Skip to content
KernelIndex
Search⌘K

Vision rotary position embedding generation backward

40 eligible runs
rope

Backward pass for vision rotary position embedding generation. Computes gradients through cos/sin operations, concatenation, indexing, and outer product to produce gradient for inverse frequencies.

Fastest reported
3.44µsmean · 95% of SOL
RIAC as wellLicense unknown · unknown

Reported evidence · last observed 2026-08-21. Reported by source; not independently reproduced. No public source. License unknown.

No public sourceRun detail →

Current records

Workloadsuite of 16 cases · mean latencyhead_dim = 72 · head_dim_half = 36 · head_dim_quarter = 1816 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 · 26 results · last observed 2026-08-22Record history →
#
Implementation
Latency
vs #1
Trust
Observed
1
3.44µs95% SOL
1.00×
Reported · license unknown · no source
2026-08-21

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

no sourcelicense unknownno install recipeRun detail →
2
3.92µs95% SOL
1.14×
Reported · license unknown · no source
2026-07-11

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

no sourcelicense unknownno install recipeRun detail →
3
5.01µs93% SOL
1.46×
Reported · license unknown · no source
2026-08-06

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

no sourcelicense unknownno install recipeRun detail →
4
5.05µs93% SOL
1.47×
Reported · license unknown · no source
2026-08-06

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

no sourcelicense unknownno install recipeRun detail →
5
5.43µs93% SOL
1.58×
Reported · license unknown · no source
2026-07-15

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

no sourcelicense unknownno install recipeRun detail →
6
5.46µs91% SOL
1.59×
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 →
7
5.55µs92% SOL
1.62×
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 →
8
5.69µs92% SOL
1.66×
Reported · license unknown · no source
2026-08-22

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

no sourcelicense unknownno install recipeRun detail →
9
6.11µs92% SOL
1.78×
Reported · license unknown · no source
2026-08-04

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

no sourcelicense unknownno install recipeRun detail →
10
6.52µs91% SOL
1.90×
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
6.62µs89% SOL
1.93×
Reported · license unknown · no source
2026-07-20

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

no sourcelicense unknownno install recipeRun detail →
12
7.44µs90% SOL
2.17×
Reported · license unknown · no source
2026-08-17

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

no sourcelicense unknownno install recipeRun detail →
13
7.44µs90% SOL
2.17×
Reported · license unknown · no source
2026-08-18

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

no sourcelicense unknownno install recipeRun detail →
14
7.47µs90% SOL
2.17×
Reported · license unknown · no source
2026-08-19

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

no sourcelicense unknownno install recipeRun detail →
15
7.70µs89% SOL
2.24×
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 →
16
8.45µs88% SOL
2.46×
Reported · license unknown · no source
2026-07-13

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

no sourcelicense unknownno install recipeRun detail →
17
8.49µs88% SOL
2.47×
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 →
18
8.67µs88% SOL
2.52×
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 →
19
8.97µs88% SOL
2.61×
Reported · license unknown · no source
2026-07-13

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

no sourcelicense unknownno install recipeRun detail →
20
9.09µs88% SOL
2.64×
Reported · license unknown · no source
2026-07-13

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

no sourcelicense unknownno install recipeRun detail →
21
9.27µs87% SOL
2.70×
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 →
22
11.2µs85% SOL
3.27×
Reported · license unknown · no source
2026-08-12

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

no sourcelicense unknownno install recipeRun detail →
23
11.5µs85% SOL
3.33×
Reported · license unknown · no source
2026-07-27

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

no sourcelicense unknownno install recipeRun detail →
24
12.6µs83% SOL
3.68×
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 →
25
13.1µs82% SOL
3.80×
Reported · license unknown · no source
2026-08-02

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

no sourcelicense unknownno install recipeRun detail →
26
15.1µs81% SOL
4.38×
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
3.44µs
1.00×
Reported
license unknown · no source
unknown
4.02µs
1.17×
Reported
license unknown · no source
unknown
5.07µs
1.48×
Reported
license unknown · no source
unknown
5.43µs
1.58×
Reported
license unknown · no source
unknown
5.55µs
1.62×
Reported
license unknown · no source
unknown
5.69µs
1.66×
Reported
license unknown · no source
unknown
6.62µs
1.93×
Reported
license unknown · no source
unknown
8.41µs
2.45×
Reported
license unknown · no source
Show all 40 implementations ›
unknown
8.45µs
2.46×
Reported
license unknown · no source
unknown
8.88µs
2.58×
Reported
license unknown · no source
unknown
8.49µs
2.47×
Reported
license unknown · no source
unknown
9.53µs
2.77×
Reported
license unknown · no source
unknown
13.1µs
3.80×
Reported
license unknown · no source
unknown
14.8µs
4.30×
Reported
license unknown · no source
unknown
5.59µs
1.63×
Reported
license unknown · no source
unknown
5.70µs
1.66×
Reported
license unknown · no source
unknown
6.11µs
1.78×
Reported
license unknown · no source
unknown
6.52µs
1.90×
Reported
license unknown · no source
unknown
5.46µs
1.59×
Reported
license unknown · no source
unknown
7.44µs
2.17×
Reported
license unknown · no source
unknown
7.44µs
2.17×
Reported
license unknown · no source
unknown
7.47µs
2.17×
Reported
license unknown · no source
unknown
7.70µs
2.24×
Reported
license unknown · no source
unknown
8.13µs
2.36×
Reported
license unknown · no source
unknown
8.67µs
2.52×
Reported
license unknown · no source
unknown
8.71µs
2.53×
Reported
license unknown · no source
unknown
8.97µs
2.61×
Reported
license unknown · no source
unknown
9.09µs
2.64×
Reported
license unknown · no source
unknown
9.27µs
2.70×
Reported
license unknown · no source
unknown
9.29µs
2.70×
Reported
license unknown · no source
unknown
9.94µs
2.89×
Reported
license unknown · no source
unknown
11.2µs
3.27×
Reported
license unknown · no source
unknown
11.5µs
3.33×
Reported
license unknown · no source
unknown
12.1µs
3.52×
Reported
license unknown · no source
unknown
12.6µs
3.68×
Reported
license unknown · no source
unknown
15.1µs
4.38×
Reported
license unknown · no source
unknown
52.6µs
15.30×
Reported
license unknown · no source
unknown
3.92µs
1.14×
Reported
license unknown · no source
unknown
5.05µs
1.47×
Reported
license unknown · no source
unknown
5.01µs
1.46×
Reported
license unknown · no source

Semantics

Inputs and outputs
grad_cosfp32 [total_patches, head_dim]
grad_sinfp32 [total_patches, head_dim]
pos_idsint64 [total_patches, 2]
inv_freqfp32 [head_dim_quarter]
embfp32 [total_patches, head_dim]
grad_inv_freqfp32 [head_dim_quarter]
Axes and behavior
head_dimconstant = 72
num_imagesvariable
head_dim_halfconstant = 36
max_grid_sizevariable
total_patchesvariable
head_dim_quarterconstant = 18
determinismunspecified
constraintsNo mutation or aliasing
Identity
alias024_vision_rotary_position_embedding_generation_backwardmodelqwen-image-edit-2509sha256f0bf4b92837a…
Sources: NVIDIA SOL-ExecBench (2026-08-22)last observed 2026-08-22How records are decidedJSON