Skip to content

Commit 4b2448d

Browse files
committed
Fix wide-EP sweep configs: enable multi-frontend, remove buggy load-balance-method, fix max-running-requests
1 parent 562aa3f commit 4b2448d

5 files changed

Lines changed: 11 additions & 11 deletions

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-gb300-10p1d-dep4-dep32-18-c2500.yaml

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -32,7 +32,8 @@ resources:
3232

3333
frontend:
3434
type: dynamo
35-
enable_multiple_frontends: false
35+
enable_multiple_frontends: true
36+
num_additional_frontends: 8
3637
env:
3738
DYN_ROUTER_LOAD_BLOCK_SIZE: "1"
3839
args:
@@ -127,7 +128,6 @@ backend:
127128
skip-tokenizer-init: true
128129
stream-interval: 60
129130

130-
load-balance-method: "total_requests"
131131
moe-a2a-backend: "megamoe"
132132

133133
moe-dense-tp-size: 1

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-gb300-12p1d-dep4-dep24-18-c3000.yaml

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -32,7 +32,8 @@ resources:
3232

3333
frontend:
3434
type: dynamo
35-
enable_multiple_frontends: false
35+
enable_multiple_frontends: true
36+
num_additional_frontends: 8
3637
env:
3738
DYN_ROUTER_LOAD_BLOCK_SIZE: "1"
3839
args:
@@ -127,7 +128,6 @@ backend:
127128
skip-tokenizer-init: true
128129
stream-interval: 60
129130

130-
load-balance-method: "total_requests"
131131
moe-a2a-backend: "megamoe"
132132

133133
moe-dense-tp-size: 1

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-gb300-14p1d-dep4-dep16-18-c8192.yaml

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -32,7 +32,8 @@ resources:
3232

3333
frontend:
3434
type: dynamo
35-
enable_multiple_frontends: false
35+
enable_multiple_frontends: true
36+
num_additional_frontends: 8
3637
env:
3738
DYN_ROUTER_LOAD_BLOCK_SIZE: "1"
3839
args:
@@ -127,7 +128,6 @@ backend:
127128
skip-tokenizer-init: true
128129
stream-interval: 60
129130

130-
load-balance-method: "total_requests"
131131
moe-a2a-backend: "megamoe"
132132

133133
moe-dense-tp-size: 1

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-gb300-15p1d-dep4-dep12-18-c12000.yaml

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -32,7 +32,8 @@ resources:
3232

3333
frontend:
3434
type: dynamo
35-
enable_multiple_frontends: false
35+
enable_multiple_frontends: true
36+
num_additional_frontends: 8
3637
env:
3738
DYN_ROUTER_LOAD_BLOCK_SIZE: "1"
3839
args:
@@ -127,7 +128,6 @@ backend:
127128
skip-tokenizer-init: true
128129
stream-interval: 60
129130

130-
load-balance-method: "total_requests"
131131
moe-a2a-backend: "megamoe"
132132

133133
moe-dense-tp-size: 1

benchmarks/multi_node/srt-slurm-recipes/sglang/deepseek-v4/8k1k/disagg-gb300-8p1d-dep4-dep40-18-c2048.yaml

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -32,7 +32,8 @@ resources:
3232

3333
frontend:
3434
type: dynamo
35-
enable_multiple_frontends: false
35+
enable_multiple_frontends: true
36+
num_additional_frontends: 8
3637
env:
3738
DYN_ROUTER_LOAD_BLOCK_SIZE: "1"
3839
args:
@@ -127,7 +128,6 @@ backend:
127128
skip-tokenizer-init: true
128129
stream-interval: 60
129130

130-
load-balance-method: "total_requests"
131131
moe-a2a-backend: "megamoe"
132132

133133
moe-dense-tp-size: 1
@@ -145,7 +145,7 @@ backend:
145145
ep-num-redundant-experts: 16
146146
enable-dp-attention: true
147147
enable-dp-lm-head: true
148-
max-running-requests: 18432
148+
max-running-requests: 18400
149149
cuda-graph-max-bs: 1280
150150

151151

0 commit comments

Comments
 (0)