minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

Revert "Increase the KV cache pool when using indexShare by 15% (#30310)" (#30472)

Browse Source
This commit is contained in:
Mohammad Miadh Angkad
2026-07-07 23:48:35 -07:00
committed by GitHub
parent 33c3dfd7e0
commit fda87173ab
3 changed files with 11 additions and 44 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/sglang/srt/mem_cache/memory_pool.py
+10 -28
View File
@@ -3137,7 +3137,6 @@ class DSATokenToKVPool(MLATokenToKVPool):
start_layer: Optional[int] = None,
end_layer: Optional[int] = None,
index_buf_size: Optional[int] = None,
skip_topk_layers: Optional[List[bool]] = None,
):
override_dim = (
kv_cache_dim if kv_cache_dim != kv_lora_rank + qk_rope_head_dim else None
@@ -3165,13 +3164,6 @@ class DSATokenToKVPool(MLATokenToKVPool):
# num head == 1 and head dim == 128 for index_k in DSA
assert index_head_dim == 128
self.skip_topk_layers = (
list(skip_topk_layers)
if skip_topk_layers is not None
else [False] * layer_num
)
assert len(self.skip_topk_layers) == layer_num
if _is_hip:
if aiter_can_use_preshuffle_paged_mqa():
assert (
@@ -3188,10 +3180,6 @@ class DSATokenToKVPool(MLATokenToKVPool):
if self.custom_mem_pool
else nullcontext()
):
cols = self.page_size * (
index_head_dim + index_head_dim // self.quant_block_size * 4
)
num_pages = (index_buf_size + page_size + 1) // self.page_size
self.index_k_with_scale_buffer = [
torch.zeros(
# Layout:
@@ -3200,11 +3188,17 @@ class DSATokenToKVPool(MLATokenToKVPool):
# data: for page i,
# * buf[i, :page_size * head_dim] for fp8 data
# * buf[i, page_size * head_dim:].view(float32) for scale
(0 if self.skip_topk_layers[i] else num_pages, cols),
(
(index_buf_size + page_size + 1) // self.page_size,
self.page_size
* (
index_head_dim + index_head_dim // self.quant_block_size * 4
),
),
dtype=self.index_k_with_scale_buffer_dtype,
device=device,
)
for i in range(layer_num)
for _ in range(layer_num)
]
self._finalize_allocation_log(size)
@@ -3221,9 +3215,7 @@ class DSATokenToKVPool(MLATokenToKVPool):
tgt_loc_flat = tgt_loc.view(-1).long()
src_loc_flat = src_loc.view(-1).long()
for i, index_k in enumerate(self.index_k_with_scale_buffer):
if self.skip_topk_layers[i]:
continue
for index_k in self.index_k_with_scale_buffer:
index_k[tgt_loc_flat] = index_k[src_loc_flat]
def get_index_k_with_scale_buffer(self, layer_id: int) -> torch.Tensor:
@@ -3314,9 +3306,6 @@ class DSATokenToKVPool(MLATokenToKVPool):
chunk_size = self.cpu_offloading_chunk_size
page_chunk_size = max(1, chunk_size // self.page_size)
for layer_id in range(self.layer_num):
if self.skip_topk_layers[layer_id]:
index_k_cpu.append([])
continue
index_k_cpu.append([])
for i in range(0, len(page_indices), page_chunk_size):
chunk_page_indices = page_indices[i : i + page_chunk_size]
@@ -3339,8 +3328,6 @@ class DSATokenToKVPool(MLATokenToKVPool):
chunk_size = self.cpu_offloading_chunk_size
page_chunk_size = max(1, chunk_size // self.page_size)
for layer_id in range(self.layer_num):
if self.skip_topk_layers[layer_id]:
continue
for i in range(0, len(page_indices), page_chunk_size):
chunk_page_indices = page_indices[i : i + page_chunk_size]
idx_cpu = index_k_cpu[layer_id][i // page_chunk_size]
@@ -3359,12 +3346,7 @@ class DSATokenToKVPool(MLATokenToKVPool):
self.index_k_with_scale_buffer[i].nbytes for i in range(self.layer_num)
]
item_lens = [
(
0
if self.skip_topk_layers[i]
else self.index_k_with_scale_buffer[i][0].nbytes
)
for i in range(self.layer_num)
self.index_k_with_scale_buffer[i][0].nbytes for i in range(self.layer_num)
]
return data_ptrs, data_lens, item_lens
python/sglang/srt/model_executor/model_runner_kv_cache_mixin.py
-6
View File
@@ -7,7 +7,6 @@ from typing import TYPE_CHECKING, Optional
import torch
from sglang.srt.configs.model_config import (
dsa_layer_skips_topk,
get_dsa_index_head_dim,
get_minimax_sparse_attention_config,
get_minimax_sparse_disable_value_layer_ids,
@@ -958,11 +957,6 @@ class ModelRunnerKVCacheMixin:
pool_kwargs["host_to_device_ratio"] = parse_hisparse_config(
self.server_args
).host_to_device_ratio
elif not self.is_draft_worker:
pool_kwargs["skip_topk_layers"] = [
dsa_layer_skips_topk(self.model_config.hf_config, layer_id)
for layer_id in range(self.start_layer, self.end_layer)
]
self.token_to_kv_pool = PoolCls(
self.max_total_num_tokens,
page_size=self.page_size,
python/sglang/srt/model_executor/pool_configurator.py
+1 -10
View File
@@ -20,7 +20,6 @@ from typing import TYPE_CHECKING, Optional
import torch
from sglang.srt.configs.model_config import (
dsa_layer_skips_topk,
get_dsa_index_head_dim,
get_minimax_sparse_attention_config,
get_minimax_sparse_disable_value_layer_ids,
@@ -210,15 +209,7 @@ class DefaultPoolConfigurator(MemoryPoolConfigurator):
element_size = torch._utils._element_size(
DSATokenToKVPool.index_k_with_scale_buffer_dtype
)
if mr.enable_hisparse or mr.is_draft_worker:
num_indexer_layers = num_layers
else:
num_indexer_layers = sum(
1
for layer_id in range(mr.start_layer, mr.end_layer)
if not dsa_layer_skips_topk(model_config.hf_config, layer_id)
)
cell_size += indexer_size_per_token * num_indexer_layers * element_size
cell_size += indexer_size_per_token * num_layers * element_size
elif is_minimax_sparse(model_config.hf_config):
# Mirrors MiniMaxSparseKVPool: main pool (K+V all layers) + indexer pool
# (sparse-only, single-head; kv layers store K+V, k-only layers store K).