diff --git a/atom/plugin/vllm/deepseek_v4_bridge.py b/atom/plugin/vllm/deepseek_v4_bridge.py index fb79e1946e..9dd300a59b 100644 --- a/atom/plugin/vllm/deepseek_v4_bridge.py +++ b/atom/plugin/vllm/deepseek_v4_bridge.py @@ -700,7 +700,12 @@ def i32(*shape): return CpuGpuBuffer(*shape, dtype=torch.int32, device=device) # Per-seq scalars (sized to padded request count == num_slots). - self.state_slot = i32(S) + # Core DeepSeek-V4 metadata distinguishes the state slot read by the + # compressor from the slot written by this forward. vLLM does not fork + # state today, so both buffers carry the same values, but they must have + # independent addresses for CUDA/HIP graph replay. + self.state_slot_in = i32(S) + self.state_slot_out = i32(S) self.n_csa = i32(S) self.n_hca = i32(S) # Per-token mapping (sized to padded token count). int32: accepted by @@ -1191,10 +1196,10 @@ def build_atom_v4_attention_metadata( applied or out of sync) the build raises rather than reading the device block table. """ - from atom.utils.forward_context import AttentionMetaData + from atom.model_ops.attentions.deepseek_v4_attn import AttentionMetaData_DSV4 if common_attn_metadata is None: - return AttentionMetaData() + return AttentionMetaData_DSV4() state = _infer_atom_attn_state(common_attn_metadata, num_spec_tokens) is_decode = state.value == "decode" device = common_attn_metadata.seq_lens.device @@ -1228,7 +1233,7 @@ def build_atom_v4_attention_metadata( seq_lens_cpu = common_attn_metadata.seq_lens.cpu() seq_np = seq_lens_cpu[:num_reqs].numpy().astype(np.int32) batch_np = np.repeat(np.arange(num_reqs, dtype=np.int32), lens) - md = AttentionMetaData( + md = AttentionMetaData_DSV4( cu_seqlens_q=common_attn_metadata.query_start_loc, cu_seqlens_k=common_attn_metadata.query_start_loc, max_seqlen_q=int(common_attn_metadata.max_query_len), @@ -1333,8 +1338,14 @@ def build_atom_v4_attention_metadata( if decode_persistent: bufs = decode_bufs + md.state_slot_out_cpu = slot_arr + md.state_slot_out = bufs.stage(bufs.state_slot_out, slot_arr) + md.state_slot_in = bufs.stage(bufs.state_slot_in, slot_arr) + # Bridge-local helpers still use the legacy single-slot spelling. It is + # the write slot because paged-index generation addresses the state + # populated by this forward. md.state_slot_mapping_cpu = slot_arr - md.state_slot_mapping = bufs.stage(bufs.state_slot, slot_arr) + md.state_slot_mapping = md.state_slot_out # Per-token seq map padded to T_pad with the -1 sentinel tail. if total: bufs.batch_id.np[:total] = batch_np @@ -1404,7 +1415,10 @@ def build_atom_v4_attention_metadata( return md # ---- eager path: prefill, or decode without persistent buffers ---- - md.state_slot_mapping = torch.from_numpy(slot_arr).to(device) + md.state_slot_out = torch.from_numpy(slot_arr).to(device) + md.state_slot_in = md.state_slot_out.clone() + md.state_slot_out_cpu = slot_arr + md.state_slot_mapping = md.state_slot_out md.state_slot_mapping_cpu = slot_arr md.batch_id_per_token = torch.from_numpy(batch_np).to(device) md.n_committed_csa_per_seq = torch.from_numpy(n_csa_cpu).to(device) diff --git a/atom/plugin/vllm/models/deepseek_v4.py b/atom/plugin/vllm/models/deepseek_v4.py index cc50a07381..b737919181 100644 --- a/atom/plugin/vllm/models/deepseek_v4.py +++ b/atom/plugin/vllm/models/deepseek_v4.py @@ -12,19 +12,26 @@ variant, then restores it. """ -from typing import Optional - import torch from atom.models import deepseek_v4 as deepseek_v4_base + +# isort: off from atom.models.deepseek_v4 import ( DeepseekV4Attention as DeepseekV4AttentionBase, DeepseekV4ForCausalLM as DeepseekV4ForCausalLMBase, DeepseekV4Model as DeepseekV4ModelBase, Indexer as IndexerBase, ) + +# isort: on +from atom.plugin.vllm.deepseek_v4_bridge import ATOM_DEEPSEEK_V4_BLOCK_SIZE from atom.utils.forward_context import AttnState, get_forward_context +# Compressor.forward and CSA translation read this module global at runtime, so +# it must remain aligned with the vLLM proxy page geometry after construction. +deepseek_v4_base._V4_BLOCK_SIZE = ATOM_DEEPSEEK_V4_BLOCK_SIZE + class IndexerVllm(IndexerBase): """DeepSeek-V4 sparse indexer with vLLM mixed-batch decode/prefill split. diff --git a/tests/plugin/test_vllm_deepseek_v4_metadata_contract.py b/tests/plugin/test_vllm_deepseek_v4_metadata_contract.py new file mode 100644 index 0000000000..616b04ea96 --- /dev/null +++ b/tests/plugin/test_vllm_deepseek_v4_metadata_contract.py @@ -0,0 +1,22 @@ +# SPDX-License-Identifier: MIT +# Copyright (C) 2024-2026, Advanced Micro Devices, Inc. All rights reserved. + +from pathlib import Path + +BRIDGE_SOURCE = ( + Path(__file__).parents[2] / "atom/plugin/vllm/deepseek_v4_bridge.py" +).read_text() + + +def test_vllm_decode_buffers_keep_distinct_state_input_and_output_addresses(): + assert "self.state_slot_in = i32(S)" in BRIDGE_SOURCE + assert "self.state_slot_out = i32(S)" in BRIDGE_SOURCE + assert "bufs.stage(bufs.state_slot_in, slot_arr)" in BRIDGE_SOURCE + assert "bufs.stage(bufs.state_slot_out, slot_arr)" in BRIDGE_SOURCE + + +def test_vllm_eager_metadata_satisfies_split_state_slot_contract(): + assert "md = AttentionMetaData_DSV4(" in BRIDGE_SOURCE + assert "md.state_slot_out = torch.from_numpy(slot_arr).to(device)" in BRIDGE_SOURCE + assert "md.state_slot_in = md.state_slot_out.clone()" in BRIDGE_SOURCE + assert "md.state_slot_mapping = md.state_slot_out" in BRIDGE_SOURCE diff --git a/tests/test_mtp_deferred_status_queue.py b/tests/test_mtp_deferred_status_queue.py index 736598f09a..818c574d39 100644 --- a/tests/test_mtp_deferred_status_queue.py +++ b/tests/test_mtp_deferred_status_queue.py @@ -2,6 +2,9 @@ from unittest import mock import numpy as np +import pytest + +pytest.importorskip("aiter", reason="requires AITER to import model_runner") from atom.model_engine.model_runner import tokenIDProcessor from atom.model_engine.scheduler import ScheduledBatch