For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /max/get-started.md).

Mojo function

naive_gemv

def naive_gemv(c_buf: TileTensor[Storage=c_buf.Storage, address_space=c_buf.address_space, linear_idx_type=c_buf.linear_idx_type], a_buf: TileTensor[Storage=a_buf.Storage, address_space=a_buf.address_space, linear_idx_type=a_buf.linear_idx_type], b_buf: TileTensor[Storage=b_buf.Storage, address_space=b_buf.address_space, linear_idx_type=b_buf.linear_idx_type])

Computes a reference matrix-vector product C = A * b using a scalar nested loop.

Iterates over K then M, accumulating each A[m, k] * b[k] into c[m]. Intended for correctness testing rather than performance-critical paths.

Args:

c_buf (TileTensor[Storage=c_buf.Storage, address_space=c_buf.address_space, linear_idx_type=c_buf.linear_idx_type]): Output TileTensor of length M, zero-filled on entry.
a_buf (TileTensor[Storage=a_buf.Storage, address_space=a_buf.address_space, linear_idx_type=a_buf.linear_idx_type]): Input matrix TileTensor of shape (M, K).
b_buf (TileTensor[Storage=b_buf.Storage, address_space=b_buf.address_space, linear_idx_type=b_buf.linear_idx_type]): Input vector TileTensor of length K.