Skip to content
KernelIndex
Search⌘K
Sign in

SiLU activation backward

26 eligible runs
mlp

Backward pass for SiLU (Swish) activation function. Computes gradient with respect to input: grad_input = grad_output * sigmoid(x) * [1 + x * (1 - sigmoid(x))]

Fastest reported
4.94µsmean · 56% of SOL
RIAC as wellLicense unknown · unknown

Reported evidence · last observed 2026-08-20. Reported by source; not independently reproduced. No public source. License unknown.

No public sourceRun detail →

Current records

Workloadsuite of 16 cases · mean latency16 cases
num_elements
dtype
131fp322053fp323089fp324096fp3216384fp3240960fp32163840fp32262144fp32655360fp32786432fp322097152fp324194304fp325242880fp328388608fp3210485760fp3216777216fp32
scrolls · 16 cases total
Aggregatessuite of 16 cases · mean latency
HardwareBest knownImplementationRuns
NVIDIA B200 · env 14.94 µsRIAC as well17NVIDIA B200 · env 25.40 µsDatabricks9

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 · 17 results · last observed 2026-08-24Record history →
#
Implementation
Latency
vs #1
Trust
Observed
1
RIAC as well
4.94µs56% SOL
1.00×
Reported · license unknown · no source
2026-08-20

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

no sourcelicense unknownno install recipeRun detail →
2
Geometric
5.46µs53% SOL
1.11×
Reported · license unknown · no source
2026-08-24

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

no sourcelicense unknownno install recipeRun detail →
3
doubleAI
5.49µs53% SOL
1.11×
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 →
4
Recursive
5.53µs52% SOL
1.12×
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 →
5
TH3CHARLie | TileGym
5.62µs52% SOL
1.14×
Reported · license unknown · no source
2026-07-28

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

no sourcelicense unknownno install recipeRun detail →
6
Databricks
5.65µs52% SOL
1.15×
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
TH3CHARLie | TileGym
5.70µs52% SOL
1.15×
Reported · license unknown · no source
2026-07-28

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

no sourcelicense unknownno install recipeRun detail →
8
Amir M. Mir | SF Tensor
5.80µs51% SOL
1.17×
Reported · license unknown · no source
2026-08-07

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

no sourcelicense unknownno install recipeRun detail →
9
Amir M. Mir | SF Tensor
5.87µs51% SOL
1.19×
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 →
10
Abhinav Gorrepati
6.09µs50% SOL
1.23×
Reported · license unknown · no source
2026-07-17

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

no sourcelicense unknownno install recipeRun detail →
11
Hyra
6.09µs50% SOL
1.23×
Reported · license unknown · no source
2026-07-16

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

no sourcelicense unknownno install recipeRun detail →
12
Geometric
6.12µs49% SOL
1.24×
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 →
13
ac4k
6.33µs48% SOL
1.28×
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 →
14
Runic Kingfisher
6.34µs48% SOL
1.28×
Reported · license unknown · no source
2026-08-01

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

no sourcelicense unknownno install recipeRun detail →
15
Abhinav Gorrepati
6.39µs48% SOL
1.29×
Reported · license unknown · no source
2026-07-17

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

no sourcelicense unknownno install recipeRun detail →
16
Cursor
6.73µs47% SOL
1.36×
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 →
17
Runic Kingfisher
7.82µs43% SOL
1.58×
Reported · license unknown · no source
2026-08-01

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

no sourcelicense unknownno install recipeRun detail →

Implementations

Implementation
Runtime
Best latency
Evidence
Availability
RIAC as well
unknown
4.94µs
1.00×
Reported
license unknown · no source
Geometric
unknown
5.46µs
1.11×
Reported
license unknown · no source
TH3CHARLie | TileGym
unknown
5.62µs
1.14×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
5.80µs
1.17×
Reported
license unknown · no source
Hyra
unknown
6.09µs
1.23×
Reported
license unknown · no source
Abhinav Gorrepati
unknown
6.09µs
1.23×
Reported
license unknown · no source
ac4k
unknown
6.14µs
1.24×
Reported
license unknown · no source
Argus
unknown
6.18µs
1.25×
Reported
license unknown · no source
Show all 26 implementations ›
gnehiz
unknown
6.31µs
1.28×
Reported
license unknown · no source
Runic Kingfisher
unknown
6.34µs
1.28×
Reported
license unknown · no source
Cursor
unknown
6.71µs
1.36×
Reported
license unknown · no source
Theoric.com
unknown
7.76µs
1.57×
Reported
license unknown · no source
Databricks
unknown
5.65µs
1.15×
Reported
license unknown · no source
Recursive
unknown
5.66µs
1.15×
Reported
license unknown · no source
TH3CHARLie | TileGym
unknown
5.70µs
1.15×
Reported
license unknown · no source
doubleAI
unknown
5.74µs
1.16×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
5.87µs
1.19×
Reported
license unknown · no source
Geometric
unknown
6.12µs
1.24×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
6.16µs
1.25×
Reported
license unknown · no source
ac4k
unknown
6.33µs
1.28×
Reported
license unknown · no source
Abhinav Gorrepati
unknown
6.39µs
1.29×
Reported
license unknown · no source
Cursor
unknown
6.73µs
1.36×
Reported
license unknown · no source
Runic Kingfisher
unknown
7.82µs
1.58×
Reported
license unknown · no source
Databricks
unknown
5.40µs
1.09×
Reported
license unknown · no source
doubleAI
unknown
5.49µs
1.11×
Reported
license unknown · no source
Recursive
unknown
5.53µs
1.12×
Reported
license unknown · no source

Semantics

Inputs and outputs
grad_outputfp32 [num_elements]
xfp32 [num_elements]
sigmoid_xfp32 [num_elements]
grad_inputfp32 [num_elements]
Axes and behavior
num_elementsvariable
determinismunspecified
constraintsNo mutation or aliasing
Identity
alias084_silu_activation_backwardmodelstable-diffusion-xl-base-1-0sha256046d16d38d2f…
Sources: NVIDIA SOL-ExecBench (2026-08-24)last observed 2026-08-24How records are decidedJSON