submission 553064

from task import input_t, output_t

import torch
import helion
import helion.language as hl


import os
os.environ["HELION_AUTOTUNE_EFFORT"] = "none"

SHAPE_CONFIGS: dict[tuple, helion.Config] = {
    (1, 256, 64):    helion.Config(block_sizes=[1], num_warps=1, num_stages=1),
    (4, 512, 128):   helion.Config(block_sizes=[1], num_warps=1, num_stages=1),
    (16, 1024, 64):  helion.Config(block_sizes=[1], num_warps=1, num_stages=1),
    (1, 4096, 128):  helion.Config(block_sizes=[1], num_warps=1, num_stages=1),
    (8, 4096, 128):  helion.Config(block_sizes=[1], num_warps=1, num_stages=1),
    (256, 4096, 128):  helion.Config(block_sizes=[1], num_warps=1, num_stages=1),
    (256, 8192, 128):  helion.Config(block_sizes=[1], num_warps=1, num_stages=1),
    (4096, 7168, 128): helion.Config(block_sizes=[1], num_warps=1, num_stages=1),
}


#     helion.Config(..., advanced_controls_file="/opt/booster_pack/fp8_group_quant_0.acf")


def _make_kernel(config: helion.Config):
    @helion.kernel(static_shapes=True, config=config)
    def kernel(
        data: torch.Tensor,        # [N, group_size] flattened groups
        scales_out: torch.Tensor,  # [N] output scale per group
    ) -> torch.Tensor:
        nrows = data.size(0)
        ncols = hl.specialize(data.size(1))
        MAX_VAL = 448.0

        qout = torch.empty(nrows, ncols, dtype=torch.float32, device=data.device)

        for rr in hl.tile(nrows):
            row = data[rr, :].to(torch.float32)
            absmax = torch.amax(row.abs(), dim=-1).clamp(min=1e-10)
            scale = absmax / MAX_VAL
            qout[rr, :] = (row / scale[:, None]).clamp(-MAX_VAL, MAX_VAL)
            scales_out[rr] = scale

        return qout

    return kernel


_KERNELS = {shape: _make_kernel(cfg) for shape, cfg in SHAPE_CONFIGS.items()}


def custom_kernel(data: input_t) -> output_t:
    x, x_q, x_s = data
    T, H = x.shape
    G = x_s.shape[1]
    gsz = H // G
    N = T * G

    kernel = _KERNELS[(T, H, gsz)]

    flat_in = x.reshape(N, gsz)
    flat_s = x_s.reshape(N)

    flat_q = kernel(flat_in, flat_s)

    x_q[...] = flat_q.reshape(T, H)
    x_s[...] = flat_s.reshape(T, G)
    return x_q, x_s