Skip to content
KernelIndex
Search⌘K

Group limited topk routing

22 eligible runs
moe

Group-limited top-k expert routing for Mixture of Experts. Implements a two-stage selection: first organizes 256 experts into 8 groups, computes group scores by summing top-2 expert scores within each group, selects top-4 groups, then picks top-8 experts from selected groups. Returns selected expert indices and normalized routing weights scaled by routed_scaling_factor.

Fastest reported
21.1µsmean · 88% of SOL
Amir M. Mir | SF TensorLicense unknown · unknown

Reported evidence · last observed 2026-08-01. Reported by source; not independently reproduced. No public source. License unknown.

No public sourceRun detail →

Current records

Workloadsuite of 16 cases · mean latencyn_group = 8 · gating_dim = 4096 · topk_group = 4 · num_experts = 256 · experts_per_group = 32 · num_experts_per_tok = 816 cases

Not measured on H100 for this workload. Challenges →

Source-native comparison · GPU NVIDIA B200 · Workload suite of 16 cases · mean latency · Protocol SOL-ExecBench evaluation stack v1.1 · mean · 14 results · last observed 2026-08-15Record history →
#
Implementation
Latency
vs #1
Trust
Observed
1
21.1µs88% SOL
1.00×
Reported · license unknown · no source
2026-08-01

Measured exactly what you asked. Reported by source; not independently reproduced.

no sourcelicense unknownno install recipeRun detail →
2
21.5µs88% SOL
1.02×
Reported · license unknown · no source
2026-08-15

Measured exactly what you asked. Reported by source; not independently reproduced.

no sourcelicense unknownno install recipeRun detail →
3
21.6µs88% SOL
1.02×
Reported · license unknown · no source
2026-08-15

Measured exactly what you asked. Reported by source; not independently reproduced.

no sourcelicense unknownno install recipeRun detail →
4
21.6µs88% SOL
1.02×
Reported · license unknown · no source
2026-08-15

Measured exactly what you asked. Reported by source; not independently reproduced.

no sourcelicense unknownno install recipeRun detail →
5
21.6µs88% SOL
1.02×
Reported · license unknown · no source
2026-07-23

Measured exactly what you asked. Reported by source; not independently reproduced.

no sourcelicense unknownno install recipeRun detail →
6
21.8µs88% SOL
1.03×
Reported · license unknown · no source
2026-07-17

Measured exactly what you asked. Reported by source; not independently reproduced.

no sourcelicense unknownno install recipeRun detail →
7
22.8µs87% SOL
1.08×
Reported · license unknown · no source
2026-07-16

Measured exactly what you asked. Reported by source; not independently reproduced.

no sourcelicense unknownno install recipeRun detail →
8
22.9µs87% SOL
1.08×
Reported · license unknown · no source
2026-07-25

Measured exactly what you asked. Reported by source; not independently reproduced.

no sourcelicense unknownno install recipeRun detail →
9
24.2µs86% SOL
1.14×
Reported · license unknown · no source
2026-07-10

Measured exactly what you asked. Reported by source; not independently reproduced.

no sourcelicense unknownno install recipeRun detail →
10
24.6µs86% SOL
1.16×
Reported · license unknown · no source
2026-07-10

Measured exactly what you asked. Reported by source; not independently reproduced.

no sourcelicense unknownno install recipeRun detail →
11
27.9µs84% SOL
1.32×
Reported · license unknown · no source
2026-07-10

Measured exactly what you asked. Reported by source; not independently reproduced.

no sourcelicense unknownno install recipeRun detail →
12
28.2µs84% SOL
1.34×
Reported · license unknown · no source
2026-07-10

Measured exactly what you asked. Reported by source; not independently reproduced.

no sourcelicense unknownno install recipeRun detail →
13
55.3µs71% SOL
2.62×
Reported · license unknown · no source
2026-07-10

Measured exactly what you asked. Reported by source; not independently reproduced.

no sourcelicense unknownno install recipeRun detail →
14
148.9µs46% SOL
7.04×
Reported · license unknown · no source
2026-07-10

Measured exactly what you asked. Reported by source; not independently reproduced.

no sourcelicense unknownno install recipeRun detail →

Implementations

Implementation
Runtime
Best latency
Evidence
Availability
unknown
21.5µs
1.03×
Reported
license unknown · no source
unknown
21.6µs
1.04×
Reported
license unknown · no source
unknown
21.6µs
1.04×
Reported
license unknown · no source
unknown
21.6µs
1.04×
Reported
license unknown · no source
unknown
21.8µs
1.05×
Reported
license unknown · no source
unknown
22.8µs
1.10×
Reported
license unknown · no source
unknown
27.6µs
1.33×
Reported
license unknown · no source
unknown
55.3µs
2.66×
Reported
license unknown · no source
Show all 22 implementations ›
unknown
109.8µs
5.28×
Reported
license unknown · no source
unknown
22.9µs
1.10×
Reported
license unknown · no source
unknown
24.2µs
1.16×
Reported
license unknown · no source
unknown
24.6µs
1.18×
Reported
license unknown · no source
unknown
27.9µs
1.34×
Reported
license unknown · no source
unknown
28.2µs
1.36×
Reported
license unknown · no source
unknown
28.2µs
1.36×
Reported
license unknown · no source
unknown
62.3µs
2.99×
Reported
license unknown · no source
unknown
148.9µs
7.16×
Reported
license unknown · no source
unknown
182.6µs
8.78×
Reported
license unknown · no source
unknown
248.3µs
11.95×
Reported
license unknown · no source
unknown
20.8µs
1.00×
Reported
license unknown · no source
unknown
20.9µs
1.01×
Reported
license unknown · no source
unknown
21.1µs
1.02×
Reported
license unknown · no source

Semantics

Inputs and outputs
hidden_statesbf16 [num_tokens, gating_dim]
weightbf16 [num_experts, gating_dim]
expert_biasbf16 [num_experts]
routed_scaling_factorfp32 scalar
topk_idxint64 [num_tokens, num_experts_per_tok]
topk_weightfp32 [num_tokens, num_experts_per_tok]
Axes and behavior
n_groupconstant = 8
gating_dimconstant = 4096
num_tokensvariable
topk_groupconstant = 4
num_expertsconstant = 256
experts_per_groupexperts_per_group = num_experts // n_group
num_experts_per_tokconstant = 8
determinismunspecified
constraintsNo mutation or aliasing
Identity
alias049_group_limited_topk_routingmodelring-flash-2-0sha2561bdb6c1569e6…
Sources: NVIDIA SOL-ExecBench (2026-08-15)last observed 2026-08-15How records are decidedJSON