Stop using global mesh for custom_partitioning. (#1112)

Signed-off-by: Frederic Bastien <fbastien@nvidia.com>
Co-authored-by: Phuong Nguyen <36155692+phu0ngng@users.noreply.github.com>

transformer_engine/jax/cpp_extensions/activation.py

@@ -467,7 +467,7 @@ class ActLuFp8Primitive(BasePrimitive):
             local_x, local_amax = ActLuFp8Primitive.impl(
                 x, amax, scale, scale_inv, out_dtype=out_dtype, act_enum=act_enum
             )
-            global_updated_amax = all_reduce_max_along_all_axes_except_PP(local_amax)
+            global_updated_amax = all_reduce_max_along_all_axes_except_PP(local_amax, mesh)
 
             return local_x, global_updated_amax
 

transformer_engine/jax/cpp_extensions/attention.py

@@ -1011,7 +1011,7 @@ class FusedAttnBwdPrimitive(BasePrimitive):
             )
             global_dbias = local_dbias
             if attn_bias_type is not NVTE_Bias_Type.NVTE_NO_BIAS:
-                global_dbias = all_reduce_sum_along_dp_fsdp(local_dbias)
+                global_dbias = all_reduce_sum_along_dp_fsdp(local_dbias, mesh)
             return local_dq, local_dk, local_dv, global_dbias
 
         return mesh, sharded_impl, out_shardings, arg_shardings

transformer_engine/jax/cpp_extensions/normalization.py

@@ -533,8 +533,8 @@ class LayerNormBwdPrimitive(BasePrimitive):
             local_dx, local_dgamma, local_dbeta = LayerNormBwdPrimitive.impl(
                 dz, x, mu, rsigma, gamma, zero_centered_gamma=zero_centered_gamma, epsilon=epsilon
             )
-            global_dgamma = all_reduce_sum_along_dp_fsdp(local_dgamma)
-            global_dbeta = all_reduce_sum_along_dp_fsdp(local_dbeta)
+            global_dgamma = all_reduce_sum_along_dp_fsdp(local_dgamma, mesh)
+            global_dbeta = all_reduce_sum_along_dp_fsdp(local_dbeta, mesh)
             return local_dx, global_dgamma, global_dbeta
 
         return mesh, sharded_impl, out_shardings, arg_shardings
@@ -935,7 +935,7 @@ class RmsNormBwdPrimitive(BasePrimitive):
 
         def sharded_impl(dz, x, rsigma, gamma):
             local_dx, local_dgamma = RmsNormBwdPrimitive.impl(dz, x, rsigma, gamma, epsilon=epsilon)
-            global_dgamma = all_reduce_sum_along_dp_fsdp(local_dgamma)
+            global_dgamma = all_reduce_sum_along_dp_fsdp(local_dgamma, mesh)
             return local_dx, global_dgamma
 
         return mesh, sharded_impl, out_shardings, arg_shardings
@@ -1228,7 +1228,7 @@ class LayerNormFwdFp8Primitive(BasePrimitive):
                 zero_centered_gamma=zero_centered_gamma,
                 epsilon=epsilon,
             )
-            global_updated_amax = all_reduce_max_along_all_axes_except_PP(local_amax)
+            global_updated_amax = all_reduce_max_along_all_axes_except_PP(local_amax, mesh)
 
             return local_x, local_mu, local_rsigma, global_updated_amax
 
@@ -1481,7 +1481,7 @@ class RmsNormFwdFp8Primitive(BasePrimitive):
             local_x, local_rsigma, local_amax = RmsNormFwdFp8Primitive.impl(
                 x, gamma, amax, scale, scale_inv, out_dtype=out_dtype, epsilon=epsilon
             )
-            global_updated_amax = all_reduce_max_along_all_axes_except_PP(local_amax)
+            global_updated_amax = all_reduce_max_along_all_axes_except_PP(local_amax, mesh)
 
             return local_x, local_rsigma, global_updated_amax
 

transformer_engine/jax/cpp_extensions/quantization.py

@@ -157,7 +157,7 @@ class CastFP8Primitive(BasePrimitive):
             local_cx, local_updated_amax = CastFP8Primitive.impl(
                 x, amax, scale, scale_inv, out_dtype=out_dtype
             )
-            global_updated_amax = all_reduce_max_along_all_axes_except_PP(local_updated_amax)
+            global_updated_amax = all_reduce_max_along_all_axes_except_PP(local_updated_amax, mesh)
 
             return local_cx, global_updated_amax
 

transformer_engine/jax/cpp_extensions/transpose.py

@@ -390,7 +390,7 @@ class CastTransposePrimitive(BasePrimitive):
                 static_axis_boundary=static_axis_boundary,
                 transpose_axis_boundary=transpose_axis_boundary,
             )
-            global_updated_amax = all_reduce_max_along_all_axes_except_PP(local_updated_amax)
+            global_updated_amax = all_reduce_max_along_all_axes_except_PP(local_updated_amax, mesh)
 
             return local_cx, local_cxt, global_updated_amax
 
@@ -646,8 +646,8 @@ class DBiasCastTransposePrimitive(BasePrimitive):
                 static_axis_boundary=static_axis_boundary,
                 transpose_axis_boundary=transpose_axis_boundary,
             )
-            global_dbias = all_reduce_sum_along_dp_fsdp(local_dbias)
-            global_updated_amax = all_reduce_max_along_all_axes_except_PP(local_amax)
+            global_dbias = all_reduce_sum_along_dp_fsdp(local_dbias, mesh)
+            global_updated_amax = all_reduce_max_along_all_axes_except_PP(local_amax, mesh)
             return local_out, local_t_out, global_dbias, global_updated_amax
 
         return mesh, sharded_impl, out_shardings, arg_shardings
@@ -981,8 +981,8 @@ class DActLuDBiasCastTransposePrimitive(BasePrimitive):
                     act_enum=act_enum,
                 )
             )
-            global_dbias = all_reduce_sum_along_dp_fsdp(local_dbias)
-            global_updated_amax = all_reduce_max_along_all_axes_except_PP(local_amax)
+            global_dbias = all_reduce_sum_along_dp_fsdp(local_dbias, mesh)
+            global_updated_amax = all_reduce_max_along_all_axes_except_PP(local_amax, mesh)
             return local_out, local_t_out, global_dbias, global_updated_amax
 
         return mesh, sharded_impl, out_shardings, arg_shardings
@@ -1225,7 +1225,7 @@ class DgatedActLuCastTransposePrimitive(BasePrimitive):
                 static_axis_boundary=static_axis_boundary,
                 act_enum=act_enum,
             )
-            global_updated_amax = all_reduce_max_along_all_axes_except_PP(local_amax)
+            global_updated_amax = all_reduce_max_along_all_axes_except_PP(local_amax, mesh)
             return local_out, local_t_out, global_updated_amax
 
         return mesh, sharded_impl, out_shardings, arg_shardings

transformer_engine/jax/sharding.py

@@ -30,8 +30,7 @@ W_TP_AXES = "nvte_w_tp"
 W_JOINED_AXES = "nvte_w_joined"
 
 
-def _get_mesh_info(resource: str):
-    mesh = _PXLA_THREAD_RESOURCES.env.physical_mesh
+def _get_mesh_info(resource: str, mesh: jax.sharding.Mesh):
     assert resource in mesh.axis_names, f"{resource} is not in the axis_names of Mesh {mesh}."
     return mesh.shape[resource], resource
 
@@ -132,12 +131,12 @@ def get_padded_spec(spec, ndim):
     return spec + (None,) * (ndim - len(spec))
 
 
-def lax_paral_op(x: jnp.array, ops: Callable, mesh_resource: str):
+def lax_paral_op(x: jnp.array, ops: Callable, mesh_resource: str, mesh: jax.sharding.Mesh):
     """
     A wrapper function to invoke lax.p* operations, like psum.
     """
     if mesh_resource is not None:
-        _, resource = _get_mesh_info(mesh_resource)
+        _, resource = _get_mesh_info(mesh_resource, mesh)
         return ops(x, resource)
     return x
 
@@ -201,22 +200,22 @@ def global_mesh_resource() -> MeshResource:
     return _GLOBAL_MESH_RESOURCE
 
 
-def all_reduce_sum_along_dp_fsdp(x: jnp.array):
+def all_reduce_sum_along_dp_fsdp(x: jnp.array, mesh: jax.sharding.Mesh):
     """
     All-Reduce (Sum) along DP and FSDP mesh axes.
     """
-    x = lax_paral_op(x, jax.lax.psum, global_mesh_resource().dp_resource)
-    return lax_paral_op(x, jax.lax.psum, global_mesh_resource().fsdp_resource)
+    x = lax_paral_op(x, jax.lax.psum, global_mesh_resource().dp_resource, mesh)
+    return lax_paral_op(x, jax.lax.psum, global_mesh_resource().fsdp_resource, mesh)
 
 
-def all_reduce_max_along_all_axes_except_PP(x: jnp.array):
+def all_reduce_max_along_all_axes_except_PP(x: jnp.array, mesh: jax.sharding.Mesh):
     """
     All-Reduce (Max) along all mesh axes.
     """
     all_axes = get_all_mesh_axes()
     for axis in all_axes:
         if axis != global_mesh_resource().pp_resource:
-            x = lax_paral_op(x, jax.lax.pmax, axis)
+            x = lax_paral_op(x, jax.lax.pmax, axis, mesh)
     return x