diff --git a/MEMO b/MEMO
index 20f7895ec6cb60f181409b2e0da813c6acebb433..dd7220c4b6d21848aa359fed8cc7b4cfc0b74584 100644
--- a/MEMO
+++ b/MEMO
@@ -70,9 +70,14 @@ Things to consider
-> Only reduction
- Slab decomposition for parallel dimensions
+ - implement at the outermost nesting level regardless
+ - bound *all* tagged inames
- Sharing of checks across ILP instances
+- Loop bounds currently may not depend on parallel dimensions
+ Does it make sense to relax this?
+
Dealt with
^^^^^^^^^^
diff --git a/loopy/__init__.py b/loopy/__init__.py
index add477149517a64ecf588b8f58477edfbbd27050..6f1c9b930e51177ccde0f892a577ae3efcb7eace 100644
--- a/loopy/__init__.py
+++ b/loopy/__init__.py
@@ -79,7 +79,6 @@ def split_dimension(kernel, iname, inner_length, padded_length=None,
.project_out(name_dim_type, name_idx, 1))
new_domain = process_set(kernel.domain)
- new_assumptions = process_set(kernel.assumptions)
from pymbolic import var
inner = var(inner_iname)
@@ -121,7 +120,6 @@ def split_dimension(kernel, iname, inner_length, padded_length=None,
iname_slab_increments[outer_iname] = outer_slab_increments
result = (kernel
.copy(domain=new_domain,
- assumptions=new_assumptions,
iname_slab_increments=iname_slab_increments,
iname_to_dim=None,
instructions=new_insns))
@@ -380,7 +378,6 @@ def realize_cse(kernel, cse_tag, dtype, duplicate_inames=[], parallel_inames=Non
return result, new_iname_to_dim
new_domain, new_iname_to_dim = realize_duplication(kernel.domain)
- new_assumptions, _ = realize_duplication(kernel.assumptions)
# }}}
@@ -397,7 +394,7 @@ def realize_cse(kernel, cse_tag, dtype, duplicate_inames=[], parallel_inames=Non
from loopy.symbolic import pw_aff_to_expr
target_var_shape.append(static_max_of_pw_aff(
- upper_bound_pw_aff - lower_bound_pw_aff + 1))
+ upper_bound_pw_aff - lower_bound_pw_aff + 1, constants_only=True))
target_var_base_indices.append(pw_aff_to_expr(lower_bound_pw_aff))
from loopy.kernel import TemporaryVariable
@@ -417,7 +414,6 @@ def realize_cse(kernel, cse_tag, dtype, duplicate_inames=[], parallel_inames=Non
return kernel.copy(
domain=new_domain,
- assumptions=new_assumptions,
instructions=new_insns,
temporary_variables=new_temporary_variables,
iname_to_dim=new_iname_to_dim,
@@ -426,81 +422,7 @@ def realize_cse(kernel, cse_tag, dtype, duplicate_inames=[], parallel_inames=Non
-def realize_reduction(kernel, inames=None, reduction_tag=None):
- new_insns = []
- new_temporary_variables = kernel.temporary_variables.copy()
-
- def map_reduction(expr, rec):
- sub_expr = rec(expr.expr)
-
- if reduction_tag is not None and expr.tag != reduction_tag:
- return
-
- if inames is not None and set(inames) != set(expr.inames):
- return
-
- from pymbolic import var
-
- target_var_name = kernel.make_unique_var_name("red",
- extra_used_vars=set(tv for tv in new_temporary_variables))
- target_var = var(target_var_name)
-
- from loopy.kernel import Instruction
-
- from loopy.kernel import TemporaryVariable
- new_temporary_variables[target_var_name] = TemporaryVariable(
- name=target_var_name,
- dtype=expr.operation.dtype,
- shape=(),
- is_local=False)
-
- init_insn = Instruction(
- id=kernel.make_unique_instruction_id(
- extra_used_ids=set(ni.id for ni in new_insns)),
- assignee=target_var,
- forced_iname_deps=list(insn.all_inames() - set(expr.inames)),
- expression=expr.operation.neutral_element)
-
- new_insns.append(init_insn)
-
- reduction_insn = Instruction(
- id=kernel.make_unique_instruction_id(
- extra_used_ids=set(ni.id for ni in new_insns)),
- assignee=target_var,
- expression=expr.operation(target_var, sub_expr),
- insn_deps=[init_insn.id],
- forced_iname_deps=list(insn.all_inames()))
-
- new_insns.append(reduction_insn)
-
- new_insn_insn_deps.append(reduction_insn.id)
- new_insn_removed_inames.extend(expr.inames)
-
- return target_var
-
- from loopy.symbolic import ReductionCallbackMapper
- cb_mapper = ReductionCallbackMapper(map_reduction)
-
- for insn in kernel.instructions:
- new_insn_insn_deps = []
- new_insn_removed_inames = []
-
- new_expression = cb_mapper(insn.expression)
-
- new_insns.append(
- insn.copy(
- expression=new_expression,
- insn_deps=insn.insn_deps
- + new_insn_insn_deps))
-
- return kernel.copy(
- instructions=new_insns,
- temporary_variables=new_temporary_variables)
-
-
-
-
-def get_problems(kernel, parameters, emit_warnings=True):
+def get_problems(kernel, parameters):
"""
:return: *(max_severity, list of (severity, msg))*, where *severity* ranges from 1-5.
'5' means 'will certainly not run'.
@@ -508,15 +430,9 @@ def get_problems(kernel, parameters, emit_warnings=True):
msgs = []
def msg(severity, s):
- if emit_warnings:
- from warnings import warn
- from loopy import LoopyAdvisory
- warn(s, LoopyAdvisory)
-
msgs.append((severity, s))
- glens = kernel.tag_type_lengths(TAG_GROUP_IDX, allow_parameters=True)
- llens = kernel.tag_type_lengths(TAG_LOCAL_IDX, allow_parameters=False)
+ glens, llens = kernel.get_grid_sizes_as_exprs()
from pymbolic import evaluate
glens = evaluate(glens, parameters)
@@ -534,6 +450,7 @@ def get_problems(kernel, parameters, emit_warnings=True):
if product(llens) > kernel.device.max_work_group_size:
msg(5, "work group too big")
+ import pyopencl as cl
from pyopencl.characterize import usable_local_mem_size
if kernel.local_mem_use() > usable_local_mem_size(kernel.device):
if kernel.device.local_mem_type == cl.device_local_mem_type.LOCAL:
@@ -554,6 +471,23 @@ def get_problems(kernel, parameters, emit_warnings=True):
max_severity = max(sev, max_severity)
return max_severity, msgs
+
+
+
+def check_kernels(kernel_gen, parameters, kill_level_min=3,
+ warn_level_min=1):
+ for kernel in kernel_gen:
+ max_severity, msgs = get_problems(kernel, parameters)
+
+ for severity, msg in msgs:
+ if severity >= warn_level_min:
+ from warnings import warn
+ from loopy import LoopyAdvisory
+ warn(msg, LoopyAdvisory)
+
+ if max_severity < kill_level_min:
+ yield kernel
+
# }}}
# {{{ high-level modifiers
diff --git a/loopy/codegen/__init__.py b/loopy/codegen/__init__.py
index 65a07c6d0a96939cffa5b9604618a687bf78605c..b15178dc06f4ead28575f39794f0c0fc558bdfe6 100644
--- a/loopy/codegen/__init__.py
+++ b/loopy/codegen/__init__.py
@@ -217,6 +217,9 @@ def generate_code(kernel):
( (a) - ( ((a)<0) ? ((b)-1) : 0 ) ) / (b) \
)
+ #define lid(N) ((int) get_local_id(N))
+ #define gid(N) ((int) get_group_id(N))
+
"""),
Line()])
@@ -243,15 +246,23 @@ def generate_code(kernel):
from loopy.codegen.dispatch import build_loop_nest
+ from islpy import align_spaces
+ initial_implemented_domain = align_spaces(kernel.assumptions, kernel.domain)
gen_code = build_loop_nest(kernel, 0,
- CodeGenerationState(kernel.assumptions, c_code_mapper=ccm))
- body.extend([Line(), gen_code.ast])
+ CodeGenerationState(initial_implemented_domain, c_code_mapper=ccm))
+
+ body.append(Line())
+
+ if isinstance(gen_code.ast, Block):
+ body.extend(gen_code.ast.contents)
+ else:
+ body.append(gen_code.ast)
from loopy.symbolic import pw_aff_to_expr
mod.append(
FunctionBody(
CLRequiredWorkGroupSize(
- tuple(pw_aff_to_expr(sz) for sz in kernel.fix_grid_sizes()[1]),
+ tuple(pw_aff_to_expr(sz) for sz in kernel.get_grid_sizes()[1]),
CLKernel(FunctionDeclaration(
Value("void", kernel.name), args))),
body))
diff --git a/loopy/codegen/bounds.py b/loopy/codegen/bounds.py
index 0b97179b5aed6c0c3bc82f6ecac89da181d697a2..06bef83e7ef36014a708c1621491a46003bc4c74 100644
--- a/loopy/codegen/bounds.py
+++ b/loopy/codegen/bounds.py
@@ -31,10 +31,6 @@ def get_bounds_constraints(set, iname, admissible_inames, allow_parameters):
bset, = basic_sets
- # FIXME: hackety hack--elimination leaves the set in an
- # invalid ('non-final'?) state
- bset = bset.intersect(isl.BasicSet.universe(bset.get_space()))
-
# FIXME perhaps use some form of hull here if there's more than one
# basic set?
@@ -156,11 +152,6 @@ def generate_bounds_checks(domain, check_vars, implemented_domain):
.coalesce()
.get_basic_sets())
- # FIXME: hackety hack--elimination leaves the set in an
- # invalid ('non-final'?) state
- domain_bset = domain_bset.intersect(
- isl.BasicSet.universe(domain_bset.get_space()))
-
return filter_necessary_constraints(
implemented_domain, domain_bset.get_constraints())
@@ -248,7 +239,7 @@ def wrap_in_for_from_constraints(ccm, iname, constraint_bset, stmt):
# {{{ on which variables may a conditional depend?
-def get_defined_inames(kernel, sched_index, allow_ilp, exclude_tag_classes=()):
+def get_defined_inames(kernel, sched_index, allow_tag_classes=()):
"""
:param exclude_tags: a tuple of tag classes to exclude
"""
@@ -264,16 +255,10 @@ def get_defined_inames(kernel, sched_index, allow_ilp, exclude_tag_classes=()):
elif isinstance(sched_item, LeaveLoop):
result.remove(sched_item.iname)
- from loopy.kernel import TAG_ILP, ParallelTagWithAxis
for iname in kernel.all_inames():
tag = kernel.iname_to_tag.get(iname)
- if isinstance(tag, exclude_tag_classes):
- continue
-
- if isinstance(tag, ParallelTagWithAxis):
- result.add(iname)
- elif isinstance(tag, TAG_ILP) and allow_ilp:
+ if isinstance(tag, allow_tag_classes):
result.add(iname)
return result
diff --git a/loopy/codegen/instruction.py b/loopy/codegen/instruction.py
index 5651a24d5f5df198de3aa66f7ce0be332b74f6a2..e9ddcf2c0cca2f19b53232c0697dade535ccf421 100644
--- a/loopy/codegen/instruction.py
+++ b/loopy/codegen/instruction.py
@@ -57,11 +57,11 @@ def generate_ilp_instances(kernel, insn, codegen_state):
tag = kernel.iname_to_tag.get(iname)
if isinstance(tag, TAG_LOCAL_IDX):
- hw_axis_expr = var("(int) get_local_id")(tag.axis)
+ hw_axis_expr = var("lid")(tag.axis)
hw_axis_size = local_size[tag.axis]
elif isinstance(tag, TAG_GROUP_IDX):
- hw_axis_expr = var("(int) get_group_id")(tag.axis)
+ hw_axis_expr = var("gid")(tag.axis)
hw_axis_size = global_size[tag.axis]
else:
@@ -70,12 +70,12 @@ def generate_ilp_instances(kernel, insn, codegen_state):
bounds = kernel.get_iname_bounds(iname)
from loopy.isl import make_slab
- impl_domain = impl_domain.intersect(
- make_slab(impl_domain.get_space(), iname,
- bounds.lower_bound_pw_aff, bounds.lower_bound_pw_aff+hw_axis_size))
+ slab = make_slab(impl_domain.get_space(), iname,
+ bounds.lower_bound_pw_aff, bounds.lower_bound_pw_aff+hw_axis_size)
+ impl_domain = impl_domain.intersect(slab)
from loopy.symbolic import pw_aff_to_expr
- assignments[iname] = pw_aff_to_expr(bounds.lower_bound_pw_aff + hw_axis_expr)
+ assignments[iname] = pw_aff_to_expr(bounds.lower_bound_pw_aff) + hw_axis_expr
# }}}
diff --git a/loopy/codegen/loop.py b/loopy/codegen/loop.py
index 0260ca858c5108f9e53ca07315166441b191f251..e667a9d76de15e15759629b226939e3eee80342a 100644
--- a/loopy/codegen/loop.py
+++ b/loopy/codegen/loop.py
@@ -17,7 +17,7 @@ def get_simple_loop_bounds(kernel, sched_index, iname, implemented_domain):
lower_constraints_orig, upper_constraints_orig, equality_constraints_orig = \
get_bounds_constraints(kernel.domain, iname,
frozenset([iname])
- | frozenset(get_defined_inames(kernel, sched_index+1, allow_ilp=False)),
+ | frozenset(get_defined_inames(kernel, sched_index+1)),
allow_parameters=True)
assert not equality_constraints_orig
@@ -27,6 +27,9 @@ def get_simple_loop_bounds(kernel, sched_index, iname, implemented_domain):
return lb_cns_orig, ub_cns_orig
+
+
+
# {{{ conditional-minimizing slab decomposition
def get_slab_decomposition(kernel, sched_index, exec_domain):
diff --git a/loopy/compiled.py b/loopy/compiled.py
index 864683bdd13605d8d619a76e14ea13f77d3c7874..ed80d23719416648f52f57d68632158729d51bf8 100644
--- a/loopy/compiled.py
+++ b/loopy/compiled.py
@@ -49,10 +49,7 @@ class CompiledKernel:
self.size_args = size_args
from loopy.kernel import TAG_GROUP_IDX, TAG_LOCAL_IDX
- gsize_expr = tuple(self.kernel.tag_type_lengths(
- TAG_GROUP_IDX, allow_parameters=True))
- lsize_expr = tuple(self.kernel.tag_type_lengths(
- TAG_LOCAL_IDX, allow_parameters=False))
+ gsize_expr, lsize_expr = kernel.get_grid_sizes_as_exprs()
if not gsize_expr: gsize_expr = (1,)
if not lsize_expr: lsize_expr = (1,)
diff --git a/loopy/isl.py b/loopy/isl.py
index fac42ecf60793aa8205005a7c533c4c815d43789..ef1b6cf5153c3e2c8ff2d2f0ece8e3681933f912 100644
--- a/loopy/isl.py
+++ b/loopy/isl.py
@@ -90,65 +90,81 @@ def pw_aff_to_aff(pw_aff):
assert isinstance(pw_aff, isl.PwAff)
pieces = pw_aff.get_pieces()
- if len(pieces) != 1:
- raise NotImplementedError("only single-piece PwAff instances are supported here")
+ if len(pieces) == 0:
+ raise RuntimeError("PwAff does not have any pieces")
+ if len(pieces) > 1:
+ _, first_aff = pieces[0]
+ for _, other_aff in pieces[1:]:
+ if not first_aff.plain_is_equal(other_aff):
+ raise NotImplementedError("only single-valued piecewise affine "
+ "expressions are supported here--encountered "
+ "multi-valued expression '%s'" % pw_aff)
+
+ return first_aff
return pieces[0][1]
-def make_slab(space, iname, start, stop):
- if isinstance(start, isl.PwAff): start = pw_aff_to_aff(start)
- if isinstance(stop, isl.PwAff): stop = pw_aff_to_aff(stop)
+def dump_local_space(ls):
+ return " ".join("%s: %d" % (dt, ls.dim(getattr(dim_type, dt)))
+ for dt in dim_type.names)
+def make_slab(space, iname, start, stop):
zero = isl.Aff.zero_on_domain(space)
+ from islpy import align_spaces
+ if isinstance(start, isl.PwAff):
+ start = align_spaces(pw_aff_to_aff(start), zero)
+ if isinstance(stop, isl.PwAff):
+ stop = align_spaces(pw_aff_to_aff(stop), zero)
+
if isinstance(start, int): start = zero + start
if isinstance(stop, int): stop = zero + stop
iname_dt, iname_idx = zero.get_space().get_var_dict()[iname]
iname_aff = zero.add_coefficient(iname_dt, iname_idx, 1)
- return (isl.Set.universe(space)
- # start <= inner
+ result = (isl.Set.universe(space)
+ # start <= iname
.add_constraint(isl.Constraint.inequality_from_aff(
iname_aff - start))
- # inner < stop
+ # iname < stop
.add_constraint(isl.Constraint.inequality_from_aff(
stop-1 - iname_aff)))
+ return result
-def set_is_universe(set):
- bs = set.get_basic_sets()
- if len(bs) == 1:
- return bs[0].is_universe()
- else:
- return isl.Set.universe_like(set).is_subset(set)
-
+def static_extremum_of_pw_aff(pw_aff, constants_only, set_method, what):
+ pieces = pw_aff.get_pieces()
+ if len(pieces) == 1:
+ return pieces[0][1]
+ agg_domain = pw_aff.get_aggregate_domain()
+ for set, candidate_aff in pieces:
+ if constants_only and not candidate_aff.is_cst():
+ continue
-def static_min_of_pw_aff(pw_aff):
- for set, candidate_aff in pw_aff.get_pieces():
- if set_is_universe(candidate_aff.le_set(pw_aff)):
+ if set_method(pw_aff, candidate_aff) == agg_domain:
return candidate_aff
- raise ValueError("a static minimum was not found for PwAff '%s'"
- % pw_aff)
+ raise ValueError("a static %s was not found for PwAff '%s'"
+ % (what, pw_aff))
-def static_max_of_pw_aff(pw_aff):
- for set, candidate_aff in pw_aff.get_pieces():
- if set_is_universe(candidate_aff.ge_set(pw_aff)):
- return candidate_aff
+def static_min_of_pw_aff(pw_aff, constants_only):
+ return static_extremum_of_pw_aff(pw_aff, constants_only, isl.PwAff.ge_set,
+ "minimum")
- raise ValueError("a static maximum was not found for PwAff '%s'"
- % pw_aff)
+def static_max_of_pw_aff(pw_aff, constants_only):
+ return static_extremum_of_pw_aff(pw_aff, constants_only, isl.PwAff.le_set,
+ "maximum")
diff --git a/loopy/kernel.py b/loopy/kernel.py
index 1dcfd43cd931cb9ae267ef01187f4ed91e14e2e7..762a4c4b6777e0e40f40d64552fdb1b9a72b779a 100644
--- a/loopy/kernel.py
+++ b/loopy/kernel.py
@@ -468,16 +468,17 @@ class LoopKernel(Record):
# }}}
if assumptions is None:
- assumptions = isl.Set.universe(domain.get_space())
+ assumptions_space = domain.get_space()
+ assumptions_space = assumptions_space.remove_dims(
+ dim_type.set, 0, assumptions_space.dim(dim_type.set))
+ assumptions = isl.Set.universe(assumptions_space)
elif isinstance(assumptions, str):
s = domain.get_space()
assumptions = isl.BasicSet.read_from_str(domain.get_ctx(),
- "[%s] -> {[%s]: %s}"
- % (",".join(s.get_name(dim_type.param, i)
- for i in range(s.size(dim_type.param))),
- ",".join(s.get_name(dim_type.set, i)
- for i in range(s.size(dim_type.set))),
- assumptions))
+ "[%s] -> { : %s}"
+ % (",".join(s.get_dim_name(dim_type.param, i)
+ for i in range(s.dim(dim_type.param))),
+ assumptions))
Record.__init__(self,
device=device, domain=domain, instructions=insns,
@@ -548,8 +549,8 @@ class LoopKernel(Record):
if self.args is None:
return []
else:
- loop_arg_names = [self.space.get_name(dim_type.param, i)
- for i in range(self.space.size(dim_type.param))]
+ loop_arg_names = [self.space.get_dim_name(dim_type.param, i)
+ for i in range(self.space.dim(dim_type.param))]
return [arg.name for arg in self.args if isinstance(arg, ScalarArg)
if arg.name in loop_arg_names]
@@ -565,22 +566,26 @@ class LoopKernel(Record):
@memoize_method
def get_iname_bounds(self, iname):
lower_bound_pw_aff = (self.domain
+ .intersect(self.assumptions)
.dim_min(self.iname_to_dim[iname][1])
.coalesce())
upper_bound_pw_aff = (self.domain
+ .intersect(self.assumptions)
.dim_max(self.iname_to_dim[iname][1])
.coalesce())
class BoundsRecord(Record):
pass
- size = upper_bound_pw_aff - lower_bound_pw_aff + 1
+ size = (upper_bound_pw_aff - lower_bound_pw_aff + 1)
+ size = size.intersect_domain(self.assumptions)
return BoundsRecord(
lower_bound_pw_aff=lower_bound_pw_aff,
upper_bound_pw_aff=upper_bound_pw_aff,
size=size)
+ @memoize_method
def get_grid_sizes(self):
all_inames_by_insns = set()
for insn in self.instructions:
@@ -605,30 +610,28 @@ class LoopKernel(Record):
elif isinstance(tag, TAG_LOCAL_IDX):
tgt_dict = local_sizes
elif isinstance(tag, TAG_AUTO_LOCAL_IDX):
- #raise RuntimeError("cannot find grid sizes if AUTO_LOCAL_IDX tags are "
- #"present")
- pass
- tgt_dict = None
+ raise RuntimeError("cannot find grid sizes if AUTO_LOCAL_IDX tags are "
+ "present")
else:
tgt_dict = None
if tgt_dict is None:
continue
- bounds = self.get_iname_bounds(iname)
+ size = self.get_iname_bounds(iname).size
- size = bounds.size
+ if tag.axis in tgt_dict:
+ size = tgt_dict[tag.axis].max(size)
from loopy.isl import static_max_of_pw_aff
try:
- size = static_max_of_pw_aff(size)
+ # insist block size is constant
+ size = static_max_of_pw_aff(size,
+ constants_only=isinstance(tag, TAG_LOCAL_IDX))
except ValueError:
pass
- if tag.axis in tgt_dict:
- tgt_dict[tag.axis] = tgt_dict[tag.axis].max(size)
- else:
- tgt_dict[tag.axis] = size
+ tgt_dict[tag.axis] = size
max_dims = self.device.max_work_item_dimensions
@@ -655,6 +658,15 @@ class LoopKernel(Record):
return (to_dim_tuple(global_sizes, "global"),
to_dim_tuple(local_sizes, "local"))
+ def get_grid_sizes_as_exprs(self):
+ grid_size, group_size = self.get_grid_sizes()
+
+ def tup_to_exprs(tup):
+ from loopy.symbolic import pw_aff_to_expr
+ return tuple(pw_aff_to_expr(i) for i in tup)
+
+ return tup_to_exprs(grid_size), tup_to_exprs(group_size)
+
def local_mem_use(self):
return sum(lv.nbytes for lv in self.temporary_variables.itervalues()
if lv.is_local)
diff --git a/loopy/schedule.py b/loopy/schedule.py
index 7371b5ae8fb9f53c67fda2ec7d46d99577900042..f761bc1c95fcf3e2271d5eb9021881ce13643513 100644
--- a/loopy/schedule.py
+++ b/loopy/schedule.py
@@ -26,6 +26,80 @@ class Barrier(Record):
+def realize_reduction(kernel, inames=None, reduction_tag=None):
+ new_insns = []
+ new_temporary_variables = kernel.temporary_variables.copy()
+
+ def map_reduction(expr, rec):
+ sub_expr = rec(expr.expr)
+
+ if reduction_tag is not None and expr.tag != reduction_tag:
+ return
+
+ if inames is not None and set(inames) != set(expr.inames):
+ return
+
+ from pymbolic import var
+
+ target_var_name = kernel.make_unique_var_name("acc",
+ extra_used_vars=set(tv for tv in new_temporary_variables))
+ target_var = var(target_var_name)
+
+ from loopy.kernel import Instruction
+
+ from loopy.kernel import TemporaryVariable
+ new_temporary_variables[target_var_name] = TemporaryVariable(
+ name=target_var_name,
+ dtype=expr.operation.dtype,
+ shape=(),
+ is_local=False)
+
+ init_insn = Instruction(
+ id=kernel.make_unique_instruction_id(
+ extra_used_ids=set(ni.id for ni in new_insns)),
+ assignee=target_var,
+ forced_iname_deps=list(insn.all_inames() - set(expr.inames)),
+ expression=expr.operation.neutral_element)
+
+ new_insns.append(init_insn)
+
+ reduction_insn = Instruction(
+ id=kernel.make_unique_instruction_id(
+ extra_used_ids=set(ni.id for ni in new_insns)),
+ assignee=target_var,
+ expression=expr.operation(target_var, sub_expr),
+ insn_deps=[init_insn.id],
+ forced_iname_deps=list(insn.all_inames()))
+
+ new_insns.append(reduction_insn)
+
+ new_insn_insn_deps.append(reduction_insn.id)
+ new_insn_removed_inames.extend(expr.inames)
+
+ return target_var
+
+ from loopy.symbolic import ReductionCallbackMapper
+ cb_mapper = ReductionCallbackMapper(map_reduction)
+
+ for insn in kernel.instructions:
+ new_insn_insn_deps = []
+ new_insn_removed_inames = []
+
+ new_expression = cb_mapper(insn.expression)
+
+ new_insns.append(
+ insn.copy(
+ expression=new_expression,
+ insn_deps=insn.insn_deps
+ + new_insn_insn_deps))
+
+ return kernel.copy(
+ instructions=new_insns,
+ temporary_variables=new_temporary_variables)
+
+
+
+
def check_double_use_of_hw_dimensions(kernel):
from loopy.kernel import UniqueTag
@@ -602,7 +676,6 @@ def insert_parallel_dim_check_points(kernel, schedule):
def generate_loop_schedules(kernel):
- from loopy import realize_reduction
kernel = realize_reduction(kernel)
check_double_use_of_hw_dimensions(kernel)
diff --git a/test/test_matmul.py b/test/test_matmul.py
index 4c51518b2e61e6a8612061e57af0510426b7ef63..13a81e0abfab0b12c5d6f5da1ae8e421b7b96ca6 100644
--- a/test/test_matmul.py
+++ b/test/test_matmul.py
@@ -211,7 +211,7 @@ def test_plain_matrix_mul_new_ui(ctx_factory):
lp.ArrayArg("c", dtype, shape=(n, n), order=order),
lp.ScalarArg("n", np.int32, approximately=n),
],
- name="matmul")
+ name="matmul", assumptions="n >= 1")
knl = lp.split_dimension(knl, "i", 16,
outer_tag="g.0", inner_tag="l.1", no_slabs=True)
@@ -222,12 +222,8 @@ def test_plain_matrix_mul_new_ui(ctx_factory):
knl = lp.realize_cse(knl, "lhsmat", dtype, ["k_inner", "i_inner"])
knl = lp.realize_cse(knl, "rhsmat", dtype, ["j_inner", "k_inner"])
- #print
- #for insn in knl.instructions:
- #print insn
- #assert lp.get_problems(knl, {})[0] <= 2
-
kernel_gen = lp.generate_loop_schedules(knl)
+ kernel_gen = lp.check_kernels(kernel_gen, dict(n=n), kill_level_min=6)
a = make_well_conditioned_dev_matrix(queue, n, dtype=dtype, order=order)
b = make_well_conditioned_dev_matrix(queue, n, dtype=dtype, order=order)
@@ -235,7 +231,7 @@ def test_plain_matrix_mul_new_ui(ctx_factory):
refsol = np.dot(a.get(), b.get())
def launcher(kernel, gsize, lsize, check):
- evt = kernel(queue, gsize(), lsize(), a.data, b.data, c.data,
+ evt = kernel(queue, gsize(n), lsize(n), a.data, b.data, c.data, n,
g_times_l=True)
if check: