jax-ml/jax · error · ValueError

`next_power_of_2` requires a non-negative integer.

Error message

`next_power_of_2` requires a non-negative integer.

What it means

pallas.utils.next_power_of_2(x) computes the smallest power of two >= x and only accepts non-negative integers. A negative input raises ValueError because bit_length-based computation is undefined for negatives.

Source

Thrown at jax/_src/pallas/utils.py:107

    >>> align_to(5, 4)  # 5 is not a multiple of 4, so rounds up to 8
    8
  """
  return cdiv(a, alignment) * alignment


def strides_from_shape(shape: tuple[int, ...]) -> tuple[int, ...]:
  size = np.prod(shape)
  strides = []
  for s in shape:
    size = size // s
    strides.append(int(size))
  return tuple(strides)


def next_power_of_2(x: int) -> int:
  """Returns the next power of two greater than or equal to `x`."""
  if x < 0:
    raise ValueError("`next_power_of_2` requires a non-negative integer.")
  return 1 if x == 0 else 2 ** (x - 1).bit_length()


def pattern_match_scan_to_fori_loop(
    jaxpr: jax_core.Jaxpr, num_consts: int, num_carry: int
) -> tuple[jax_core.Jaxpr, bool]:
  num_extensive_inputs = len(jaxpr.invars) - num_consts - num_carry
  num_extensive_outputs = len(jaxpr.outvars) - num_carry
  if num_extensive_outputs:
    raise ValueError(
        f"Scan with {num_extensive_outputs} extensive output(s) is not"
        " supported."
    )
  if num_extensive_inputs:
    raise ValueError(
        f"Scan with {num_extensive_inputs} extensive argument(s) is not"
        f" supported. Found {num_consts} consts and {num_carry} carry"
        " arguments."

View on GitHub (pinned to 1e1c6a8fc0)

Solutions

1. Fix the upstream computation so the argument is non-negative
2. Guard with max(0, x) if negatives are legitimately possible
3. Validate block sizes with an assert before calling

Example fix

# before
next_power_of_2(n - block_size)
# after
next_power_of_2(max(0, n - block_size))

assert isinstance(x, int) and x >= 0, 'next_power_of_2 needs a non-negative int'

When it happens

Trigger: Calling next_power_of_2 with a negative number, e.g. next_power_of_2(-8); typically the negative comes from a computed block size or shape expression (e.g. padding - kernel).

Common situations: Kernel grid/block-size math producing negative values when problem dimensions are smaller than assumed; accidental sign error in a size formula.

Understand the failure class

Background: "Must be a positive integer", "Invalid value", "Unsupported": the invalid-argument-value error family, when a library rejects the value you pass — this error's family across 35 libraries.

Related errors

• Sum of sizes {n} must be equal to dimension {axis} of the op
• Cannot pull iota along dimension {dimension} with None block
• Invalid memory space: {memory_space!r}
• dimension_semantics must be the same length as grid.
• Grid must consist of Python integers and JAX Arrays: {grid_t

AI-assisted analysis of jax-ml/jax@1e1c6a8fc0 (2026-08-27). Data as JSON: /api/errors/8cc102d3e90f2121. Report an issue: GitHub.