Always treat all mesh axes controlled by xmap as MANUAL

PiperOrigin-RevId: 430192736

jax/experimental/maps.py

@@ -731,10 +731,12 @@ def make_xmap_callable(fun: lu.WrappedFun,
         for ax, av, ips in safe_zip(mesh_in_axes, in_avals, in_positional_semantics)
     ]
     in_is_gda = [ips == _PositionalSemantics.GLOBAL for ips in in_positional_semantics]
+    tiling_method: pxla.TilingMethod
     if config.experimental_xmap_spmd_lowering_manual:
-      tiling_method = pxla.TilingMethod.MANUAL
+      manual_mesh_axes = frozenset(it.chain.from_iterable(plan.physical_axis_resources.values()))
+      tiling_method = pxla.TileManual(manual_mesh_axes)
     else:
-      tiling_method = pxla.TilingMethod.VECTORIZE
+      tiling_method = pxla.TileVectorize()
     return pxla.lower_mesh_computation(
         f, name, mesh,
         mesh_in_axes, mesh_out_axes, donated_invars,
@@ -1519,6 +1521,7 @@ def _xmap_lowering_rule_spmd_manual(ctx, *global_in_nodes,
   xla.check_backend_matches(backend, ctx.module_context.platform)
   plan = EvaluationPlan.from_axis_resources(
       axis_resources, resource_env, global_axis_sizes, in_positional_semantics)
+  manual_mesh_axes = frozenset(it.chain.from_iterable(plan.physical_axis_resources.values()))
 
   resource_call_jaxpr = plan.subst_axes_with_resources(call_jaxpr)
   f = lu.wrap_init(core.jaxpr_as_fun(core.ClosedJaxpr(resource_call_jaxpr, ())))
@@ -1528,7 +1531,7 @@ def _xmap_lowering_rule_spmd_manual(ctx, *global_in_nodes,
   # NOTE: Sharding constraints are handled entirely by vtile_manual!
   mesh_in_axes, mesh_out_axes = plan.to_mesh_axes(in_axes, out_axes)
   mesh = resource_env.physical_mesh
-  f = pxla.vtile_manual(f, mesh, mesh_in_axes, mesh_out_axes)
+  f = pxla.vtile_manual(f, tuple(manual_mesh_axes), mesh, mesh_in_axes, mesh_out_axes)
 
   # NOTE: We don't extend the resource env with the mesh shape, because those
   #       resources are already in scope! It's the outermost xmap that introduces
@@ -1539,7 +1542,6 @@ def _xmap_lowering_rule_spmd_manual(ctx, *global_in_nodes,
 
   # We in-line here rather than generating a Call HLO as in the xla_call
   # translation rule just because the extra tuple stuff is a pain.
-  manual_mesh_axes = frozenset(it.chain.from_iterable(plan.physical_axis_resources.values()))
   assert isinstance(ctx.module_context.axis_context, mlir.SPMDAxisContext)
   sub_ctx = ctx.module_context.replace(
       name_stack=xla.extend_name_stack(ctx.module_context.name_stack,

jax/interpreters/pxla.py

@@ -37,9 +37,8 @@ from functools import partial
 import itertools as it
 import operator as op
 import threading
-from typing import (Any, Callable, Dict, List, NamedTuple, Optional,
+from typing import (Any, Callable, Dict, List, NamedTuple, Optional, FrozenSet,
                     Sequence, Set, Tuple, Type, Union, Iterable, Mapping, cast)
-import enum
 import sys
 
 from absl import logging
@@ -2047,11 +2046,11 @@ def vtile_by_mesh(fun: lu.WrappedFun,
 full_to_shard_p = core.Primitive('full_to_shard')
 
 @full_to_shard_p.def_abstract_eval
-def _full_to_shard_abstract_eval(x, axes, mesh):
+def _full_to_shard_abstract_eval(x, axes, mesh, **_):
   # TODO: Assert x is a global aval! Or ideally check that it's global in dims from axes!
   return tile_aval_nd(mesh.shape, axes, x)
 
-def _manual_proto(aval, axes, mesh):
+def _manual_proto(aval: core.ShapedArray, manual_axes_set: FrozenSet[MeshAxisName], mesh: Mesh):
   """Create an OpSharding proto that declares all mesh axes from `axes` as manual
   and all others as replicated.
   """
@@ -2059,8 +2058,8 @@ def _manual_proto(aval, axes, mesh):
   mesh_shape = list(named_mesh_shape.values())
   axis_order = {axis: i for i, axis in enumerate(mesh.axis_names)}
 
-  manual_axes = list(axes)
-  replicated_axes = list(axis for axis in mesh.axis_names if axis not in axes)
+  manual_axes = list(sorted(manual_axes_set, key=str))
+  replicated_axes = list(axis for axis in mesh.axis_names if axis not in manual_axes_set)
 
   tad_perm = ([axis_order[a] for a in replicated_axes] +
               [axis_order[a] for a in manual_axes])
@@ -2077,29 +2076,29 @@ def _manual_proto(aval, axes, mesh):
   return proto
 
 @partial(mlir.register_lowering, full_to_shard_p)
-def _full_to_shard_lowering(ctx, x, *, axes: ArrayMapping, mesh: Mesh):
+def _full_to_shard_lowering(ctx, x, *, axes: ArrayMapping, mesh: Mesh, manual_axes: FrozenSet[MeshAxisName]):
   # TODO: Can we short-circuit for replicated values? Probably not.
   aval_in, = ctx.avals_in
   aval_out, = ctx.avals_out
   sharding_proto = mesh_sharding_specs(mesh.shape, mesh.axis_names)(aval_in, axes).sharding_proto()
   unspecified_dims = set(range(aval_in.ndim)) - set(axes.values())
   sx = mlir.wrap_with_sharding_op(x, sharding_proto, unspecified_dims=unspecified_dims)
-  manual_proto = _manual_proto(aval_in, axes, mesh)
+  manual_proto = _manual_proto(aval_in, manual_axes, mesh)
   result_type, = mlir.aval_to_ir_types(aval_out)
   return mlir.wrap_with_full_to_shard_op(result_type, sx, manual_proto, unspecified_dims=unspecified_dims),
 
 shard_to_full_p = core.Primitive('shard_to_full')
 
 @shard_to_full_p.def_abstract_eval
-def _shard_to_full_abstract_eval(x, axes, mesh):
+def _shard_to_full_abstract_eval(x, axes, mesh, **_):
   # TODO: Assert x is a global aval! Or ideally check that it's global in dims from axes!
   return untile_aval_nd(mesh.shape, axes, x)
 
 @partial(mlir.register_lowering, shard_to_full_p)
-def _shard_to_full_lowering(ctx, x, *, axes: ArrayMapping, mesh: Mesh):
+def _shard_to_full_lowering(ctx, x, *, axes: ArrayMapping, mesh: Mesh, manual_axes: FrozenSet[MeshAxisName]):
   aval_in, = ctx.avals_in
   aval_out, = ctx.avals_out
-  manual_proto = _manual_proto(aval_in, axes, mesh)
+  manual_proto = _manual_proto(aval_in, manual_axes, mesh)
   result_type, = mlir.aval_to_ir_types(aval_out)
   unspecified_dims = set(range(aval_in.ndim)) - set(axes.values())
   sx = mlir.wrap_with_sharding_op(x, manual_proto, unspecified_dims=unspecified_dims)
@@ -2107,18 +2106,29 @@ def _shard_to_full_lowering(ctx, x, *, axes: ArrayMapping, mesh: Mesh):
   return mlir.wrap_with_shard_to_full_op(result_type, sx, sharding_proto, unspecified_dims),
 
 @lu.transformation
-def vtile_manual(mesh: Mesh,
+def vtile_manual(manual_axes: FrozenSet[MeshAxisName],
+                 mesh: Mesh,
                  in_axes: Sequence[ArrayMapping],
                  out_axes: Sequence[ArrayMapping],
                  *args):
-  tiled_args = [full_to_shard_p.bind(arg, axes=axes, mesh=mesh)
+  tiled_args = [full_to_shard_p.bind(arg, axes=axes, mesh=mesh, manual_axes=manual_axes)
                 for arg, axes in zip(args, in_axes)]
   tiled_outs = yield tiled_args, {}
-  outs = [shard_to_full_p.bind(out, axes=axes, mesh=mesh)
+  outs = [shard_to_full_p.bind(out, axes=axes, mesh=mesh, manual_axes=manual_axes)
           for out, axes in zip(tiled_outs, out_axes)]
   yield outs
 
-TilingMethod = enum.Enum("TilingMethod", ["VECTORIZE", "MANUAL"])
+
+@dataclasses.dataclass(frozen=True)
+class TileVectorize:
+  pass
+
+@dataclasses.dataclass(frozen=True)
+class TileManual:
+  manual_axes: FrozenSet[MeshAxisName]
+
+TilingMethod = Union[TileVectorize, TileManual]
+
 
 @profiler.annotate_function
 def lower_mesh_computation(
@@ -2152,17 +2162,17 @@ def lower_mesh_computation(
   if spmd_lowering:
     # TODO: Consider handling xmap's 'vectorize' in here. We can vmap once instead of vtile twice!
     if tiling_method is not None:
-      if tiling_method is TilingMethod.VECTORIZE:
+      if isinstance(tiling_method, TileVectorize):
         tiling_transform = vtile_by_mesh
-      elif tiling_method is TilingMethod.MANUAL:
-        tiling_transform = vtile_manual
+      elif isinstance(tiling_method, TileManual):
+        tiling_transform = lambda f, *args: vtile_manual(f, tiling_method.manual_axes, *args)  # type: ignore
       else:
         raise NotImplementedError(f"Unrecognized tiling method: {tiling_method}")
       assert not callable(out_axes)
       fun = tiling_transform(fun, mesh, in_axes, out_axes)
     in_jaxpr_avals = global_in_avals
   else:
-    assert tiling_method is TilingMethod.VECTORIZE
+    assert isinstance(tiling_method, TileVectorize)
     in_jaxpr_avals = in_tiled_avals
   with core.extend_axis_env_nd(mesh.shape.items()):
     with dispatch.log_elapsed_time(f"Finished tracing + transforming {name_stack} "

tests/xmap_test.py

@@ -717,6 +717,14 @@ class XMapTestManualSPMD(ManualSPMDTestMixin, XMapTestCase):
     x = jnp.arange(20, dtype=jnp.float32)
     self.assertAllClose(fx(x), f(x))
 
+  @jtu.with_mesh([('x', 2)])
+  def testReplicated(self):
+    # TODO(apaszke): This seems to be failing if I try to have a replicated and a mapped argument?
+    f = lambda x: jnp.sin(jnp.cos(x) + x) * x
+    fx = xmap(f, in_axes=[...], out_axes=[...], axis_sizes={'i': 4}, axis_resources={'i': 'x'})
+    x = jnp.arange(20, dtype=jnp.float32)
+    self.assertAllClose(fx(x), f(x))
+
   @jtu.with_mesh([('x', 2), ('y', 1)])
   def testInPJit(self):
     f = xmap(lambda x: jnp.sin(x) + x, in_axes=['i'], out_axes=['i'], axis_resources={'i': 'x'})
@@ -724,6 +732,13 @@ class XMapTestManualSPMD(ManualSPMDTestMixin, XMapTestCase):
     x = jnp.arange(20, dtype=jnp.float32)
     self.assertAllClose(h(x), jnp.sin(x * x) + x * x + x)
 
+  @jtu.with_mesh([('x', 2), ('y', 1)])
+  def testInPJitReplicated(self):
+    f = xmap(lambda x: jnp.sin(x) + x, in_axes={}, out_axes={}, axis_sizes={'i': 4}, axis_resources={'i': 'x'})
+    h = pjit(lambda x: f(x * x) + x, in_axis_resources=P('y'), out_axis_resources=None)
+    x = jnp.arange(20, dtype=jnp.float32)
+    self.assertAllClose(h(x), jnp.sin(x * x) + x * x + x)
+
   @jtu.with_mesh([('x', 2), ('y', 1)])
   def testNestedConstraint(self):
     # TODO(b/219691408): Using P('y') instead of P() causes an XLA crash!