Index manipulation not working in te.sum

Troubleshooting
#1

The adding index manipulation in te.sum seems not working with tvm 0.8.dev0.

n = 1024
dtype = "float32"
A = tvm.te.placeholder((n, n), dtype=dtype, name='A')
k = tvm.te.reduce_axis((0, 8), name='k')
B = tvm.te.compute((8,), lambda i: tvm.te.sum(A[i, k + i - (i % 8)], axis=k), name='B')

s = tvm.te.create_schedule(B.op)
print(tvm.lower(s, [A, B], simple_mode=True))

primfn(A_1: handle, B_1: handle) -> ()
  attr = {"global_symbol": "main", "tir.noalias": True}
  buffers = {A: Buffer(A_2: Pointer(float32), float32, [1024, 1024], []),
             B: Buffer(B_2: Pointer(float32), float32, [8], [])}
  buffer_map = {A_1: A, B_1: B} {
  for (i: int32, 0, 8) {
    B_2[i] = 0f32
    for (k: int32, 0, 8) {
      B_2[i] = ((float32*)B_2[i] + (float32*)A_2[((i*1024) + k)])
    }
  }
}

Expected to

primfn(A_1: handle, B_1: handle) -> ()
  attr = {"global_symbol": "main", "tir.noalias": True}
  buffers = {A: Buffer(A_2: Pointer(float32), float32, [1024, 1024], []),
             B: Buffer(B_2: Pointer(float32), float32, [8], [])}
  buffer_map = {A_1: A, B_1: B} {
  for (i: int32, 0, 8) {
    B_2[i] = 0f32
    for (k: int32, 0, 8) {
      B_2[i] = ((float32*)B_2[i] + (float32*)A_2[((i*1024) + k + i - (i % 8))])
    }
  }
}

Is this kind of tiling allowed or error should generated here?

#2

Sorry the problem solved with typo here. The code was auto optimized.

n = 1024
dtype = "float32"
A = tvm.te.placeholder((n, n), dtype=dtype, name='A')
k = tvm.te.reduce_axis((0, 8), name='k')
B = tvm.te.compute((n,), lambda i: tvm.te.sum(A[i, k + i - (i % 8)], axis=k), name='B')

s = tvm.te.create_schedule(B.op)
print(tvm.lower(s, [A, B], simple_mode=True))

primfn(A_1: handle, B_1: handle) -> ()
  attr = {"global_symbol": "main", "tir.noalias": True}
  buffers = {A: Buffer(A_2: Pointer(float32), float32, [1024, 1024], []),
             B: Buffer(B_2: Pointer(float32), float32, [1024], [])}
  buffer_map = {A_1: A, B_1: B} {
  for (i: int32, 0, 1024) {
    B_2[i] = 0f32
    for (k: int32, 0, 8) {
      B_2[i] = ((float32*)B_2[i] + (float32*)A_2[(((i*1025) + k) - floormod(i, 8))])
    }
  }
}