diff --git a/vllm/models/qwen4_exp/nvidia/ops/qsa_prepare.py b/vllm/models/qwen4_exp/nvidia/ops/qsa_prepare.py index 1b26fd8145..d5e809ee0e 100644 --- a/vllm/models/qwen4_exp/nvidia/ops/qsa_prepare.py +++ b/vllm/models/qwen4_exp/nvidia/ops/qsa_prepare.py @@ -222,6 +222,7 @@ def _qsa_prepare_kernel( num_k_work, num_requests, # bounds-guard instrumentation guard_ctr, # bounds-guard instrumentation: persistent int64[3] + ring_skip_ctr, # ring-skip telemetry, persistent int64[3], SITE_QSA_RING_SKIP HQ: tl.constexpr, D: tl.constexpr, TILE_T_Q: tl.constexpr, @@ -555,6 +556,19 @@ def _qsa_prepare_kernel( valid_slot = ( valid_token & (slot >= 0) & (slot < num_state_blocks * STATE_SIZE) ) + # Ring-skip telemetry: a live token whose slot fails the range check + # has its ring row silently NOT written (every store below is + # masked by valid_slot). That token's raw key is then absent + # from the ring the next step reads (a candidate cause). Count + # it -- the write stays skipped, the reconciliation is the ring-row tag fix. + skipped = valid_token & ~valid_slot + tl.atomic_add(ring_skip_ctr, 1, mask=skipped) + tl.atomic_max( + ring_skip_ctr + 1, token.to(tl.int64) + _GUARD_BIAS, mask=skipped + ) + tl.atomic_max( + ring_skip_ctr + 2, slot.to(tl.int64) + _GUARD_BIAS, mask=skipped + ) safe_slot = tl.maximum(slot, 0) state_row = ( state_cache_ptr @@ -712,6 +726,7 @@ def qsa_prepare( num_k_work, min(query_start_loc.numel() - 1, state_block_table.shape[0]), guard_telemetry.counter(guard_telemetry.SITE_QSA_PRE_INDEXER, q.device), + guard_telemetry.counter(guard_telemetry.SITE_QSA_RING_SKIP, q.device), HQ=num_q_heads, D=head_dim, TILE_T_Q=TILE_T_Q,