[NIXL] Use int32 array for indices to avoid intermediate conversion (#51952)

Signed-off-by: Ilia Yastrebov <iyastrebov@nvidia.com>
This commit is contained in:
Ilia Yastrebov
2026-09-02 16:30:16 +02:00
committed by GitHub
parent c6bca6e585
commit 41848caa63
9 changed files with 29 additions and 22 deletions
+1 -1
View File
@@ -527,7 +527,7 @@ class NixlEplbCommunicator(EplbCommunicator):
remote_desc,
)
indices = list(range(len(local_descs)))
indices = np.arange(len(local_descs), dtype=np.int32)
xfer_handle = self._nixl_wrapper.make_prepped_xfer(
"READ",
local_handle,
@@ -141,9 +141,11 @@ class NixlBaseConnectorWorker:
# read across all regions, same for [3], but group0-group1 blocks will
# always differ (different areas). Therefore we can just flatten the
# block_ids and compute the descs ids for all groups at once.
block_arr = np.concatenate(block_ids)[None, :]
region_ids = np.arange(self.num_regions)[:, None]
return (region_ids * num_blocks + block_arr).flatten()
block_arr = np.concatenate(
[np.asarray(g, dtype=np.int32) for g in block_ids]
)[None, :]
region_ids = np.arange(self.num_regions, dtype=np.int32)[:, None]
return (region_ids * num_blocks + block_arr).ravel()
# Compute desc ids per group using the right stride: FA descs have
# num_blocks entries per region (kernel granularity, expanded by
@@ -153,18 +155,18 @@ class NixlBaseConnectorWorker:
logical_blocks = dst_num_blocks // physical_blocks_per_logical
all_descs: list[np.ndarray] = []
for i, group in enumerate(block_ids):
group_arr = np.asarray(group)
group_arr = np.asarray(group, dtype=np.int32)
spec_type = self._group_spec_types[i]
if _is_attention_spec(spec_type):
# A scratch cache lives only in its own regions; every other
# attention group spans all of them.
fa_region_ids = (
np.asarray(self._scratch_region_indices, dtype=np.int64)
np.asarray(self._scratch_region_indices, dtype=np.int32)
if spec_type is CircularBufferSpec
else np.arange(self.num_regions)
else np.arange(self.num_regions, dtype=np.int32)
)[:, None]
all_descs.append(
(fa_region_ids * num_blocks + group_arr[None, :]).flatten()
(fa_region_ids * num_blocks + group_arr[None, :]).ravel()
)
elif _is_ssm_spec(spec_type):
# NOTE (NickLucche) SSM and Attention block regions can
@@ -175,16 +177,16 @@ class NixlBaseConnectorWorker:
# P and D can have different num_blocks (and thus
# different FA desc counts).
ssm_region_ids = (
np.arange(num_ssm_regions, num_ssm_regions + 1)
np.arange(num_ssm_regions, num_ssm_regions + 1, dtype=np.int32)
if i == self._ple_group_index
else np.arange(num_ssm_regions)
else np.arange(num_ssm_regions, dtype=np.int32)
)[:, None]
all_descs.append(
(
ssm_region_ids * logical_blocks
+ group_arr[None, :]
+ num_fa_descs
).flatten()
).ravel()
)
else:
raise ValueError(