_xpu_ops.py

# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project

from typing import TYPE_CHECKING

import torch
from vllm_xpu_kernels.flash_attn_interface import flash_attn_varlen_func

from vllm.logger import init_logger
from vllm.platforms import current_platform
from vllm.utils.torch_utils import direct_register_custom_op

logger = init_logger(__name__)

if TYPE_CHECKING:

    def register_fake(fn):
        return lambda name: fn
else:
    try:
        from torch.library import register_fake
    except ImportError:
        from torch.library import impl_abstract as register_fake

if hasattr(torch.ops._xpu_C, "fp8_gemm_w8a16"):

    @register_fake("_xpu_C::fp8_gemm_w8a16")
    def _fp8_gemm_w8a16_fake(
        input: torch.Tensor,
        q_weight: torch.Tensor,
        weight_scale: torch.Tensor,
        bias: torch.Tensor | None = None,
    ) -> torch.Tensor:
        input_2d = input.view(-1, input.shape[-1])
        M = input_2d.size(0)
        N = q_weight.size(1)
        return torch.empty((M, N), dtype=input.dtype, device=input.device)


if hasattr(torch.ops._xpu_C, "int4_gemm_w4a8"):

    @register_fake("_xpu_C::int4_gemm_w4a8")
    def _int4_gemm_w4a8_fake(
        input: torch.Tensor,
        input_scales: torch.Tensor,
        input_zero_points: torch.Tensor,
        q_weight: torch.Tensor,
        weight_scale: torch.Tensor,
        weight_zp: torch.Tensor,
        group_size: int,
        g_idx: torch.Tensor | None = None,
        bias: torch.Tensor | None = None,
    ) -> torch.Tensor:
        input_2d = input.view(-1, input.shape[-1])
        M = input_2d.size(0)
        N = q_weight.size(1)
        return torch.empty((M, N), dtype=torch.float16, device=input.device)


if hasattr(torch.ops._xpu_C, "int4_gemm_w4a16"):

    @register_fake("_xpu_C::int4_gemm_w4a16")
    def _int4_gemm_w4a16_fake(
        input: torch.Tensor,
        q_weight: torch.Tensor,
        bias: torch.Tensor | None,
        weight_scale: torch.Tensor,
        qzeros: torch.Tensor,
        group_size: int,
        group_idx: torch.Tensor | None = None,
    ) -> torch.Tensor:
        input_2d = input.view(-1, input.shape[-1])
        M = input_2d.size(0)
        N = q_weight.size(1)
        return torch.empty((M, N), dtype=input.dtype, device=input.device)


def _xpu_ops_deepseek_scaling_rope_impl(
    positions: torch.Tensor,
    query: torch.Tensor,
    key: torch.Tensor | None,
    offsets: torch.Tensor | None,
    cos_sin_cache: torch.Tensor | None,
    rotary_dim: int,
    is_neox_style: bool,
) -> tuple[torch.Tensor, torch.Tensor]:
    assert key is not None
    return torch.ops._xpu_C.deepseek_scaling_rope(
        positions, query, key, offsets, cos_sin_cache, rotary_dim, is_neox_style
    )


def _xpu_ops_deepseek_scaling_rope_fake(
    positions: torch.Tensor,
    query: torch.Tensor,
    key: torch.Tensor | None,
    offsets: torch.Tensor | None,
    cos_sin_cache: torch.Tensor | None,
    rotary_dim: int,
    is_neox_style: bool,
) -> tuple[torch.Tensor, torch.Tensor]:
    return query, key


# Global flag to ensure ops are registered only once
_OPS_REGISTERED = False


class xpu_ops:
    @staticmethod
    @torch.compile
    def dynamic_per_token_int8_quant_ref(
        input: torch.Tensor, use_sym_quant: bool, bits: int
    ):
        original_sizes = input.size()
        # view is not safe in torch.compile if input is not contiguous
        input = input.reshape(
            -1, original_sizes[-1]
        )  # Flatten except for the last dimension
        qmin = -(2 ** (bits - 1)) if use_sym_quant else 0
        qmax = 2 ** (bits - 1) - 1 if use_sym_quant else 2**bits - 1
        min_val = torch.min(input, dim=-1)[0].to(dtype=torch.float32).unsqueeze(-1)
        max_val = torch.max(input, dim=-1)[0].to(dtype=torch.float32).unsqueeze(-1)
        if use_sym_quant:
            scale = (
                torch.maximum(torch.abs(min_val), torch.abs(max_val)) / qmax
            ).clamp(min=1e-5)
            zero_point = torch.zeros_like(scale).to(dtype=torch.int32)
        else:
            scale = ((max_val - min_val) / qmax).clamp(min=1e-5)
            zero_point = -1 * torch.round(min_val / scale).to(dtype=torch.int32)
        scale = scale.to(dtype=input.dtype)
        quantized = torch.clamp(
            torch.round(input / scale.to(dtype=torch.float32) + zero_point),
            qmin,
            qmax,
        ).to(dtype=torch.int8 if use_sym_quant else torch.uint8)
        return (
            quantized.view(original_sizes),
            scale.view(original_sizes[:-1] + (1,)),
            zero_point.view(original_sizes[:-1] + (1,)),
        )

    @staticmethod
    def flash_attn_varlen_func(
        q: torch.Tensor,
        k: torch.Tensor,
        v: torch.Tensor,
        cu_seqlens_q: torch.Tensor,
        max_seqlen_q: int,
        max_seqlen_k: int,
        softmax_scale: float | None = None,
        causal: bool = False,
        out: torch.Tensor | None = None,
        block_table: torch.Tensor | None = None,
        alibi_slopes: torch.Tensor | None = None,
        window_size: list[int] | None = None,
        softcap: float | None = 0.0,
        seqused_k: torch.Tensor | None = None,
        cu_seqlens_k: torch.Tensor | None = None,
        # passed in qwen vl
        dropout_p: float = 0.0,
        # The following parameters are not used in xpu kernel currently,
        # we keep API compatible to CUDA's.
        scheduler_metadata=None,
        fa_version: int = 2,
        q_descale=None,
        k_descale=None,
        v_descale=None,
        num_splits=0,
        return_softmax_lse: bool | None = False,
        s_aux: torch.Tensor | None = None,
    ):
        assert cu_seqlens_k is not None or seqused_k is not None, (
            "cu_seqlens_k or seqused_k must be provided"
        )
        assert cu_seqlens_k is None or seqused_k is None, (
            "cu_seqlens_k and seqused_k cannot be provided at the same time"
        )
        assert block_table is None or seqused_k is not None, (
            "when enable block_table, seqused_k is needed"
        )
        assert block_table is not None or cu_seqlens_k is not None, (
            "when block_table is disabled, cu_seqlens_k is needed"
        )
        if out is None:
            out = torch.empty(q.shape, dtype=q.dtype, device=q.device)
        real_window_size: tuple[int, int]
        if window_size is None:
            real_window_size = (-1, -1)
        else:
            assert len(window_size) == 2
            real_window_size = (window_size[0], window_size[1])  # noqa: F841

        # In encode attention, k and v maybe not contiguous and current
        # kernel can't handle it
        if block_table is None:
            k = k.contiguous()
            v = v.contiguous()
        return flash_attn_varlen_func(
            out=out,
            q=q.contiguous(),
            k=k,
            v=v,
            cu_seqlens_q=cu_seqlens_q,
            cu_seqlens_k=cu_seqlens_k,
            seqused_k=seqused_k,
            max_seqlen_q=max_seqlen_q,
            max_seqlen_k=max_seqlen_k,
            softmax_scale=softmax_scale,
            causal=causal,
            block_table=block_table,
            s_aux=s_aux,
            window_size=real_window_size,
            # alibi_slopes = alibi_slopes,
            # softcap=softcap,
            return_softmax_lse=return_softmax_lse,
        )

    @staticmethod
    def get_scheduler_metadata(
        batch_size,
        max_seqlen_q,
        max_seqlen_k,
        num_heads_q,
        num_heads_kv,
        headdim,
        cache_seqlens: torch.Tensor,
        qkv_dtype=torch.bfloat16,
        headdim_v=None,
        cu_seqlens_q: torch.Tensor | None = None,
        cu_seqlens_k_new: torch.Tensor | None = None,
        cache_leftpad: torch.Tensor | None = None,
        page_size: int | None = None,
        max_seqlen_k_new=0,
        causal=False,
        window_size=(-1, -1),  # -1 means infinite context window
        has_softcap=False,
        num_splits=0,  # Can be tuned for speed
        pack_gqa=None,  # Can be tuned for speed
        sm_margin=0,  # Can be tuned if some SMs are used for communication
    ) -> None:
        logger.warning_once(
            "get_scheduler_metadata is not implemented for xpu_ops, returning None."
        )
        return None

    @staticmethod
    def indexer_k_quant_and_cache(
        k: torch.Tensor,
        kv_cache: torch.Tensor,
        slot_mapping: torch.Tensor,
        quant_block_size: int,
        scale_fmt: str | None,
    ) -> None:
        head_dim = k.shape[-1]
        k = k.view(-1, head_dim)  # [total_tokens, head_dim]

        def group_quant_torch(
            x: torch.Tensor,
            group_size: int,
            eps: float = 1e-10,
            dtype: torch.dtype | None = None,
            column_major_scales: bool = False,
            out_q: torch.Tensor | None = None,
            use_ue8m0: bool | None = None,
        ) -> tuple[torch.Tensor, torch.Tensor]:
            if use_ue8m0 is None:
                # Default fallback - could import is_deep_gemm_e8m0_used if needed
                use_ue8m0 = False

            if dtype is None:
                dtype = current_platform.fp8_dtype()

            # Validate inputs
            assert x.shape[-1] % group_size == 0, (
                f"Last dimension {x.shape[-1]} must be divisible by "
                f"group_size {group_size}"
            )
            assert x.stride(-1) == 1, "Input tensor groups must be contiguous"

            # Prepare output tensor
            if out_q is None:
                x_q = torch.empty_like(x, dtype=dtype)
            else:
                assert out_q.shape == x.shape
                x_q = out_q

            # Reshape input for group processing
            # Original shape: (..., last_dim)
            # Target shape: (..., num_groups, group_size)
            original_shape = x.shape
            num_groups = original_shape[-1] // group_size

            # Reshape to separate groups
            group_shape = original_shape[:-1] + (num_groups, group_size)
            x_grouped = x.view(group_shape)

            # Compute per-group absolute maximum values
            # Shape: (..., num_groups)
            abs_max = torch.amax(torch.abs(x_grouped), dim=-1, keepdim=False)
            abs_max = torch.maximum(
                abs_max, torch.tensor(eps, device=x.device, dtype=x.dtype)
            )

            # Compute scales
            FP8_MAX = torch.finfo(dtype).max
            FP8_MIN = torch.finfo(dtype).min
            scale_raw = abs_max / FP8_MAX

            if use_ue8m0:
                # For UE8M0 format, scales must be powers of 2
                scales = torch.pow(2.0, torch.ceil(torch.log2(scale_raw)))
            else:
                scales = scale_raw

            # Expand scales for broadcasting with grouped data
            # Shape: (..., num_groups, 1)
            scales_expanded = scales.unsqueeze(-1)

            # Quantize the grouped data
            x_scaled = x_grouped / scales_expanded
            x_clamped = torch.clamp(x_scaled, FP8_MIN, FP8_MAX)
            x_quantized = x_clamped.to(dtype)

            # Reshape back to original shape
            x_q.copy_(x_quantized.view(original_shape))

            # Prepare scales tensor in requested format
            if column_major_scales:
                # Column-major: (num_groups,) + batch_dims
                # Transpose the scales to put group dimension first
                scales_shape = (num_groups,) + original_shape[:-1]
                x_s = scales.permute(-1, *range(len(original_shape) - 1))
                x_s = x_s.contiguous().view(scales_shape)
            else:
                # Row-major: batch_dims + (num_groups,)
                x_s = scales.contiguous()

            # Ensure scales are float32
            return x_q, x_s.float()

        k_fp8, k_scale = group_quant_torch(
            k,
            group_size=quant_block_size,
            column_major_scales=False,
            use_ue8m0=(scale_fmt == "ue8m0"),
        )

        k_fp8_bytes = k_fp8.view(-1, head_dim).view(torch.uint8)
        scale_bytes = k_scale.view(torch.uint8).view(-1, 4)
        k = torch.cat(
            [k_fp8_bytes, scale_bytes], dim=-1
        )  # [total_tokens, head_dim + 4]

        slot_mapping = slot_mapping.flatten()
        # kv_cache: [num_block, block_size, head_dim + 4]
        kv_cache.view(-1, kv_cache.shape[-1]).index_copy_(0, slot_mapping, k)

    @staticmethod
    def cp_gather_indexer_k_quant_cache(
        kv_cache: torch.Tensor,
        dst_k: torch.Tensor,
        dst_scale: torch.Tensor,
        block_table: torch.Tensor,
        cu_seq_lens: torch.Tensor,
    ) -> None:
        """
        Args:
            kv_cache: [num_blocks, block_size, cache_stride] - quantized KV cache
                    Layout per block: [k_values, scale_values]
                    - k_values: [block_size * head_dim]
                    - scale_values: [block_size * head_dim * 4 / quant_block_size]
            dst_k: [num_tokens, head_dim] - output tensor for K values
            dst_scale: [num_tokens, head_dim / quant_block_size * 4]
                - output tensor for scale values
            block_table: [batch_size, num_blocks] - block table for indexing
            cu_seq_lens: [batch_size + 1] - cumulative sequence lengths
        """
        batch_size = block_table.size(0)
        num_tokens = dst_k.size(0)
        head_dim = dst_k.size(1)
        cache_block_size = kv_cache.size(1)
        quant_block_size = head_dim * 4 // dst_scale.size(1)

        # For each token, find which batch it belongs to using searchsorted
        token_indices = torch.arange(num_tokens, device=dst_k.device) + 1
        # cu_seq_lens is [batch_size + 1], we need to find which interval each
        # token belongs to
        batch_indices = torch.searchsorted(cu_seq_lens, token_indices) - 1
        batch_indices = torch.clamp(batch_indices, 0, batch_size - 1)

        # Calculate the in-batch sequence index for each token
        inbatch_seq_indices = token_indices - cu_seq_lens[batch_indices]

        # Find which block each token belongs to
        block_indices_in_table = inbatch_seq_indices // cache_block_size
        physical_block_indices = block_table[batch_indices, block_indices_in_table]

        # Calculate the offset within each block
        inblock_offsets = (inbatch_seq_indices - 1) % cache_block_size

        # Calculate strides
        block_stride = kv_cache.stride(0)  # stride for each block

        # Flatten kv_cache for easier indexing
        kv_cache_flat = kv_cache.view(-1)

        # Calculate source offset for K values for all tokens (vectorized)
        src_block_offsets = physical_block_indices * block_stride
        src_k_offsets = src_block_offsets + inblock_offsets * head_dim

        # Gather K values using advanced indexing
        # Create indices for all elements we need to gather
        k_indices = src_k_offsets.unsqueeze(1) + torch.arange(
            head_dim, device=dst_k.device
        )
        dst_k[:] = kv_cache_flat[k_indices]

        # Calculate source offset for scale values (vectorized)
        # Scales are stored after all K values for each block
        scale_size = head_dim * 4 // quant_block_size
        src_scale_offsets = src_block_offsets + head_dim + inblock_offsets * scale_size

        # Gather scale values
        scale_indices = src_scale_offsets.unsqueeze(1) + torch.arange(
            scale_size, device=dst_scale.device
        )
        dst_scale[:] = kv_cache_flat[scale_indices]

    @staticmethod
    def top_k_per_row_prefill(
        logits: torch.Tensor,
        cu_seqlen_ks: torch.Tensor,
        cu_seqlen_ke: torch.Tensor,
        raw_topk_indices: torch.Tensor,
        num_rows: int,
        stride0: int,
        strdide1: int,
        topk_tokens: int,
    ) -> torch.Tensor:
        real_topk = min(topk_tokens, logits.shape[-1])
        topk_indices = logits.topk(real_topk, dim=-1)[1].to(torch.int32)
        topk_indices -= cu_seqlen_ks[:, None]
        mask_lo = topk_indices >= 0
        mask_hi = topk_indices - (cu_seqlen_ke - cu_seqlen_ks)[:, None] < 0
        mask = torch.full_like(
            topk_indices, False, dtype=torch.bool, device=topk_indices.device
        )
        mask = mask_lo & mask_hi
        topk_indices.masked_fill_(~mask, -1)
        raw_topk_indices[: topk_indices.shape[0], : topk_indices.shape[1]] = (
            topk_indices
        )

    @staticmethod
    def top_k_per_row_decode(
        logits: torch.Tensor,
        next_n: int,
        seq_lens: torch.Tensor,
        raw_topk_indices: torch.Tensor,
        num_rows: int,
        stride0: int,
        stride1: int,
        topk_tokens: int,
    ) -> torch.Tensor:
        device = logits.device
        batch_size = seq_lens.size(0)
        # padded query len
        padded_num_tokens = batch_size * next_n
        positions = (
            torch.arange(logits.shape[-1], device=device)
            .unsqueeze(0)
            .expand(batch_size * next_n, -1)
        )
        row_indices = torch.arange(padded_num_tokens, device=device) // next_n
        next_n_offset = torch.arange(padded_num_tokens, device=device) % next_n
        index_end_pos = (seq_lens[row_indices] - next_n + next_n_offset).unsqueeze(1)
        # index_end_pos: [B * N, 1]
        mask = positions <= index_end_pos
        # mask: [B * N, L]
        logits = logits.masked_fill(~mask, float("-inf"))
        real_topk = min(topk_tokens, logits.shape[-1])
        topk_indices = logits.topk(real_topk, dim=-1)[1].to(torch.int32)  # [B * N, K]
        # ensure we don't set indices for the top k
        # that is out of range(masked already)
        # this will happen if context length is shorter than K
        topk_indices[topk_indices > index_end_pos] = -1
        raw_topk_indices[: topk_indices.shape[0], : topk_indices.shape[1]] = (
            topk_indices
        )

    @staticmethod
    def register_ops_once() -> None:
        global _OPS_REGISTERED
        if not _OPS_REGISTERED:
            # register all the custom ops here
            direct_register_custom_op(
                op_name="xpu_ops_deepseek_scaling_rope",
                op_func=_xpu_ops_deepseek_scaling_rope_impl,
                mutates_args=[],
                fake_impl=_xpu_ops_deepseek_scaling_rope_fake,
                dispatch_key=current_platform.dispatch_key,
            )

            _OPS_REGISTERED = True


xpu_ops.register_ops_once()