Skip to content
Closed
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@ name: "b200-fp4-mtp2-throughput-1p-dep4-1d-dep8"
# layout (sglang/<model>/<hw>-<precision>/<seq>/disagg/<variant>/...).

dynamo:
hash: "5b4bc1dd70965017a737c71b19db5a0aeaa88727"
wheel: "1.5.0.dev20260910"
install: true

frontend:
Expand All @@ -17,7 +17,7 @@ frontend:

model:
path: "dsr1"
container: "lmsysorg/sglang:v0.5.12.post1"
container: "lmsysorg/sglang:v0.5.19-cu130"
precision: "fp4"

resources:
Expand Down Expand Up @@ -92,7 +92,7 @@ backend:
expert-parallel-size: 4
enable-dp-attention: true
enable-dp-lm-head: true
fp4-gemm-backend: "flashinfer_trtllm"
fp4-gemm-backend: "auto"
disaggregation-transfer-backend: nixl
speculative-algorithm: "EAGLE"
speculative-num-steps: 2
Expand Down Expand Up @@ -123,7 +123,7 @@ backend:
expert-parallel-size: 8
enable-dp-attention: true
enable-dp-lm-head: true
fp4-gemm-backend: "flashinfer_trtllm"
fp4-gemm-backend: "auto"
disaggregation-transfer-backend: nixl
speculative-algorithm: "EAGLE"
speculative-num-steps: 2
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@ name: "b200-fp4-mtp2-throughput-2p-dep4-1d-dep8"
# layout (sglang/<model>/<hw>-<precision>/<seq>/disagg/<variant>/...).

dynamo:
hash: "5b4bc1dd70965017a737c71b19db5a0aeaa88727"
wheel: "1.5.0.dev20260910"
install: true

frontend:
Expand All @@ -17,7 +17,7 @@ frontend:

model:
path: "dsr1"
container: "lmsysorg/sglang:v0.5.12.post1"
container: "lmsysorg/sglang:v0.5.19-cu130"
precision: "fp4"

resources:
Expand Down Expand Up @@ -92,7 +92,7 @@ backend:
expert-parallel-size: 4
enable-dp-attention: true
enable-dp-lm-head: true
fp4-gemm-backend: "flashinfer_trtllm"
fp4-gemm-backend: "auto"
disaggregation-transfer-backend: nixl
speculative-algorithm: "EAGLE"
speculative-num-steps: 2
Expand Down Expand Up @@ -123,7 +123,7 @@ backend:
expert-parallel-size: 8
enable-dp-attention: true
enable-dp-lm-head: true
fp4-gemm-backend: "flashinfer_trtllm"
fp4-gemm-backend: "auto"
disaggregation-transfer-backend: nixl
speculative-algorithm: "EAGLE"
speculative-num-steps: 2
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@ name: "b200-fp4-mtp2-throughput-3p-dep4-1d-dep8"
# layout (sglang/<model>/<hw>-<precision>/<seq>/disagg/<variant>/...).

dynamo:
hash: "5b4bc1dd70965017a737c71b19db5a0aeaa88727"
wheel: "1.5.0.dev20260910"
install: true

frontend:
Expand All @@ -17,7 +17,7 @@ frontend:

model:
path: "dsr1"
container: "lmsysorg/sglang:v0.5.12.post1"
container: "lmsysorg/sglang:v0.5.19-cu130"
precision: "fp4"

resources:
Expand Down Expand Up @@ -92,7 +92,7 @@ backend:
expert-parallel-size: 4
enable-dp-attention: true
enable-dp-lm-head: true
fp4-gemm-backend: "flashinfer_trtllm"
fp4-gemm-backend: "auto"
disaggregation-transfer-backend: nixl
speculative-algorithm: "EAGLE"
speculative-num-steps: 2
Expand Down Expand Up @@ -123,7 +123,7 @@ backend:
expert-parallel-size: 8
enable-dp-attention: true
enable-dp-lm-head: true
fp4-gemm-backend: "flashinfer_trtllm"
fp4-gemm-backend: "auto"
disaggregation-transfer-backend: nixl
speculative-algorithm: "EAGLE"
speculative-num-steps: 2
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@ name: "b200-fp4-mtp2-throughput-4p-dep4-1d-dep8"
# layout (sglang/<model>/<hw>-<precision>/<seq>/disagg/<variant>/...).

dynamo:
hash: "5b4bc1dd70965017a737c71b19db5a0aeaa88727"
wheel: "1.5.0.dev20260910"
install: true

frontend:
Expand All @@ -17,7 +17,7 @@ frontend:

model:
path: "dsr1"
container: "lmsysorg/sglang:v0.5.12.post1"
container: "lmsysorg/sglang:v0.5.19-cu130"
precision: "fp4"

resources:
Expand Down Expand Up @@ -92,7 +92,7 @@ backend:
expert-parallel-size: 4
enable-dp-attention: true
enable-dp-lm-head: true
fp4-gemm-backend: "flashinfer_trtllm"
fp4-gemm-backend: "auto"
disaggregation-transfer-backend: nixl
speculative-algorithm: "EAGLE"
speculative-num-steps: 2
Expand Down Expand Up @@ -123,7 +123,7 @@ backend:
expert-parallel-size: 8
enable-dp-attention: true
enable-dp-lm-head: true
fp4-gemm-backend: "flashinfer_trtllm"
fp4-gemm-backend: "auto"
disaggregation-transfer-backend: nixl
speculative-algorithm: "EAGLE"
speculative-num-steps: 2
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@ name: "b200-fp4-mtp2-throughput-5p-dep4-1d-dep8"
# layout (sglang/<model>/<hw>-<precision>/<seq>/disagg/<variant>/...).

dynamo:
hash: "5b4bc1dd70965017a737c71b19db5a0aeaa88727"
wheel: "1.5.0.dev20260910"
install: true

frontend:
Expand All @@ -17,7 +17,7 @@ frontend:

model:
path: "dsr1"
container: "lmsysorg/sglang:v0.5.12.post1"
container: "lmsysorg/sglang:v0.5.19-cu130"
precision: "fp4"

resources:
Expand Down Expand Up @@ -92,7 +92,7 @@ backend:
expert-parallel-size: 4
enable-dp-attention: true
enable-dp-lm-head: true
fp4-gemm-backend: "flashinfer_trtllm"
fp4-gemm-backend: "auto"
disaggregation-transfer-backend: nixl
speculative-algorithm: "EAGLE"
speculative-num-steps: 2
Expand Down Expand Up @@ -123,7 +123,7 @@ backend:
expert-parallel-size: 8
enable-dp-attention: true
enable-dp-lm-head: true
fp4-gemm-backend: "flashinfer_trtllm"
fp4-gemm-backend: "auto"
disaggregation-transfer-backend: nixl
speculative-algorithm: "EAGLE"
speculative-num-steps: 2
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@
name: b200-fp4-dsr1_8k1k_mtp_4p1d_c2048
model:
path: "dsr1"
container: "lmsysorg/sglang:v0.5.12.post1"
container: "lmsysorg/sglang:v0.5.19-cu130"
precision: fp4
resources:
gpu_type: b200
Expand All @@ -22,7 +22,7 @@ resources:
frontend:
type: dynamo
dynamo:
hash: "5b4bc1dd70965017a737c71b19db5a0aeaa88727"
wheel: "1.5.0.dev20260910"
install: true
backend:
prefill_environment: &env
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@ name: "b200-fp4-mtp-low-latency-1p-tp4-5d-tp8"
# layout (sglang/<model>/<hw>-<precision>/<seq>/disagg/<variant>/...).

dynamo:
hash: "5b4bc1dd70965017a737c71b19db5a0aeaa88727"
wheel: "1.5.0.dev20260910"
install: true

frontend:
Expand All @@ -17,7 +17,7 @@ frontend:

model:
path: "dsr1"
container: "lmsysorg/sglang:v0.5.12.post1"
container: "lmsysorg/sglang:v0.5.19-cu130"
precision: "fp4"

resources:
Expand Down Expand Up @@ -91,7 +91,7 @@ backend:
tensor-parallel-size: 4
expert-parallel-size: 1
enable-dp-attention: false
fp4-gemm-backend: "flashinfer_trtllm"
fp4-gemm-backend: "auto"
disaggregation-transfer-backend: nixl
speculative-algorithm: "EAGLE"
speculative-num-steps: 3
Expand Down Expand Up @@ -120,7 +120,7 @@ backend:
tensor-parallel-size: 8
expert-parallel-size: 1
enable-dp-attention: false
fp4-gemm-backend: "flashinfer_trtllm"
fp4-gemm-backend: "auto"
disaggregation-transfer-backend: nixl
speculative-algorithm: "EAGLE"
speculative-num-steps: 3
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@ name: "b200-fp4-mtp-low-latency-1p-tp4-3d-tp8"
# layout (sglang/<model>/<hw>-<precision>/<seq>/disagg/<variant>/...).

dynamo:
hash: "5b4bc1dd70965017a737c71b19db5a0aeaa88727"
wheel: "1.5.0.dev20260910"
install: true

frontend:
Expand All @@ -17,7 +17,7 @@ frontend:

model:
path: "dsr1"
container: "lmsysorg/sglang:v0.5.12.post1"
container: "lmsysorg/sglang:v0.5.19-cu130"
precision: "fp4"

resources:
Expand Down Expand Up @@ -91,7 +91,7 @@ backend:
tensor-parallel-size: 4
expert-parallel-size: 1
enable-dp-attention: false
fp4-gemm-backend: "flashinfer_trtllm"
fp4-gemm-backend: "auto"
disaggregation-transfer-backend: nixl
speculative-algorithm: "EAGLE"
speculative-num-steps: 3
Expand Down Expand Up @@ -120,7 +120,7 @@ backend:
tensor-parallel-size: 8
expert-parallel-size: 1
enable-dp-attention: false
fp4-gemm-backend: "flashinfer_trtllm"
fp4-gemm-backend: "auto"
disaggregation-transfer-backend: nixl
speculative-algorithm: "EAGLE"
speculative-num-steps: 3
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@ name: "b200-fp4-mtp-low-latency-1p-tp4-1d-tp8"
# layout (sglang/<model>/<hw>-<precision>/<seq>/disagg/<variant>/...).

dynamo:
hash: "5b4bc1dd70965017a737c71b19db5a0aeaa88727"
wheel: "1.5.0.dev20260910"
install: true

frontend:
Expand All @@ -17,7 +17,7 @@ frontend:

model:
path: "dsr1"
container: "lmsysorg/sglang:v0.5.12.post1"
container: "lmsysorg/sglang:v0.5.19-cu130"
precision: "fp4"

resources:
Expand Down Expand Up @@ -91,7 +91,7 @@ backend:
tensor-parallel-size: 4
expert-parallel-size: 1
enable-dp-attention: false
fp4-gemm-backend: "flashinfer_trtllm"
fp4-gemm-backend: "auto"
disaggregation-transfer-backend: nixl
speculative-algorithm: "EAGLE"
speculative-num-steps: 3
Expand Down Expand Up @@ -120,7 +120,7 @@ backend:
tensor-parallel-size: 8
expert-parallel-size: 1
enable-dp-attention: false
fp4-gemm-backend: "flashinfer_trtllm"
fp4-gemm-backend: "auto"
disaggregation-transfer-backend: nixl
speculative-algorithm: "EAGLE"
speculative-num-steps: 3
Expand Down
4 changes: 2 additions & 2 deletions configs/nvidia-master.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -3990,13 +3990,13 @@ dsr1-fp8-b200-dynamo-sglang-mtp:
dp-attn: true

dsr1-fp4-b200-dynamo-sglang-mtp:
image: "lmsysorg/sglang:v0.5.12.post1"
image: "lmsysorg/sglang:v0.5.19-cu130"
model: deepseek-r1-fp4
model-prefix: dsr1
runner: cluster:b200-nscale
precision: fp4
framework: dynamo-sglang
router: { name: dynamo-router, version: "5b4bc1dd70965017a737c71b19db5a0aeaa88727" }
router: { name: dynamo-router, version: "1.5.0.dev20260910" }
kv-p2p-transfer: nixl
multinode: true
disagg: true
Expand Down