Skip to content
KernelIndex
Search⌘K
Sign in

Residual coupling flow block

39 eligible runs
other

Residual coupling block for normalizing flow with affine coupling transformations. Splits channels in half, transforms one half conditioned on the other, and combines. Supports forward and reverse passes for flow-based generative modeling.

Fastest reported
179.6µsmean · 80% 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 latencychannels = 192 · n_layers = 4 · kernel_size = 5 · half_channels = 96 · hidden_channels = 19216 cases
time
batch_size
dtype
1281fp322562fp322932fp323844fp325414fp326912fp327688fp3210248fp3210874fp32153616fp3217218fp32204816fp32244716fp3226894fp32363764fp32819264fp32
scrolls · 16 cases total
Aggregatessuite of 16 cases · mean latency
HardwareBest knownImplementationRuns
NVIDIA B200 · env 1179.6 µsRIAC as well28NVIDIA B200 · env 2194.8 µsdoubleAI11

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 · 28 results · last observed 2026-08-23Record history →
#
Implementation
Latency
vs #1
Trust
Observed
1
RIAC as well
179.6µs80% 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
Hyra
194.5µs79% SOL
1.08×
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 →
3
doubleAI
201.9µs78% 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 →
4
ac4k
252.1µs74% SOL
1.40×
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 →
5
Amir M. Mir | SF Tensor
259.1µs73% SOL
1.44×
Reported · license unknown · no source
2026-07-22

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

no sourcelicense unknownno install recipeRun detail →
6
Recursive
291.2µs71% 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 →
7
Humble Pegasus
303.1µs70% SOL
1.69×
Reported · license unknown · no source
2026-07-30

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

no sourcelicense unknownno install recipeRun detail →
8
Geometric
318.5µs69% SOL
1.77×
Reported · license unknown · no source
2026-08-23

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

no sourcelicense unknownno install recipeRun detail →
9
Geometric
318.8µs69% SOL
1.78×
Reported · license unknown · no source
2026-08-23

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

no sourcelicense unknownno install recipeRun detail →
10
Amir M. Mir | SF Tensor
329.8µs69% SOL
1.84×
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 →
11
Geometric
333.1µs68% SOL
1.85×
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 →
12
Geometric
333.2µs68% SOL
1.86×
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 →
13
Geometric
333.2µs68% SOL
1.86×
Reported · license unknown · no source
2026-08-23

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

no sourcelicense unknownno install recipeRun detail →
14
Geometric
334.9µs68% SOL
1.87×
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 →
15
Geometric
337.4µs68% SOL
1.88×
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 →
16
Geometric
344.1µs67% SOL
1.92×
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 →
17
Geometric
374.5µs65% SOL
2.09×
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 →
18
Geometric
397.2µs64% SOL
2.21×
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 →
19
Geometric
403.2µs64% SOL
2.25×
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 →
20
Geometric
516.6µs57% SOL
2.88×
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 →
21
Geometric
536.7µs57% SOL
2.99×
Reported · license unknown · no source
2026-08-16

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

no sourcelicense unknownno install recipeRun detail →
22
Proud Wren @ NVIDIA
561.9µs55% SOL
3.13×
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 →
23
Proud Wren @ NVIDIA
614.1µs53% SOL
3.42×
Reported · license unknown · no source
2026-07-12

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

no sourcelicense unknownno install recipeRun detail →
24
Cursor
645.5µs52% SOL
3.60×
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
bismuth83
657.4µs51% SOL
3.66×
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 →
26
Geometric
691.9µs50% SOL
3.85×
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 →
27
Amir M. Mir | SF Tensor
867.4µs44% SOL
4.83×
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 →
28
Amir M. Mir | SF Tensor
957.1µs42% SOL
5.33×
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
RIAC as well
unknown
179.6µs
1.00×
Reported
license unknown · no source
ac4k
unknown
252.1µs
1.40×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
259.1µs
1.44×
Reported
license unknown · no source
Recursive
unknown
280.8µs
1.56×
Reported
license unknown · no source
Gilded Quail
unknown
281.2µs
1.57×
Reported
license unknown · no source
Humble Pegasus
unknown
303.1µs
1.69×
Reported
license unknown · no source
Geometric
unknown
318.5µs
1.77×
Reported
license unknown · no source
Proud Wren @ NVIDIA
unknown
561.9µs
3.13×
Reported
license unknown · no source
Show all 39 implementations ›
Cursor
unknown
617.9µs
3.44×
Reported
license unknown · no source
bismuth83
unknown
657.4µs
3.66×
Reported
license unknown · no source
Recursive
unknown
291.2µs
1.62×
Reported
license unknown · no source
Geometric
unknown
318.8µs
1.78×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
329.8µs
1.84×
Reported
license unknown · no source
Geometric
unknown
333.1µs
1.85×
Reported
license unknown · no source
Geometric
unknown
333.2µs
1.86×
Reported
license unknown · no source
Geometric
unknown
333.2µs
1.86×
Reported
license unknown · no source
Geometric
unknown
334.9µs
1.87×
Reported
license unknown · no source
Geometric
unknown
337.4µs
1.88×
Reported
license unknown · no source
Geometric
unknown
344.1µs
1.92×
Reported
license unknown · no source
Geometric
unknown
374.5µs
2.09×
Reported
license unknown · no source
Geometric
unknown
397.2µs
2.21×
Reported
license unknown · no source
Geometric
unknown
403.2µs
2.25×
Reported
license unknown · no source
Geometric
unknown
516.6µs
2.88×
Reported
license unknown · no source
Geometric
unknown
536.7µs
2.99×
Reported
license unknown · no source
doubleAI
unknown
608.1µs
3.39×
Reported
license unknown · no source
Proud Wren @ NVIDIA
unknown
614.1µs
3.42×
Reported
license unknown · no source
Cursor
unknown
645.5µs
3.60×
Reported
license unknown · no source
bismuth83
unknown
671.0µs
3.74×
Reported
license unknown · no source
bismuth83
unknown
671.6µs
3.74×
Reported
license unknown · no source
bismuth83
unknown
672.6µs
3.75×
Reported
license unknown · no source
Geometric
unknown
691.9µs
3.85×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
857.2µs
4.77×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
867.4µs
4.83×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
889.9µs
4.96×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
957.1µs
5.33×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
973.6µs
5.42×
Reported
license unknown · no source
doubleAI
unknown
194.8µs
1.08×
Reported
license unknown · no source
Hyra
unknown
194.5µs
1.08×
Reported
license unknown · no source
doubleAI
unknown
201.9µs
1.12×
Reported
license unknown · no source

Semantics

Inputs and outputs
xfp32 [batch_size, channels, time]
x_maskfp32 [batch_size, 1, time]
reversebool scalar
transform_0_conv0_weightfp32 [hidden_channels, half_channels, kernel_size]
transform_0_conv0_biasfp32 [hidden_channels]
transform_0_conv1_weightfp32 [hidden_channels, hidden_channels, kernel_size]
transform_0_conv1_biasfp32 [hidden_channels]
transform_0_conv2_weightfp32 [half_channels, hidden_channels, kernel_size]
transform_0_conv2_biasfp32 [half_channels]
transform_1_conv0_weightfp32 [hidden_channels, half_channels, kernel_size]
transform_1_conv0_biasfp32 [hidden_channels]
transform_1_conv1_weightfp32 [hidden_channels, hidden_channels, kernel_size]
transform_1_conv1_biasfp32 [hidden_channels]
transform_1_conv2_weightfp32 [half_channels, hidden_channels, kernel_size]
transform_1_conv2_biasfp32 [half_channels]
transform_2_conv0_weightfp32 [hidden_channels, half_channels, kernel_size]
transform_2_conv0_biasfp32 [hidden_channels]
transform_2_conv1_weightfp32 [hidden_channels, hidden_channels, kernel_size]
transform_2_conv1_biasfp32 [hidden_channels]
transform_2_conv2_weightfp32 [half_channels, hidden_channels, kernel_size]
transform_2_conv2_biasfp32 [half_channels]
transform_3_conv0_weightfp32 [hidden_channels, half_channels, kernel_size]
transform_3_conv0_biasfp32 [hidden_channels]
transform_3_conv1_weightfp32 [hidden_channels, hidden_channels, kernel_size]
transform_3_conv1_biasfp32 [hidden_channels]
transform_3_conv2_weightfp32 [half_channels, hidden_channels, kernel_size]
transform_3_conv2_biasfp32 [half_channels]
outputfp32 [batch_size, channels, time]
Axes and behavior
timevariable
channelsconstant = 192
n_layersconstant = 4
batch_sizevariable
kernel_sizeconstant = 5
half_channelsconstant = 96
hidden_channelsconstant = 192
determinismunspecified
constraintsNo mutation or aliasing
Identity
alias057_residual_coupling_flow_blockmodelopenvoicesha25642e012198576…
Sources: NVIDIA SOL-ExecBench (2026-08-23)last observed 2026-08-23How records are decidedJSON