MoE sparse expert dispatch
Sparse MoE expert computation with top-k routing across 128 routed experts. Each token is routed to 8 experts via sigmoid-based gating with group-wise selection. Includes SwiGLU activation (gate * up projection) and shared expert computation.
Reported evidence · last observed 2026-08-23. Reported by source; not independently reproduced. No public source. License unknown.
Current records
Workloadsuite of 16 cases · mean latencyn_group = 1 · topk_group = 1 · hidden_size = 4096 · n_routed_experts = 128 · num_experts_per_tok = 8 · shared_intermediate = 1408 · moe_intermediate_size = 140816 cases
Not measured on H100 for this workload. Challenges →
1751.0µs95% SOL1.00×Reported · license unknown · no source2026-08-23
Measured exactly what you asked. Reported by source; not independently reproduced.
2822.5µs94% SOL1.10×Reported · license unknown · no source2026-07-17
Measured exactly what you asked. Reported by source; not independently reproduced.
3884.2µs92% SOL1.18×Reported · license unknown · no source2026-07-30
Measured exactly what you asked. Reported by source; not independently reproduced.
4887.8µs92% SOL1.18×Reported · license unknown · no source2026-07-18
Measured exactly what you asked. Reported by source; not independently reproduced.
5890.4µs92% SOL1.19×Reported · license unknown · no source2026-08-03
Measured exactly what you asked. Reported by source; not independently reproduced.
6910.5µs92% SOL1.21×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
7915.8µs90% SOL1.22×Reported · license unknown · no source2026-07-12
Measured exactly what you asked. Reported by source; not independently reproduced.
8936.0µs91% SOL1.25×Reported · license unknown · no source2026-07-12
Measured exactly what you asked. Reported by source; not independently reproduced.
91.01ms88% SOL1.34×Reported · license unknown · no source2026-08-22
Measured exactly what you asked. Reported by source; not independently reproduced.
101.04ms87% SOL1.39×Reported · license unknown · no source2026-08-21
Measured exactly what you asked. Reported by source; not independently reproduced.
111.07ms86% SOL1.42×Reported · license unknown · no source2026-08-17
Measured exactly what you asked. Reported by source; not independently reproduced.
121.10ms85% SOL1.46×Reported · license unknown · no source2026-08-19
Measured exactly what you asked. Reported by source; not independently reproduced.
131.11ms85% SOL1.48×Reported · license unknown · no source2026-08-18
Measured exactly what you asked. Reported by source; not independently reproduced.
141.12ms85% SOL1.49×Reported · license unknown · no source2026-08-08
Measured exactly what you asked. Reported by source; not independently reproduced.
151.17ms83% SOL1.55×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
161.21ms81% SOL1.61×Reported · license unknown · no source2026-08-03
Measured exactly what you asked. Reported by source; not independently reproduced.
171.92ms65% SOL2.55×Reported · license unknown · no source2026-07-28
Measured exactly what you asked. Reported by source; not independently reproduced.
182.12ms62% SOL2.82×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
192.26ms60% SOL3.01×Reported · license unknown · no source2026-07-23
Measured exactly what you asked. Reported by source; not independently reproduced.
203.18ms48% SOL4.24×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
213.44ms46% SOL4.59×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.