[rocm-libraries] ROCm/rocm-libraries#6914 (commit b791478)

[CK_TILE][FMHA] Fix sink un-mask under right-window and emit
 fp8bf16 batch_prefill sink kernels (#6914)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

## Summary

Two related fixes to `ck_tile` FMHA so that StreamLLM-sink +
sliding-window
  batch-prefill works correctly for fp8 KV / bf16 compute.

  Review the commits in this order:

  1. `fmha: emit sink kernels for fp8bf16 batch_prefill`
Extends `example/ck_tile/01_fmha/codegen/ops/fmha_batch_prefill.py` so
     the fp8(KV) / bf16(QO) batch-prefill codegen also emits the
`mask=mask_enum::generic_with_sink` variant. Without this the runtime
     could not dispatch to a sink-aware kernel for the fp8bf16 path.

  2. `fmha: respect right-window in IsOutOfSinkBound`
The sink un-mask in `GenericAttentionMask::IsOutOfSinkBound` (local-mask
branch) used `(i_y + x) > 1` as the gate, which conditioned on the row
     index instead of the column index. As a result, queries `1..sink-1`
could attend to *future* sink positions (violating causal /
right-window),
while query `0` fell back to the plain causal mask. The fix replaces the
guard with `i_x < i_y + x` so every query only sees sink columns up to
     its own right-window boundary.

  3. `fmha: clarify IsOutOfSinkBound predicate comment`
Doc-only follow-up that rewrites the comment above the predicate as a
     clause-by-clause explanation (`i_x < sink`, `i_x < i_y + x`,
     `y < y_total`, `i_y < x_total`).

  ## Test plan

- [x] Repro on aiter `op_tests/test_batch_prefill.py` (fp8 +
bf16_dequant
        modes with `sink=4`, `win_left=1023`, `softcap=0.0`, `sal=True`)
        now passes for all parametrized shapes.
- [x] Existing fp16/bf16 batch-prefill paths (no sink) unchanged —
codegen
diff only adds the `generic_with_sink` variant for fp8bf16; existing
        kernel object lists unaffected.

## Submission Checklist

- [x] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.
This commit is contained in:
Linjun-AMD
2026-05-07 02:40:45 +00:00
committed by assistant-librarian[bot]
parent 207a95d5e4
commit 33b62ed087
2 changed files with 30 additions and 5 deletions

View File

@@ -242,16 +242,27 @@ struct GenericAttentionMask
index_t x_start = -y + i_y + 1;
index_t x_end = min(i_y + x, x_total);
// Sink un-mask predicate, clause by clause:
// i_x < sink : the column lives inside the StreamLLM sink prefix.
// i_x < i_y + x : the column is not in the masked-out future of the
// window (= < x_end modulo the min with x_total);
// without this, queries <= sink-1 would be allowed
// to look at later sink columns/positions than they
// should under causality / right-window.
// y < y_total : the local window doesn't already span everything
// (otherwise sink un-mask is meaningless).
// i_y < x_total : the query row is in-range vs. the key sequence
// (handles seqlen_q > seqlen_k padding).
if constexpr(IsLocal)
{
if((i_x < sink) && (y < y_total) && ((i_y + x) > 1) && i_y < x_total)
if((i_x < sink) && (i_x < i_y + x) && (y < y_total) && i_y < x_total)
return false;
else
return i_x < x_start || i_x >= x_end;
}
else
{
if((i_x < sink) && (y < y_total) && ((i_y + x) > 1) && i_y < x_total)
if((i_x < sink) && (i_x < i_y + x) && (y < y_total) && i_y < x_total)
return false;
else
return i_x >= x_end || i_y >= y_total;
@@ -498,7 +509,9 @@ struct SimplifiedGenericAttentionMask
return i_x >= x_total;
index_t x_start = -y + i_y + 1; // this could be negative, but it's fine
index_t x_end = min(i_y + x, x_total); // need min in case x is padded
if((i_x < sink) && (y < y_total) && ((i_y + x) > 1) && i_y < x_total)
// See note in the local-mask IsOutOfSinkBound: the sink column i_x is
// only valid up to the right-window boundary i_y + x.
if((i_x < sink) && (i_x < i_y + x) && (y < y_total) && i_y < x_total)
return false;
else
return i_x < x_start || i_x >= x_end || i_y >= y_total;