Skip to content
KernelIndex
Search⌘K

Flux feedforward GELU approximate

31 eligible runs
mlp

FLUX feedforward network with GELU approximate activation (tanh approximation). Performs Linear projection to 4x hidden dimension, GELU-tanh activation, and Linear projection back to hidden dimension. This appears 76 times per forward pass in FLUX transformer blocks.

Fastest reported
518.7µsmean · 72% of SOL
RecursiveLicense unknown · unknown

Reported evidence · last observed 2026-07-10. Reported by source; not independently reproduced. No public source. License unknown.

No public sourceRun detail →

Current records

Workloadsuite of 16 cases · mean latencyhidden_dim = 3072 · mlp_hidden_dim = 1228816 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 · 17 results · last observed 2026-08-19Record history →
#
Implementation
Latency
vs #1
Trust
Observed
1
518.7µs72% SOL
1.00×
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 →
2
524.4µs71% SOL
1.01×
Reported · license unknown · no source
2026-07-21

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

no sourcelicense unknownno install recipeRun detail →
3
526.9µs71% SOL
1.02×
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 →
4
529.9µs70% SOL
1.02×
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 →
5
532.7µs70% SOL
1.03×
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
537.1µs70% SOL
1.04×
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
553.1µs68% SOL
1.07×
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 →
8
563.1µs68% SOL
1.09×
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 →
9
570.5µs67% 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
575.0µs67% 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 →
11
601.6µs64% SOL
1.16×
Reported · license unknown · no source
2026-08-10

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

no sourcelicense unknownno install recipeRun detail →
12
611.4µs63% SOL
1.18×
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
706.9µs57% SOL
1.36×
Reported · license unknown · no source
2026-07-23

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

no sourcelicense unknownno install recipeRun detail →
14
766.5µs54% SOL
1.48×
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 →
15
815.5µs52% SOL
1.57×
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 →
16
866.5µs48% SOL
1.67×
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
868.0µs51% SOL
1.67×
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
524.4µs
1.11×
Reported
license unknown · no source
unknown
472.8µs
1.00×
Reported
license unknown · no source
unknown
516.0µs
1.09×
Reported
license unknown · no source
unknown
526.9µs
1.11×
Reported
license unknown · no source
unknown
529.9µs
1.12×
Reported
license unknown · no source
unknown
532.7µs
1.13×
Reported
license unknown · no source
unknown
545.9µs
1.15×
Reported
license unknown · no source
unknown
541.9µs
1.15×
Reported
license unknown · no source
Show all 31 implementations ›
unknown
601.6µs
1.27×
Reported
license unknown · no source
unknown
766.5µs
1.62×
Reported
license unknown · no source
unknown
830.0µs
1.76×
Reported
license unknown · no source
unknown
844.8µs
1.79×
Reported
license unknown · no source
unknown
11.5ms
24.32×
Reported
license unknown · no source
unknown
11.7ms
24.70×
Reported
license unknown · no source
unknown
537.1µs
1.14×
Reported
license unknown · no source
unknown
551.4µs
1.17×
Reported
license unknown · no source
unknown
553.1µs
1.17×
Reported
license unknown · no source
unknown
563.1µs
1.19×
Reported
license unknown · no source
unknown
570.5µs
1.21×
Reported
license unknown · no source
unknown
575.0µs
1.22×
Reported
license unknown · no source
unknown
591.5µs
1.25×
Reported
license unknown · no source
unknown
611.4µs
1.29×
Reported
license unknown · no source
unknown
706.9µs
1.50×
Reported
license unknown · no source
unknown
815.5µs
1.72×
Reported
license unknown · no source
unknown
868.0µs
1.84×
Reported
license unknown · no source
unknown
849.2µs
1.80×
Reported
license unknown · no source
unknown
866.5µs
1.83×
Reported
license unknown · no source
unknown
9.94ms
21.01×
Reported
license unknown · no source
unknown
11.7ms
24.67×
Reported
license unknown · no source
unknown
506.8µs
1.07×
Reported
license unknown · no source
unknown
518.7µs
1.10×
Reported
license unknown · no source

Semantics

Inputs and outputs
hidden_statesfp32 [batch_size, seq_len, hidden_dim]
fc1_weightfp32 [mlp_hidden_dim, hidden_dim]
fc1_biasfp32 [mlp_hidden_dim]
fc2_weightfp32 [hidden_dim, mlp_hidden_dim]
fc2_biasfp32 [hidden_dim]
outputfp32 [batch_size, seq_len, hidden_dim]
Axes and behavior
seq_lenvariable
batch_sizevariable
hidden_dimconstant = 3072
mlp_hidden_dimconstant = 12288
determinismunspecified
constraintsNo mutation or aliasing
Identity
alias037_flux_feedforward_gelu_approximatemodelflux-1-devsha2560a98672cef6f…
Sources: NVIDIA SOL-ExecBench (2026-08-19)last observed 2026-08-19How records are decidedJSON