[Lazy evaluation] Pytato Array Context with transformations by kaushikcfd · Pull Request #248 · inducer/meshmode

kaushikcfd · 2021-07-06T15:31:15Z

Draft because:

Implement a global reduction parallelization strategy
Tighter write race checking loopy#224
Implements a utility to get the index of ReturnFromKernel at a point in the linearized kernel loopy#476
Defines PytatoPyOpenCLArrayContext.transform_dag arraycontext#68
Show a DAG manipulating example: materialization pytato#127

Related/linked PRs:

kaushikcfd · 2021-07-06T18:19:52Z

meshmode/array_context.py


+
+@for_each_kernel
+def _single_grid_work_group_transform(kernel, cl_device):


@inducer said in a personal meeting that this logic should also handle reductions kernels.

inducer

Thanks! A few quick drive-by thoughts below.

inducer · 2021-07-06T23:56:14Z

meshmode/array_context.py

+                kernel = lp.split_iname(kernel, iname,
+                                        ngroups * l_zero_size * l_one_size)
+                kernel = lp.split_iname(kernel, f"{iname}_inner",
+                                        l_zero_size, inner_tag="l.0")
+                kernel = lp.split_iname(kernel, f"{iname}_inner_outer",
+                                        l_one_size, inner_tag="l.1",
+                                        outer_tag="g.0")


Doesn't this assume that each iname is only used by exactly one statement?

inducer · 2021-07-06T23:56:26Z

meshmode/array_context.py

+            if len(insn.within_inames) == 0:
+                continue
+
+            if len(insn.within_inames) == 1:


Which type of kernels are those?

I would like it if this were more guided by metadata, along the lines of what the pyopencl actx does.

We could borrow the unifier from FusionContractorArrayContext, but that would add a dependency on #284 and inducer/pytato#224. If those go in before this PR, I will put that here.

inducer · 2021-07-12T21:08:06Z

meshmode/array_context.py

        t_unit = _single_grid_work_group_transform(t_unit, self.queue.device)
        t_unit = lp.set_options(t_unit, "insert_gbarriers")
+        t_unit = lp.linearize(lp.preprocess_kernel(t_unit))
+        t_unit = _alias_global_temporaries(t_unit)


My recollection of our discussion was that we'd do this without aliasing... am I remembering wrong? If not, what made you change your mind?

I thought we wouldn't alias at the pytato's CodeGenMapper stage, but post-linearization grabbing hold of dead-temporaries would be trivial. I.e. I had thought we would alias the global temporaries not at the pytato-level but downstream as loopy transformation.

x-ref: inducer/pytato#122

matthiasdiener · 2021-08-01T18:16:56Z

When trying this on the mirgecom examples, I'm getting this type of error:

  File "/Users/mdiener/Work/emirge/grudge/grudge/discretization.py", line 623, in nodes
    return self.discr_from_dd(dd).nodes()
  File "/Users/mdiener/Work/emirge/meshmode/meshmode/discretization/__init__.py", line 610, in nodes
    result = make_obj_array([
  File "/Users/mdiener/Work/emirge/meshmode/meshmode/discretization/__init__.py", line 611, in <listcomp>
    _DOFArray(None, tuple([
  File "/Users/mdiener/Work/emirge/meshmode/meshmode/discretization/__init__.py", line 612, in <listcomp>
    actx.freeze(resample_mesh_nodes(grp, iaxis)) for grp in self.groups
  File "/Users/mdiener/Work/emirge/arraycontext/arraycontext/impl/pytato/__init__.py", line 133, in freeze
    pt_prg = pt_prg.with_transformed_program(self.transform_loopy_program)
  File "/Users/mdiener/Work/emirge/pytato/pytato/target/loopy/__init__.py", line 124, in with_transformed_program
    return self.copy(program=f(self.program))
  File "/Users/mdiener/Work/emirge/meshmode/meshmode/array_context.py", line 482, in transform_loopy_program
    t_unit = lp.linearize(lp.preprocess_kernel(t_unit))
  File "/Users/mdiener/Work/emirge/loopy/loopy/schedule/__init__.py", line 2201, in linearize
    knl = get_one_linearized_kernel(knl,
  File "/Users/mdiener/Work/emirge/loopy/loopy/schedule/__init__.py", line 2178, in get_one_linearized_kernel
    result = _get_one_scheduled_kernel_inner(kernel,
  File "/Users/mdiener/Work/emirge/loopy/loopy/schedule/__init__.py", line 2146, in _get_one_scheduled_kernel_inner
    return next(iter(generate_loop_schedules(kernel, callables_table)))
  File "/Users/mdiener/Work/emirge/loopy/loopy/schedule/__init__.py", line 1953, in generate_loop_schedules
    yield from generate_loop_schedules_inner(kernel,
  File "/Users/mdiener/Work/emirge/loopy/loopy/schedule/__init__.py", line 2084, in generate_loop_schedules_inner
    gen_sched = insert_barriers(kernel, gen_sched,
  File "/Users/mdiener/Work/emirge/loopy/loopy/schedule/__init__.py", line 1919, in insert_barriers
    result = insert_barriers_at_outer_level(result)
  File "/Users/mdiener/Work/emirge/loopy/loopy/schedule/__init__.py", line 1838, in insert_barriers_at_outer_level
    append_barrier_or_raise_error(
  File "/Users/mdiener/Work/emirge/loopy/loopy/schedule/__init__.py", line 1760, in append_barrier_or_raise_error
    raise MissingBarrierError(
loopy.diagnostic.MissingBarrierError: _pt_kernel: Dependency '_pt_out_store depends on call_lp_nodes' (for variable '_pt_temp') requires synchronization by a global barrier (add a 'no_sync_with' instruction option to state that no synchronization is needed)

kaushikcfd · 2021-08-01T21:55:52Z

@matthiasdiener: Sorry, the transformations in this branch depend on some PRs on the pytato/loopy/arraycontext end. I've edited the description to record that.

kaushikcfd · 2023-01-24T05:01:41Z

Closing this as inducer/arraycontext#216 provides a cleaner implementation.

kaushikcfd commented Jul 6, 2021

View reviewed changes

inducer reviewed Jul 6, 2021

View reviewed changes

inducer reviewed Jul 12, 2021

View reviewed changes

kaushikcfd force-pushed the pytato-array-context-transforms branch from 32a7d59 to bd964da Compare August 6, 2021 22:44

kaushikcfd force-pushed the pytato-array-context-transforms branch from c166b2f to 0c55627 Compare August 17, 2021 20:10

kaushikcfd force-pushed the pytato-array-context-transforms branch 2 times, most recently from 9fb8370 to 8c65687 Compare September 4, 2021 16:10

thomasgibson changed the title ~~Pytato Array Context with transformations~~ [Lazy evaluation] Pytato Array Context with transformations Oct 18, 2021

This was referenced Oct 18, 2021

[Lazy evaluation] Pytato array context transforms inducer/pytato#169

Closed

[Lazy evaluation] Pytato array context transforms inducer/arraycontext#104

Closed

thomasgibson added the lazy evaluation Anything related to lazy evaluation label Oct 19, 2021

kaushikcfd force-pushed the pytato-array-context-transforms branch from 824caee to 405bb8b Compare January 3, 2022 18:42

thomasgibson mentioned this pull request Jan 14, 2022

Entropy stable DG and flux-differencing inducer/grudge#214

Closed

2 tasks

kaushikcfd force-pushed the pytato-array-context-transforms branch 4 times, most recently from 32b0a78 to 0dcc37c Compare March 12, 2022 17:19

kaushikcfd added 3 commits March 12, 2022 12:27

simple_dg: use a single actx

c644f0b

implements SingleGridPytatoArrayContext

a6ecc4d

REVERT BEFORE MERGE: Switch to the right loopy branch for CI

6bfe6af

kaushikcfd force-pushed the pytato-array-context-transforms branch from 0dcc37c to 6bfe6af Compare March 12, 2022 18:27

inducer mentioned this pull request Jun 10, 2022

Remove FirstAxisIsElementsTag and related transform infrastructure #333

Open

kaushikcfd closed this Jan 24, 2023

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[Lazy evaluation] Pytato Array Context with transformations#248

[Lazy evaluation] Pytato Array Context with transformations#248
kaushikcfd wants to merge 3 commits intoinducer:mainfrom
kaushikcfd:pytato-array-context-transforms

kaushikcfd commented Jul 6, 2021 •

edited by inducer

Loading

Uh oh!

kaushikcfd Jul 6, 2021

Uh oh!

inducer left a comment

Uh oh!

inducer Jul 6, 2021

Uh oh!

inducer Jul 6, 2021

Uh oh!

kaushikcfd Mar 12, 2022 •

edited

Loading

Uh oh!

inducer Jul 12, 2021

Uh oh!

kaushikcfd Jul 12, 2021

Uh oh!

kaushikcfd Jul 12, 2021

Uh oh!

matthiasdiener commented Aug 1, 2021

Uh oh!

kaushikcfd commented Aug 1, 2021

Uh oh!

kaushikcfd commented Jan 24, 2023

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants



		@for_each_kernel
		def _single_grid_work_group_transform(kernel, cl_device):

Conversation

kaushikcfd commented Jul 6, 2021 • edited by inducer Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

kaushikcfd Jul 6, 2021

Choose a reason for hiding this comment

Uh oh!

inducer left a comment

Choose a reason for hiding this comment

Uh oh!

inducer Jul 6, 2021

Choose a reason for hiding this comment

Uh oh!

inducer Jul 6, 2021

Choose a reason for hiding this comment

Uh oh!

kaushikcfd Mar 12, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

inducer Jul 12, 2021

Choose a reason for hiding this comment

Uh oh!

kaushikcfd Jul 12, 2021

Choose a reason for hiding this comment

Uh oh!

kaushikcfd Jul 12, 2021

Choose a reason for hiding this comment

Uh oh!

matthiasdiener commented Aug 1, 2021

Uh oh!

kaushikcfd commented Aug 1, 2021

Uh oh!

kaushikcfd commented Jan 24, 2023

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

kaushikcfd commented Jul 6, 2021 •

edited by inducer

Loading

kaushikcfd Mar 12, 2022 •

edited

Loading