Fused cos sin embedding generation
Fused cos/sin embedding generation from frequencies for RoPE. Takes frequency tensor (result of inv_freq @ position_ids) and generates cos/sin embeddings with attention scaling. Concatenates frequencies with themselves, computes cos and sin with scaling in a single pass.
Reported evidence · last observed 2026-07-08. Reported by source; not independently reproduced. No public source. License unknown.
Current records
Workloadsuite of 16 cases · mean latencyhead_dim = 128 · half_head_dim = 6416 cases
Not measured on H100 for this workload. Challenges →
12.54µs92% SOL1.00×Reported · license unknown · no source2026-07-08
Measured exactly what you asked. Reported by source; not independently reproduced.
22.79µs91% SOL1.10×Reported · license unknown · no source2026-06-25
Measured exactly what you asked. Reported by source; not independently reproduced.
32.81µs91% SOL1.11×Reported · license unknown · no source2026-07-02
Measured exactly what you asked. Reported by source; not independently reproduced.
42.83µs91% SOL1.12×Reported · license unknown · no source2026-06-26
Measured exactly what you asked. Reported by source; not independently reproduced.
53.01µs91% SOL1.19×Reported · license unknown · no source2026-05-06
Measured exactly what you asked. Reported by source; not independently reproduced.
63.07µs90% SOL1.21×Reported · license unknown · no source2026-05-28
Measured exactly what you asked. Reported by source; not independently reproduced.
73.12µs90% SOL1.23×Reported · license unknown · no source2026-07-08
Measured exactly what you asked. Reported by source; not independently reproduced.
83.12µs90% SOL1.23×Reported · license unknown · no source2026-05-19
Measured exactly what you asked. Reported by source; not independently reproduced.
93.12µs90% SOL1.23×Reported · license unknown · no source2026-07-01
Measured exactly what you asked. Reported by source; not independently reproduced.
103.13µs90% SOL1.23×Reported · license unknown · no source2026-07-01
Measured exactly what you asked. Reported by source; not independently reproduced.
113.14µs90% SOL1.24×Reported · license unknown · no source2026-05-19
Measured exactly what you asked. Reported by source; not independently reproduced.
123.18µs90% SOL1.25×Reported · license unknown · no source2026-04-15
Measured exactly what you asked. Reported by source; not independently reproduced.
133.28µs89% SOL1.29×Reported · license unknown · no source2026-06-06
Measured exactly what you asked. Reported by source; not independently reproduced.
143.34µs89% SOL1.32×Reported · license unknown · no source2026-07-07
Measured exactly what you asked. Reported by source; not independently reproduced.
153.35µs89% SOL1.32×Reported · license unknown · no source2026-07-07
Measured exactly what you asked. Reported by source; not independently reproduced.
163.35µs89% SOL1.32×Reported · license unknown · no source2026-07-08
Measured exactly what you asked. Reported by source; not independently reproduced.
173.37µs89% SOL1.33×Reported · license unknown · no source2026-07-07
Measured exactly what you asked. Reported by source; not independently reproduced.
183.41µs89% SOL1.34×Reported · license unknown · no source2026-07-07
Measured exactly what you asked. Reported by source; not independently reproduced.
193.47µs89% SOL1.37×Reported · license unknown · no source2026-04-14
Measured exactly what you asked. Reported by source; not independently reproduced.
203.55µs88% SOL1.40×Reported · license unknown · no source2026-07-01
Measured exactly what you asked. Reported by source; not independently reproduced.
213.58µs88% SOL1.41×Reported · license unknown · no source2026-05-28
Measured exactly what you asked. Reported by source; not independently reproduced.
223.59µs88% SOL1.41×Reported · license unknown · no source2026-06-14
Measured exactly what you asked. Reported by source; not independently reproduced.
233.86µs87% SOL1.52×Reported · license unknown · no source2026-04-15
Measured exactly what you asked. Reported by source; not independently reproduced.
243.91µs87% SOL1.54×Reported · license unknown · no source2026-05-06
Measured exactly what you asked. Reported by source; not independently reproduced.
24=3.91µs87% SOL1.54×Reported · license unknown · no source2026-04-15
Measured exactly what you asked; tied ranks share a number. Reported by source; not independently reproduced.
254.40µs85% SOL1.73×Reported · license unknown · no source2026-04-15
Measured exactly what you asked. Reported by source; not independently reproduced.
264.40µs85% SOL1.73×Reported · license unknown · no source2026-04-15
Measured exactly what you asked. Reported by source; not independently reproduced.
275.01µs83% SOL1.97×Reported · license unknown · no source2026-04-22
Measured exactly what you asked. Reported by source; not independently reproduced.
289.90µs70% SOL3.90×Reported · license unknown · no source2026-04-15
Measured exactly what you asked. Reported by source; not independently reproduced.
2937.7µs36% SOL14.85×Reported · license unknown · no source2026-04-15
Measured exactly what you asked. Reported by source; not independently reproduced.