Merge remote-tracking branch 'origin/main'

Author: connorjward

.basedpyright/baseline.json

@@ -10,7 +10,7 @@ jobs:
         if: startsWith(github.repository, 'inducer/')
         runs-on: ubuntu-latest
         steps:
-        -   uses: actions/checkout@v4
+        -   uses: actions/checkout@v5
         -   run: |
                 curl -L -O https://tiker.net/ci-support-v0
                 . ./ci-support-v0

.github/workflows/autopush.yml

@@ -16,10 +16,10 @@ jobs:
         name: Ruff
         runs-on: ubuntu-latest
         steps:
-        -   uses: actions/checkout@v4
+        -   uses: actions/checkout@v5
             with:
                 submodules: true
-        -   uses: astral-sh/setup-uv@v6
+        -   uses: astral-sh/setup-uv@v7
         -   name: "Main Script"
             run: |
                 uv run --only-dev ruff check
@@ -28,13 +28,13 @@ jobs:
         name: Typos
         runs-on: ubuntu-latest
         steps:
-        -   uses: actions/checkout@v4
+        -   uses: actions/checkout@v5
         -   uses: crate-ci/typos@master
 
     basedpyright:
         runs-on: ubuntu-latest
         steps:
-        -   uses: actions/checkout@v4
+        -   uses: actions/checkout@v5
         -   name: "Main Script"
             run: |
                 EXTRA_INSTALL="pytest types-colorama types-Pygments scipy-stubs"
@@ -54,7 +54,7 @@ jobs:
           matrix:
             os: [ubuntu-latest, macos-latest]
         steps:
-        -   uses: actions/checkout@v4
+        -   uses: actions/checkout@v5
         -   name: "Main Script"
             run: |
                 curl -L -O https://gitlab.tiker.net/inducer/ci-support/raw/main/build-and-test-py-project-within-miniconda.sh
@@ -64,7 +64,7 @@ jobs:
         name: Conda Pytest with Intel CL
         runs-on: ubuntu-latest
         steps:
-        -   uses: actions/checkout@v4
+        -   uses: actions/checkout@v5
         -   name: "Main Script"
             run: |
                 curl -L -O https://raw.githubusercontent.com/illinois-scicomp/machine-shop-maintenance/main/install-intel-icd.sh
@@ -83,7 +83,7 @@ jobs:
         name: Conda Pytest without arg check
         runs-on: ubuntu-latest
         steps:
-        -   uses: actions/checkout@v4
+        -   uses: actions/checkout@v5
         -   name: "Main Script"
             run: |
                 curl -L -O https://gitlab.tiker.net/inducer/ci-support/raw/main/build-and-test-py-project-within-miniconda.sh
@@ -94,7 +94,7 @@ jobs:
         name: Conda Pytest Twice (for cache behavior)
         runs-on: ubuntu-latest
         steps:
-        -   uses: actions/checkout@v4
+        -   uses: actions/checkout@v5
         -   name: "Main Script"
             run: |
                 # This test makes sure that loopy can run with kernels loaded from disk cache.
@@ -112,7 +112,7 @@ jobs:
         name: Conda Examples
         runs-on: ubuntu-latest
         steps:
-        -   uses: actions/checkout@v4
+        -   uses: actions/checkout@v5
         -   name: "Main Script"
             run: |
                 curl -L -O https://tiker.net/ci-support-v0
@@ -130,9 +130,9 @@ jobs:
         name: Documentation
         runs-on: ubuntu-latest
         steps:
-        -   uses: actions/checkout@v4
+        -   uses: actions/checkout@v5
         -
-            uses: actions/setup-python@v5
+            uses: actions/setup-python@v6
             with:
                 python-version: '3.x'
         -   name: "Main Script"
@@ -150,7 +150,7 @@ jobs:
         name: Tests for downstream project ${{ matrix.downstream_project }}
         runs-on: ubuntu-latest
         steps:
-        -   uses: actions/checkout@v4
+        -   uses: actions/checkout@v5
         -   name: "Main Script"
             env:
                 DOWNSTREAM_PROJECT: ${{ matrix.downstream_project }}
@@ -194,15 +194,14 @@ jobs:
                 # patch so exception messages get shown
                 curl -L https://gist.githubusercontent.com/inducer/17d7134ace215f0df1f3627eac4195c7/raw/ec5470a7d8587b6e1f336f3ef1d0ece5e26f236a/firedrake-debug-patch.diff | patch -p1
 
-                sed -i 's/@mpiexec/@mpiexec --oversubscribe/' Makefile
-                make check
+                firedrake-check --mpiexec 'mpiexec --oversubscribe -n'
 
     validate_cff:
             name: Validate CITATION.cff
             runs-on: ubuntu-latest
             steps:
-            -   uses: actions/checkout@v4
-            -   uses: actions/setup-python@v5
+            -   uses: actions/checkout@v5
+            -   uses: actions/setup-python@v6
             -   run: |
                     pip install cffconvert
                     cffconvert -i CITATION.cff --validate

.github/workflows/ci.yml

@@ -58,7 +58,7 @@ def data_gen():
             for _ in f():
                 self.tick()
 
-                for ary, plot in zip(self.arrays, plots):
+                for ary, plot in zip(self.arrays, plots, strict=True):
                     plot.set_array(ary.get_plot_data())
 
                 fig.canvas.draw()
@@ -85,7 +85,7 @@ def __init__(self, ctx, name, shape, strides, elements_per_row=None):
         if elements_per_row is None:
             if len(shape) > 1:
                 minstride = min(strides)
-                for sh_i, st_i in zip(shape, strides):
+                for sh_i, st_i in zip(shape, strides, strict=True):
                     if st_i == minstride:
                         elements_per_row = sh_i
                         break
@@ -119,7 +119,7 @@ def reshape_ind(ind):
 
         lin_index = sum(
                 ind_i * stride_i
-                for ind_i, stride_i in zip(index, self.strides))
+                for ind_i, stride_i in zip(index, self.strides, strict=True))
 
         if not isinstance(lin_index, np.ndarray):
             subscript = [np.newaxis] * self.ctx.ind_length

TODO

@@ -2,8 +2,8 @@
 import numpy.linalg as la
 
 import pyopencl as cl
-import pyopencl.array
-import pyopencl.clrandom
+import pyopencl.array as cla
+import pyopencl.clrandom as clrandom
 
 import loopy as lp
 
@@ -21,11 +21,11 @@ def main():
     queue = cl.CommandQueue(ctx)
 
     n = 2048
-    a = cl.array.empty(queue, (n, n), dtype=np.float64, order="F")
-    b = cl.array.empty(queue, (n, n), dtype=np.float64, order="F")
-    c = cl.array.zeros(queue, (n, n), dtype=np.float64, order="F")
-    cl.clrandom.fill_rand(a)
-    cl.clrandom.fill_rand(b)
+    a = cla.empty(queue, (n, n), dtype=np.float64, order="F")
+    b = cla.empty(queue, (n, n), dtype=np.float64, order="F")
+    c = cla.zeros(queue, (n, n), dtype=np.float64, order="F")
+    clrandom.fill_rand(a)
+    clrandom.fill_rand(b)
 
     dgemm = lp.set_options(dgemm, write_code=True)
 

contrib/mem-pattern-explorer/pattern_vis.py

@@ -1,5 +1,4 @@
 import ctypes
-import ctypes.util
 import os
 from tempfile import TemporaryDirectory
 from time import time

examples/fortran/matmul-driver.py

@@ -580,12 +580,9 @@ def make_copy_kernel(new_dim_tags, old_dim_tags=None):
     command_indices = ", ".join(indices)
     bounds = " and ".join(
             f"0<={ind}<{shape_i}"
-            for ind, shape_i in zip(indices, shape))
+            for ind, shape_i in zip(indices, shape, strict=True))
 
-    set_str = "{{[{}]: {} }}".format(
-                command_indices,
-                bounds
-                )
+    set_str = f"{{[{command_indices}]: {bounds} }}"
     result = make_kernel(set_str,
             "output[%s] = input[%s]"
             % (command_indices, command_indices),
@@ -659,7 +656,7 @@ def make_einsum(spec, arg_names, **knl_creation_kwargs):
     lhs = var("out")[tuple(var(i) for i in out_spec)]
 
     rhs = 1
-    for arg_name, argsp in zip(arg_names, arg_specs):
+    for arg_name, argsp in zip(arg_names, arg_specs, strict=True):
         rhs = rhs * var(arg_name)[tuple(var(i) for i in argsp)]
 
     if sum_indices:
@@ -694,7 +691,7 @@ def set_default_target(target: Callable[[], TargetBase] | None):
 
 def _set_up_default_target():
     try:
-        import pyopencl  # noqa
+        import pyopencl  # noqa: F401
     except ImportError:
         from loopy.target.opencl import OpenCLTarget
         target = OpenCLTarget

examples/python/ispc-stream-harness.py

@@ -158,7 +158,7 @@ def make_ref_args(kernel, queue, parameters):
                 strides = evaluate(get_strides(arg), parameters)
 
                 alloc_size = sum(astrd*(alen-1) if astrd != 0 else alen-1
-                        for alen, astrd in zip(shape, strides)) + 1
+                        for alen, astrd in zip(shape, strides, strict=True)) + 1
 
                 if dtype is None:
                     raise LoopyError("dtype for argument '%s' is not yet "
@@ -229,7 +229,7 @@ def make_args(kernel, queue, ref_arg_data, parameters):
     from loopy.kernel.data import ArrayArg, ConstantArg, ImageArg, ValueArg
 
     args = {}
-    for arg, arg_desc in zip(kernel.args, ref_arg_data):
+    for arg, arg_desc in zip(kernel.args, ref_arg_data, strict=True):
         if isinstance(arg, ValueArg):
             arg_value = parameters[arg.name]
 
@@ -264,7 +264,7 @@ def make_args(kernel, queue, ref_arg_data, parameters):
             numpy_strides = [itemsize*s for s in strides]
 
             alloc_size = sum(astrd*(alen-1) if astrd != 0 else alen-1
-                    for alen, astrd in zip(shape, strides)) + 1
+                    for alen, astrd in zip(shape, strides, strict=True)) + 1
 
             # use contiguous array to transfer to host
             host_ref_contig_array = arg_desc.ref_pre_run_storage_array.get()
@@ -429,8 +429,9 @@ def auto_test_vs_ref(
         raise LoopyError("ref_prog and test_prog do not have the same number "
                 "of arguments")
 
-    for i, (ref_arg, test_arg) in enumerate(zip(ref_prog[ref_entrypoint].args,
-            test_prog[test_entrypoint].args)):
+    for i, (ref_arg, test_arg) in enumerate(
+            zip(ref_prog[ref_entrypoint].args,
+                test_prog[test_entrypoint].args, strict=True)):
         if ref_arg.name != test_arg.name:
             raise LoopyError("ref_prog and test_prog argument lists disagree at "
                     "index %d (1-based)" % (i+1))
@@ -477,8 +478,8 @@ def auto_test_vs_ref(
                 properties=cl.command_queue_properties.PROFILING_ENABLE)
         ref_codegen_result = lp.generate_code_v2(ref_prog)
 
-        logger.info("{} (ref): trying {} for the reference calculation".format(
-            ref_entrypoint, dev))
+        logger.info("%s (ref): trying %s for the reference calculation",
+                    ref_entrypoint, dev)
 
         if not quiet and print_ref_code:
             print(75*"-")
@@ -513,8 +514,8 @@ def auto_test_vs_ref(
 
         ref_queue.finish()
 
-        logger.info("{} (ref): using {} for the reference calculation".format(
-            ref_entrypoint, dev))
+        logger.info("%s (ref): using %s for the reference calculation",
+                    ref_entrypoint, dev)
         logger.info("%s (ref): run" % ref_entrypoint)
 
         ref_start = time()
@@ -665,8 +666,8 @@ def auto_test_vs_ref(
     logger.info("%s: timing run done" % (test_entrypoint))
 
     rates = ""
-    for cnt, lbl in zip(op_count, op_label):
-        rates += " {:g} {}/s".format(cnt/elapsed_wall, lbl)
+    for cnt, lbl in zip(op_count, op_label, strict=True):
+        rates += f" {cnt / elapsed_wall:g} {lbl}/s"
 
     if not quiet:
         def format_float_or_none(v):
@@ -683,8 +684,8 @@ def format_float_or_none(v):
 
     if do_check:
         ref_rates = ""
-        for cnt, lbl in zip(op_count, op_label):
-            rates += " {:g} {}/s".format(cnt/elapsed_wall, lbl)
+        for cnt, lbl in zip(op_count, op_label, strict=True):
+            rates += f" {cnt/elapsed_wall:g} {lbl}/s"
 
         if not quiet:
             print("elapsed: %s s event, %s s marker-event %s s wall "
@@ -695,11 +696,12 @@ def format_float_or_none(v):
 
         if do_check:
             ref_rates = ""
-            for cnt, lbl in zip(op_count, op_label):
-                ref_rates += " {:g} {}/s".format(cnt/ref_elapsed_event, lbl)
+            for cnt, lbl in zip(op_count, op_label, strict=True):
+                ref_rates += f" {cnt/ref_elapsed_wall:g} {lbl}/s"
+
             if not quiet:
-                print("ref: elapsed: {:g} s event, {:g} s wall{}".format(
-                        ref_elapsed_event, ref_elapsed_wall, ref_rates))
+                print(f"ref: elapsed: {ref_elapsed_event:g}s event, "
+                      f"{ref_elapsed_wall:g} s wall{ref_rates}")
 
     # }}}
 

loopy/__init__.py

@@ -339,8 +339,8 @@ def map_subscript(self, expr):
                         "Type inference did not find type of '%s'"
                         % idx)
             if not type_inf_result[0].is_integral():
-                raise LoopyError("Non-integral array indices obtained in"
-                        " {}.".format(expr))
+                raise LoopyError(
+                    f"Non-integral array indices obtained in {expr}")
 
         return self.rec(expr.aggregate)
 
@@ -489,8 +489,8 @@ def check_multiple_tags_allowed(kernel: LoopKernel) -> None:
     for iname in kernel.inames.values():
         for comb in illegal_combinations:
             if len(filter_iname_tags_by_type(iname.tags, comb)) > 1:
-                raise LoopyError("iname {} has illegal combination of "
-                                 "tags: {}".format(iname.name, iname.tags))
+                raise LoopyError(f"iname {iname.name} has illegal combination of "
+                                 f"tags: {iname.tags}")
 
 
 def _check_for_double_use_of_hw_axes_inner(
@@ -612,7 +612,7 @@ def check_for_write_races(kernel: LoopKernel) -> None:
     for insn in kernel.instructions:
         for assignee_name, assignee_indices in zip(
                 insn.assignee_var_names(),
-                insn.assignee_subscript_deps()):
+                insn.assignee_subscript_deps(), strict=True):
             assignee_inames = assignee_indices & kernel.all_inames()
             if not assignee_inames <= insn.within_inames:
                 raise LoopyError(
@@ -1359,7 +1359,7 @@ def check_for_nested_base_storage(kernel: LoopKernel) -> None:
 
     for ary in kernel.temporary_variables.values():
         if ary.base_storage:
-            storage_array = name_to_array.get(ary.base_storage, None)
+            storage_array = name_to_array.get(ary.base_storage)
 
             if storage_array is None:
                 raise LoopyError("Nothing known about storage array "
@@ -1716,16 +1716,16 @@ def _are_sub_array_refs_equivalent(
     from pymbolic.mapper.substitutor import make_subst_func
 
     from loopy.symbolic import SubstitutionMapper, simplify_via_aff
-    subst_func = make_subst_func({iname1.name:  iname2
-                                  for iname1, iname2 in zip(sar1.swept_inames,
-                                                            sar2.swept_inames)
-                                  })
+    subst_func = make_subst_func({
+        iname1.name: iname2
+        for iname1, iname2 in zip(sar1.swept_inames, sar2.swept_inames, strict=True)
+    })
 
     # subst_mapper: maps swept inames from sar1 to sar2
     subst_mapper = SubstitutionMapper(subst_func)
 
     for idx1, idx2 in zip(sar1.subscript.index_tuple,
-                          sar2.subscript.index_tuple):
+                          sar2.subscript.index_tuple, strict=True):
         subst_idx1 = subst_mapper(idx1)
         assert is_arithmetic_expression(subst_idx1)
         assert is_arithmetic_expression(idx2)
@@ -1991,11 +1991,9 @@ def check_implemented_domains(
                         iname, pt.get_coordinate_val(tp, dim).to_python()))
 
                 lines.append(
-                        "sample point in {} but not {}: {}".format(
-                            bigger, smaller, ", ".join(point_axes)))
+                    f"sample point in {bigger} but not {smaller}: {point_axes}")
                 lines.append(
-                        "gist of constraints in {} but not {}: {}".format(
-                            bigger, smaller, gist_domain))
+                    f"gist of constraints in {bigger} but not {smaller}: {gist_domain}")
 
             if code is not None:
                 print(79*"-")