From e16b2619341e1505e0746f9553be73e2ad0dc01b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Ren=C3=A9=20He=C3=9F?= <rene.hess@iwr.uni-heidelberg.de>
Date: Thu, 15 Nov 2018 21:07:51 +0100
Subject: [PATCH] [skip ci][wip] Move input permutation to interface classes
So far only for fastdg. This should also happen in the non-fastdg case.
---
python/dune/codegen/sumfact/basis.py | 14 ++++++++-
python/dune/codegen/sumfact/realization.py | 24 +++++++---------
python/dune/codegen/sumfact/symbolic.py | 33 +++++++++++++++++++---
3 files changed, 52 insertions(+), 19 deletions(-)
diff --git a/python/dune/codegen/sumfact/basis.py b/python/dune/codegen/sumfact/basis.py
index 3ce2fb24..ccc34406 100644
--- a/python/dune/codegen/sumfact/basis.py
+++ b/python/dune/codegen/sumfact/basis.py
@@ -24,7 +24,8 @@ from dune.codegen.sumfact.tabulation import (basis_functions_per_direction,
name_polynomials,
polynomial_degree,
)
-from dune.codegen.sumfact.permutation import (permute_forward,
+from dune.codegen.sumfact.permutation import (permute_backward,
+ permute_forward,
sumfact_cost_permutation_strategy,
sumfact_quadrature_permutation_strategy,
)
@@ -144,6 +145,17 @@ class LFSSumfactKernelInput(SumfactKernelInterfaceBase, ImmutableRecord):
return prim.Subscript(prim.Variable(arg), inames)
+ def realize_input(self, shape, inames, which=0):
+ if self.direct_is_possible:
+ shape = permute_backward(shape, self.cost_permutation)
+ shape = permute_backward(shape, self.quadrature_permutation)
+ inames = permute_backward(inames, self.cost_permutation)
+ inames = permute_backward(inames, self.quadrature_permutation)
+
+ return self.realize_direct(shape, inames)
+ else:
+ raise NotImplementedError("TODO")
+
@property
def function_name_suffix(self):
if get_form_option("fastdg"):
diff --git a/python/dune/codegen/sumfact/realization.py b/python/dune/codegen/sumfact/realization.py
index 1d951eca..b9d209c3 100644
--- a/python/dune/codegen/sumfact/realization.py
+++ b/python/dune/codegen/sumfact/realization.py
@@ -160,8 +160,12 @@ def realize_sumfact_kernel_function(sf):
for l, matrix in enumerate(matrix_sequence):
# Compute the correct shapes of in- and output matrices of this matrix-matrix multiplication
# and get inames that realize the product.
- inp_shape = (matrix.cols,) + tuple(mat.cols for mat in matrix_sequence[l + 1:]) + tuple(mat.rows for mat in matrix_sequence[:l])
- out_shape = (matrix.rows,) + tuple(mat.cols for mat in matrix_sequence[l + 1:]) + tuple(mat.rows for mat in matrix_sequence[:l])
+ inp_shape = (matrix.cols,) \
+ + tuple(mat.cols for mat in matrix_sequence[l + 1:]) \
+ + tuple(mat.rows for mat in matrix_sequence[:l])
+ out_shape = (matrix.rows,) \
+ + tuple(mat.cols for mat in matrix_sequence[l + 1:]) \
+ + tuple(mat.rows for mat in matrix_sequence[:l])
out_inames = tuple(sumfact_iname(length, "out_inames_" + str(k)) for k, length in enumerate(out_shape))
vec_iname = ()
if matrix.vectorized:
@@ -173,11 +177,11 @@ def realize_sumfact_kernel_function(sf):
# a code generation corner case producing way too complicated code. This
# could be fixed upstream, but the loopy code realizing reductions is not
# trivial and the priority is kind of low.
- if matrix.cols != 1:
+ if matrix.cols == 1:
+ k_expr = 0
+ else:
k = sumfact_iname(matrix.cols, "red")
k_expr = prim.Variable(k)
- else:
- k_expr = 0
# Setup the input of the sum factorization kernel. In the
# first matrix multiplication this can be taken from
@@ -187,15 +191,7 @@ def realize_sumfact_kernel_function(sf):
# (vectorized + FastDGGridOperator)
input_inames = (k_expr,) + tuple(prim.Variable(j) for j in out_inames[1:])
if l == 0 and sf.stage == 1 and sf.interface.direct_is_possible:
- # One permutation for cost reduction, see comment below
- inp_shape = permute_backward(inp_shape, sf.cost_permutation)
- input_inames = permute_backward(input_inames, sf.cost_permutation)
-
- # And one more for permuted quadrature points, see comment below
- inp_shape = permute_backward(inp_shape, sf.interface.quadrature_permutation)
- input_inames = permute_backward(input_inames, sf.interface.quadrature_permutation)
-
- input_summand = sf.interface.realize_direct(inp_shape, input_inames)
+ input_summand = sf.interface.realize_input(inp_shape, input_inames)
else:
# If we did permute the order of a matrices above we also
# permuted the order of out_inames. Unfortunately the
diff --git a/python/dune/codegen/sumfact/symbolic.py b/python/dune/codegen/sumfact/symbolic.py
index 82f7e027..0cafec86 100644
--- a/python/dune/codegen/sumfact/symbolic.py
+++ b/python/dune/codegen/sumfact/symbolic.py
@@ -40,11 +40,11 @@ class SumfactKernelInterfaceBase(object):
@property
def quadrature_permutation(self):
- return ()
+ raise NotImplementedError
@property
def cost_permutation(self):
- return ()
+ raise NotImplementedError
@property
def combined_permutation(self):
@@ -63,12 +63,12 @@ class SumfactKernelInterfaceBase(object):
def permute_forward_cost(self, shape, inames):
shape = permute_forward(shape, self.cost_permutation)
inames = permute_forward(inames, self.cost_permutation)
- return shape_inames
+ return shape, inames
def permute_forward_quadrature(self, shape, inames):
shape = permute_forward(shape, self.quadrature_permutation)
inames = permute_forward(inames, self.quadrature_permutation)
- return shape_inames
+ return shape, inames
@property
def within_inames(self):
@@ -113,6 +113,20 @@ class VectorSumfactKernelInput(SumfactKernelInterfaceBase):
assert i.quadrature_permutation == self.interfaces[0].quadrature_permutation
return self.interfaces[0].quadrature_permutation
+ @property
+ def cost_permutation(self):
+ # The cost_permutation of the underlying scalar SumfactKernel can be
+ # different for each kernel.
+
+ # TODO!
+ cost_permutation = self.interfaces[0].cost_permutation
+ for i in self.interfaces:
+ assert i.cost_permutation == cost_permutation
+
+ return cost_permutation
+
+ # raise RuntimeError("cost_permutation should not be called on VectorSumfactKernelInput")
+
@property
def stage(self):
return 1
@@ -153,6 +167,17 @@ class VectorSumfactKernelInput(SumfactKernelInterfaceBase):
# need to load scalars into the SIMD vector.
raise NotImplementedError("SIMD loads from scalars not implemented!")
+ def realize_input(self, shape, inames):
+ if self.direct_is_possible:
+ shape = permute_backward(shape, self.cost_permutation)
+ shape = permute_backward(shape, self.quadrature_permutation)
+ inames = permute_backward(inames, self.cost_permutation)
+ inames = permute_backward(inames, self.quadrature_permutation)
+
+ return self.realize_direct(shape, inames)
+ else:
+ raise NotImplementedError("TODO")
+
@property
def function_args(self):
return sum((i.function_args for i in remove_duplicates(self.interfaces)), ())
--
GitLab