MoE expert token radix sort with prefix sum
MoE token sorting and bucketing using radix sort with prefix sum. Performs O(N) counting sort on bounded expert indices (0-255), computes prefix sum for cumulative offsets, and returns sorted token indices and expert boundaries. Critical for efficient MoE token-to-expert routing.
Reported evidence · last observed 2026-07-16. Reported by source; not independently reproduced. No public source. License unknown.
Current records
Workloadsuite of 16 cases · mean latencynum_experts = 256 · num_experts_per_tok = 8 · num_experts_plus_one = 25716 cases
Not measured on H100 for this workload. Challenges →
15.13µs92% SOL1.00×Reported · license unknown · no source2026-07-16
Measured exactly what you asked. Reported by source; not independently reproduced.
26.00µs90% SOL1.17×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
36.43µs90% SOL1.25×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
46.50µs90% SOL1.27×Reported · license unknown · no source2026-08-24
Measured exactly what you asked. Reported by source; not independently reproduced.
56.52µs90% SOL1.27×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
66.75µs89% SOL1.32×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
77.10µs89% SOL1.38×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
87.54µs88% SOL1.47×Reported · license unknown · no source2026-07-14
Measured exactly what you asked. Reported by source; not independently reproduced.
97.57µs88% SOL1.47×Reported · license unknown · no source2026-08-18
Measured exactly what you asked. Reported by source; not independently reproduced.
108.23µs87% SOL1.60×Reported · license unknown · no source2026-08-14
Measured exactly what you asked. Reported by source; not independently reproduced.
118.29µs87% SOL1.61×Reported · license unknown · no source2026-08-17
Measured exactly what you asked. Reported by source; not independently reproduced.
128.30µs87% SOL1.62×Reported · license unknown · no source2026-07-14
Measured exactly what you asked. Reported by source; not independently reproduced.
1310.5µs84% SOL2.05×Reported · license unknown · no source2026-07-23
Measured exactly what you asked. Reported by source; not independently reproduced.
1410.9µs83% SOL2.13×Reported · license unknown · no source2026-07-25
Measured exactly what you asked. Reported by source; not independently reproduced.
1511.2µs83% SOL2.19×Reported · license unknown · no source2026-08-08
Measured exactly what you asked. Reported by source; not independently reproduced.
1612.1µs82% SOL2.35×Reported · license unknown · no source2026-08-12
Measured exactly what you asked. Reported by source; not independently reproduced.
1718.2µs75% SOL3.55×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.