Multimodal RoPE position calculation
Complex 3D position ID calculation for multi-modal inputs (text, images, videos) that computes temporal, height, and width position indices for rotary embeddings. Handles variable-length sequences with different grid sizes, performs spatial merging calculations, and creates position deltas for rope indexing.
Reported evidence · last observed 2026-06-26. Reported by source; not independently reproduced. No public source. License unknown.
Current records
Workloadsuite of 16 cases · mean latencyimage_token_id = 151655 · video_token_id = 151656 · spatial_merge_size = 2 · vision_start_token_id = 15165216 cases
Not measured on H100 for this workload. Challenges →
12.42µs98% SOL1.00×Reported · license unknown · no source2026-06-26
Measured exactly what you asked. Reported by source; not independently reproduced.
22.63µs98% SOL1.09×Reported · license unknown · no source2026-06-06
Measured exactly what you asked. Reported by source; not independently reproduced.
32.93µs98% SOL1.21×Reported · license unknown · no source2026-07-08
Measured exactly what you asked. Reported by source; not independently reproduced.
43.02µs98% SOL1.25×Reported · license unknown · no source2026-04-14
Measured exactly what you asked. Reported by source; not independently reproduced.
53.77µs97% SOL1.56×Reported · license unknown · no source2026-06-08
Measured exactly what you asked. Reported by source; not independently reproduced.
65.10µs96% SOL2.10×Reported · license unknown · no source2026-05-20
Measured exactly what you asked. Reported by source; not independently reproduced.
75.52µs93% SOL2.28×Reported · license unknown · no source2026-07-07
Measured exactly what you asked. Reported by source; not independently reproduced.
822.6µs85% SOL9.34×Reported · license unknown · no source2026-04-15
Measured exactly what you asked. Reported by source; not independently reproduced.
924.2µs83% SOL9.98×Reported · license unknown · no source2026-04-14
Measured exactly what you asked. Reported by source; not independently reproduced.
1042.9µs74% SOL17.70×Reported · license unknown · no source2026-05-25
Measured exactly what you asked. Reported by source; not independently reproduced.
11260.1µs32% SOL107.29×Reported · license unknown · no source2026-04-15
Measured exactly what you asked. Reported by source; not independently reproduced.
12269.6µs32% SOL111.22×Reported · license unknown · no source2026-05-20
Measured exactly what you asked. Reported by source; not independently reproduced.
13582.6µs19% SOL240.33×Reported · license unknown · no source2026-04-21
Measured exactly what you asked. Reported by source; not independently reproduced.
14584.9µs19% SOL241.29×Reported · license unknown · no source2026-04-21
Measured exactly what you asked. Reported by source; not independently reproduced.
151.33ms9% SOL548.17×Reported · license unknown · no source2026-04-23
Measured exactly what you asked. Reported by source; not independently reproduced.