Indexing failed on a simple load.

import torch
from nvfuser_direct import FusionDefinition, DataType
m = 1024
k = 1024 * 16

def nvfuser_fusion_id1(fd : FusionDefinition) -> None :
  T0 = fd.define_tensor(shape=[-1, -1], contiguity=[True, True], dtype=DataType.Float, is_cpu=False, stride_order=[1, 0])
  m_size = fd.ops.size(T0, 0);
  k_size = fd.ops.size(T0, 1);
  k_tile_size = fd.ops.div(k_size, 16);
  reshaped_mat1 = fd.ops.reshape(T0, [m, k//16, 16])
  scale1 = fd.ops.max(reshaped_mat1, 2)
  broadcast_scale1 = fd.ops.broadcast(scale1, [False, False, True])
  reshaped_scaled_mat1 = fd.ops.div(reshaped_mat1, broadcast_scale1)
  scaled_mat1 = fd.ops.reshape(reshaped_scaled_mat1, [m, k])
  #t_scaled_mat1 = fd.ops.permute(scaled_mat1, [1, 0])
  fp4_mat1 = fd.ops.cast(scaled_mat1, DataType.Float4_e2m1fn)
  fp4_scale1 = fd.ops.cast(scale1, DataType.Float4_e2m1fn)
  output = fd.ops.matmul(fp4_scale1, fp4_mat1)
  fd.add_output(output)

with FusionDefinition() as fd:
  nvfuser_fusion_id1(fd)

lookup = torch.randn(m, k, dtype=torch.float, device='cuda:0')
o = fd.execute([lookup])[0]
print(o)

Reverted to original example.

Indexing failed on this one. Note the the failure doesn't repro with id_model indexer.

The fusion looked pretty innocent
It's failing at the second segment on the load instruction.

Error from segmentation group 3:  INTERNAL ASSERT FAILED at /opt/pytorch/nvfuser/csrc/index_compute.cpp:1983, please report a bug with repro script to NVFuser at https://github.com/NVIDIA/Fuser/issues.
Expected index_map.find(alloc_dom[i]) != index_map.end() . Couldn't find allocation mapping for T14_l_float[iblockIdx.x123{( ceilDiv(( ceilDiv(256, blockDim.x) ), 2) )}, iblockIdx.y125{1024}, iUS126{1}, iS124{2}, ithreadIdx.x122{blockDim.x}, iS120{4}] ca_pos( 3 ) dim: 1 id: iS77{1024}, loops:  iblockIdx.x88{( ceilDiv(( ceilDiv(4096, blockDim.x) ), 2) )} iblockIdx.y90{1024} iUS91{1} iS124{2} ithreadIdx.x122{blockDim.x} iS120{4}
Exception raised from getNonGlobalConsumerStridedIndices at /opt/pytorch/nvfuser/csrc/index_compute.cpp:1983 (most recent call first):
frame #0: nvfuser::nvfCheckFail(char const*, char const*, long, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&) + 0x110 (0xfb2f01f98650 in /opt/pytorch/nvfuser/nvfuser_direct/../build/libnvfuser_codegen.so)