source tests/nvidia/cuda/fixtures/schedule/plain/gemm_8192x17408x5120_cta_grid.py
target nvidia.h200_sxm
  at:32  tf.reshard  per cta  x=Tensor[(8192, 64), "bf16", ((8192, 64), (5120, 1))]  result=Tensor[(8192, 64), "bf16",
    ShardLayout(
        layout=Layout((64, 128, 64), (0, 64, 1)),
        attrs=(S(0), B()),
        mesh=Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")),
    ), "smem"]
    refused     T.copy_async
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout((*execution_shape,), (*execution_strides,))))
    refused     T.copy_async_tensor
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout(((*_, 32),), ((*_, 1),))))
    refused     T.copy_reduce_async_tensor
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout(((*_, 32),), ((*_, 1),))))
    refused     T.copy
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout((*execution_shape,), (*execution_strides,))))
    refused     T.ldmatrix
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout(((4, 8),), ((1, 4),))))
    refused     T.stmatrix
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout(((*g, 8, 4),), ((*gs, 4, 1),))))
  bt:33  tf.reshard  per cta  x=Tensor[(64, 17408), "bf16"]  result=Tensor[(64, 17408), "bf16",
    ShardLayout(
        layout=Layout((64, 68, 256), (256, 0, 1)),
        attrs=(B(), S(1)),
        mesh=Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")),
    ), "smem"]
    refused     T.copy_async
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout((*execution_shape,), (*execution_strides,))))
    refused     T.copy_async_tensor
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout(((*_, 32),), ((*_, 1),))))
    refused     T.copy_reduce_async_tensor
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout(((*_, 32),), ((*_, 1),))))
    refused     T.copy
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout((*execution_shape,), (*execution_strides,))))
    refused     T.ldmatrix
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout(((4, 8),), ((1, 4),))))
    refused     T.stmatrix
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout(((*g, 8, 4),), ((*gs, 4, 1),))))
  part:34  tf.matmul  per cta  lhs=Tensor[(8192, 64), "bf16",
    ShardLayout(
        layout=Layout((64, 128, 64), (0, 64, 1)),
        attrs=(S(0), B()),
        mesh=Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")),
    ), "smem"]  rhs=Tensor[(64, 17408), "bf16",
    ShardLayout(
        layout=Layout((64, 68, 256), (256, 0, 1)),
        attrs=(B(), S(1)),
        mesh=Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")),
    ), "smem"]  result=Tensor[(8192, 17408), "f32",
    ShardLayout(
        layout=Layout((64, 128, 68, 256), (0, 256, 0, 1)),
        attrs=(S(0), S(2)),
        mesh=Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")),
    ), "rmem"]
    refused     T.cuda.sm90.Wgmma  dtype=f16, form=SS, a_major=MN, b_major=MN
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout(((*g, 4, 8, 4),), ((*gs, 32, 4, 1),))))
    refused     T.cuda.sm80.Mma
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout(((*g, 4, 8),), ((*gs, 1, 4),))))
    refused     T.cuda.sm89.Mma
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout(((*g, 4, 8),), ((*gs, 1, 4),))))
  acc:35  tf.binary  per cta  lhs=Tensor[(8192, 17408), "f32",
    ShardLayout(
        layout=Layout((64, 128, 68, 256), (0, 256, 0, 1)),
        attrs=(S(0), S(2)),
        mesh=Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")),
    ), "rmem"]  rhs=Tensor[(8192, 17408), "f32",
    ShardLayout(
        layout=Layout((64, 128, 68, 256), (0, 256, 0, 1)),
        attrs=(S(0), S(2)),
        mesh=Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")),
    ), "rmem"]  result=Tensor[(8192, 17408), "f32",
    ShardLayout(
        layout=Layout((64, 128, 68, 256), (0, 256, 0, 1)),
        attrs=(S(0), S(2)),
        mesh=Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")),
    ), "rmem"]
    refused     T.binary
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout((*execution_shape,), (*execution_strides,))))
  <value>:36  tf.cast  per cta  x=Tensor[(8192, 17408), "f32",
    ShardLayout(
        layout=Layout((64, 128, 68, 256), (0, 256, 0, 1)),
        attrs=(S(0), S(2)),
        mesh=Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")),
    ), "rmem"]  result=Tensor[(8192, 17408), "bf16",
    ShardLayout(
        layout=Layout((64, 128, 68, 256), (0, 256, 0, 1)),
        attrs=(S(0), S(2)),
        mesh=Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")),
    ), "rmem"]
    refused     T.cast  cast_dtype=bf16
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout((*execution_shape,), (*execution_strides,))))
  <value>:36  tf.reshard  per cta  x=Tensor[(8192, 17408), "bf16",
    ShardLayout(
        layout=Layout((64, 128, 68, 256), (0, 256, 0, 1)),
        attrs=(S(0), S(2)),
        mesh=Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")),
    ), "rmem"]  result=Tensor[(8192, 17408), "bf16"]
    refused     T.copy_async
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout((*execution_shape,), (*execution_strides,))))
    refused     T.copy_async_tensor
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout(((*_, 32),), ((*_, 1),))))
    refused     T.copy_reduce_async_tensor
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout(((*_, 32),), ((*_, 1),))))
    refused     T.copy
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout((*execution_shape,), (*execution_strides,))))
    refused     T.ldmatrix
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout(((4, 8),), ((1, 4),))))
    refused     T.stmatrix
                  mesh: Mesh((Topology("cta", 4352),), layout=(64, 68), names=("bm", "bn")) does not match Mesh(('thread',), ComposedLayout(None, p0, Layout(((*g, 8, 4),), ((*gs, 4, 1),))))
