostLayers
- for blockIndex, bs, threadIndex in zip(BLOCK_IDX, blockSize, THREAD_IDX)]
+ self._coordinates = [blockIndex * bs + threadIndex + gl[0]
+ for blockIndex, bs, threadIndex, gl in zip(BLOCK_IDX, blockSize, THREAD_IDX, ghostLayers)]
self._coordinates = self._coordinates[:field.spatialDimensions]
self._ghostLayers = ghostLayers
@@ -85,7 +85,7 @@ class BlockIndexing(AbstractIndexing):
def getCallParameters(self, arrShape):
dim = len(self._coordinates)
- arrShape = arrShape[:dim]
+ arrShape = [s - (gl[0] + gl[1]) for s, gl in zip(arrShape[:dim], self._ghostLayers)]
grid = tuple(math.ceil(length / blockSize) for length, blockSize in zip(arrShape, self._blockSize))
extendBs = (1,) * (3 - len(self._blockSize))
extendGr = (1,) * (3 - len(grid))
@@ -95,8 +95,8 @@ class BlockIndexing(AbstractIndexing):
def guard(self, kernelContent, arrShape):
dim = len(self._coordinates)
arrShape = arrShape[:dim]
- conditions = [c < shapeComponent - self._ghostLayers
- for c, shapeComponent in zip(self._coordinates, arrShape)]
+ conditions = [c < shapeComponent - gl[1]
+ for c, shapeComponent, gl in zip(self._coordinates, arrShape, self._ghostLayers)]
condition = conditions[0]
for c in conditions[1:]:
condition = sp.And(condition, c)
@@ -189,7 +189,7 @@ class LineIndexing(AbstractIndexing):
coordinates[0], coordinates[fastestCoordinate] = coordinates[fastestCoordinate], coordinates[0]
self._coordiantesNoGhostLayer = coordinates
- self._coordinates = [i + ghostLayers for i in coordinates]
+ self._coordinates = [i + gl[0] for i, gl in zip(coordinates, ghostLayers)]
self._ghostLayers = ghostLayers
@property
@@ -201,7 +201,8 @@ class LineIndexing(AbstractIndexing):
if cudaIdx not in self._coordiantesNoGhostLayer:
return 1
else:
- return arrShape[self._coordiantesNoGhostLayer.index(cudaIdx)] - 2 * self._ghostLayers
+ idx = self._coordiantesNoGhostLayer.index(cudaIdx)
+ return arrShape[idx] - (self._ghostLayers[idx][0] + self._ghostLayers[idx][1])
return {'block': tuple([getShapeOfCudaIdx(idx) for idx in THREAD_IDX]),
'grid': tuple([getShapeOfCudaIdx(idx) for idx in BLOCK_IDX])}
diff --git a/gpucuda/kernelcreation.py b/gpucuda/kernelcreation.py
index ea06701ca..03a490aad 100644
--- a/gpucuda/kernelcreation.py
+++ b/gpucuda/kernelcreation.py
@@ -5,7 +5,8 @@ from pystencils.types import TypedSymbol, BasicType, StructType
from pystencils import Field
-def createCUDAKernel(listOfEquations, functionName="kernel", typeForSymbol=None, indexingCreator=BlockIndexing):
+def createCUDAKernel(listOfEquations, functionName="kernel", typeForSymbol=None, indexingCreator=BlockIndexing,
+ ghostLayers=None):
fieldsRead, fieldsWritten, assignments = typeAllEquations(listOfEquations, typeForSymbol)
allFields = fieldsRead.union(fieldsWritten)
readOnlyFields = set([f.name for f in fieldsRead - fieldsWritten])
@@ -14,11 +15,17 @@ def createCUDAKernel(listOfEquations, functionName="kernel", typeForSymbol=None,
for eq in listOfEquations:
fieldAccesses.update(eq.atoms(Field.Access))
- requiredGhostLayers = max([fa.requiredGhostLayers for fa in fieldAccesses])
- indexing = indexingCreator(field=list(fieldsRead)[0], ghostLayers=requiredGhostLayers)
+ commonShape = getCommonShape(allFields)
+ if ghostLayers is None:
+ requiredGhostLayers = max([fa.requiredGhostLayers for fa in fieldAccesses])
+ ghostLayers = [(requiredGhostLayers, requiredGhostLayers)] * len(commonShape)
+ if isinstance(ghostLayers, int):
+ ghostLayers = [(ghostLayers, ghostLayers)] * len(commonShape)
+
+ indexing = indexingCreator(field=list(fieldsRead)[0], ghostLayers=ghostLayers)
block = Block(assignments)
- block = indexing.guard(block, getCommonShape(allFields))
+ block = indexing.guard(block, commonShape)
ast = KernelFunction(block, allFields, functionName)
ast.globalVariables.update(indexing.indexVariables)
@@ -63,7 +70,8 @@ def createdIndexedCUDAKernel(listOfEquations, indexFields, functionName="kernel"
coordinateSymbolAssignments = [getCoordinateSymbolAssignment(n) for n in coordinateNames[:spatialCoordinates]]
coordinateTypedSymbols = [eq.lhs for eq in coordinateSymbolAssignments]
- indexing = indexingCreator(field=list(indexFields)[0], ghostLayers=0)
+ idxField = list(indexFields)[0]
+ indexing = indexingCreator(field=idxField, ghostLayers=[(0, 0)] * len(idxField.shape))
functionBody = Block(coordinateSymbolAssignments + assignments)
functionBody = indexing.guard(functionBody, getCommonShape(indexFields))
diff --git a/transformations.py b/transformations.py
index a48125dea..19aebd937 100644
--- a/transformations.py
+++ b/transformations.py
@@ -77,6 +77,8 @@ def makeLoopOverDomain(body, functionName, iterationSlice=None, ghostLayers=None
if ghostLayers is None:
requiredGhostLayers = max([fa.requiredGhostLayers for fa in fieldAccesses])
ghostLayers = [(requiredGhostLayers, requiredGhostLayers)] * len(loopOrder)
+ if isinstance(ghostLayers, int):
+ ghostLayers = [(ghostLayers, ghostLayers)] * len(loopOrder)
currentBody = body
lastLoop = None
--
GitLab