[Spec Decode] Enable fused non-causal TokenSpeed MLA for DSpark (#50911)

Signed-off-by: Shreyas Misra <shreyasm@nvidia.com>
Co-authored-by: OpenAI Codex <codex@openai.com>
This commit is contained in:
NVShreyas
2026-08-04 18:05:40 +00:00
committed by GitHub
co-authored by OpenAI Codex
parent d31de3c421
commit 8ae8337ffa
2 changed files with 32 additions and 7 deletions
+24 -7
View File
@@ -827,13 +827,25 @@ def test_mock_mla_dcp_fp8_decode_gathers_quantized_query(
)
def test_tokenspeed_mla_dcp_single_token_decode_contract(monkeypatch):
def test_tokenspeed_mla_noncausal_capability():
builder = tokenspeed_mla_module.TokenspeedMLAMetadataBuilder
assert builder.supports_non_causal_multi_token_decode
assert tokenspeed_mla_module.TokenspeedMLABackend.supports_non_causal()
@pytest.mark.parametrize(
("causal", "tokens_per_decode", "dcp_world_size", "dcp_rank"),
[
pytest.param(True, 1, 2, 1, id="causal-dcp"),
pytest.param(False, 3, 1, 0, id="noncausal-multi-token"),
],
)
def test_tokenspeed_mla_decode_contract(
monkeypatch, causal, tokens_per_decode, dcp_world_size, dcp_rank
):
decode_call = None
num_decodes = 2
tokens_per_decode = 1
num_decode_tokens = num_decodes * tokens_per_decode
dcp_world_size = 2
dcp_rank = 1
num_heads = 128
kv_lora_rank = 512
qk_rope_head_dim = 64
@@ -879,6 +891,7 @@ def test_tokenspeed_mla_dcp_single_token_decode_contract(monkeypatch):
num_decodes=num_decodes,
num_decode_tokens=num_decode_tokens,
max_seq_len=max_seq_len,
causal=causal,
decode=SimpleNamespace(
block_table=torch.empty((num_decodes, 1), dtype=torch.int32),
seq_lens=torch.tensor([16, max_seq_len], dtype=torch.int32),
@@ -913,9 +926,13 @@ def test_tokenspeed_mla_dcp_single_token_decode_contract(monkeypatch):
)
torch.testing.assert_close(decode_call["seq_lens"], metadata.decode.seq_lens)
torch.testing.assert_close(decode_call["block_tables"], metadata.decode.block_table)
torch.testing.assert_close(
decode_call["causal_seqs"], metadata.decode.dcp_tot_seq_lens
)
if dcp_world_size > 1:
torch.testing.assert_close(
decode_call["causal_seqs"], metadata.decode.dcp_tot_seq_lens
)
else:
assert decode_call["causal_seqs"] is None
assert decode_call["causal_mask"] is causal
assert decode_call["return_lse"] is True
assert decode_call["cp_world"] == dcp_world_size
assert decode_call["cp_rank"] == dcp_rank