Skip to content
KernelIndex
Search⌘K

GeGLU activation

27 eligible runs
mlp

GEGLU (Gated GLU with GELU) activation function. Splits input along last dimension into two halves, applies approximate GELU (tanh-based) to the first half, and multiplies element-wise with the second half. Used in transformer feed-forward networks.

Fastest reported
17.0µsmean · 68% of SOL
GeometricLicense unknown · unknown

Reported evidence · last observed 2026-08-12. Reported by source; not independently reproduced. No public source. License unknown.

No public sourceRun detail →

Current records

Workloadsuite of 16 cases · mean latencyinner_dim = 5120 · input_dim = 1024016 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 · 16 results · last observed 2026-08-24Record history →
#
Implementation
Latency
vs #1
Trust
Observed
1
17.0µs68% SOL
1.00×
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 →
2
17.0µs68% 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 →
3
17.1µs68% SOL
1.01×
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 →
4
17.2µs67% SOL
1.01×
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
17.2µs67% SOL
1.01×
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 →
6
17.9µs66% SOL
1.05×
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 →
7
18.3µs65% SOL
1.08×
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
18.6µs65% SOL
1.10×
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 →
9
18.7µs65% SOL
1.10×
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
18.9µs64% 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 →
11
19.1µs64% SOL
1.13×
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
19.3µs64% SOL
1.14×
Reported · license unknown · no source
2026-07-29

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

no sourcelicense unknownno install recipeRun detail →
13
19.3µs63% SOL
1.14×
Reported · license unknown · no source
2026-07-14

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

no sourcelicense unknownno install recipeRun detail →
14
19.3µs63% 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 →
15
19.6µs63% SOL
1.16×
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 →
16
19.8µs63% SOL
1.16×
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
17.0µs
1.00×
Reported
license unknown · no source
unknown
17.1µs
1.01×
Reported
license unknown · no source
unknown
17.2µs
1.01×
Reported
license unknown · no source
unknown
18.2µs
1.07×
Reported
license unknown · no source
unknown
18.5µs
1.09×
Reported
license unknown · no source
unknown
18.6µs
1.10×
Reported
license unknown · no source
unknown
18.7µs
1.10×
Reported
license unknown · no source
unknown
19.1µs
1.13×
Reported
license unknown · no source
Show all 27 implementations ›
unknown
19.3µs
1.14×
Reported
license unknown · no source
unknown
19.3µs
1.14×
Reported
license unknown · no source
unknown
19.3µs
1.14×
Reported
license unknown · no source
unknown
19.6µs
1.16×
Reported
license unknown · no source
unknown
20.1µs
1.18×
Reported
license unknown · no source
unknown
35.0µs
2.06×
Reported
license unknown · no source
unknown
17.2µs
1.01×
Reported
license unknown · no source
unknown
17.9µs
1.05×
Reported
license unknown · no source
unknown
18.3µs
1.08×
Reported
license unknown · no source
unknown
18.8µs
1.10×
Reported
license unknown · no source
unknown
18.9µs
1.12×
Reported
license unknown · no source
unknown
18.8µs
1.11×
Reported
license unknown · no source
unknown
19.8µs
1.16×
Reported
license unknown · no source
unknown
19.9µs
1.17×
Reported
license unknown · no source
unknown
19.8µs
1.16×
Reported
license unknown · no source
unknown
84.9µs
5.00×
Reported
license unknown · no source
unknown
17.0µs
1.00×
Reported
license unknown · no source
unknown
17.2µs
1.01×
Reported
license unknown · no source
unknown
17.2µs
1.01×
Reported
license unknown · no source

Semantics

Inputs and outputs
xfp32 [batch_size, seq_len, input_dim]
outputfp32 [batch_size, seq_len, inner_dim]
Axes and behavior
seq_lenvariable
inner_dimconstant = 5120
input_diminput_dim = inner_dim * 2
batch_sizevariable
determinismunspecified
constraintsNo mutation or aliasing
Identity
alias085_geglu_activationmodelstable-diffusion-xl-refiner-1-0sha256569cbb7d5af1…
Sources: NVIDIA SOL-ExecBench (2026-08-24)last observed 2026-08-24How records are decidedJSON