Skip to content
KernelIndex
Search⌘K
Sign in

FP8 MLP gate up projection

50 eligible runs
mlp

FP8-quantized fused gate and up projection for standard MLP layers. Combines gate_proj and up_proj linear projections with SiLU activation: output = silu(gate_proj(x)) * up_proj(x). Uses BlockWise1x128 scaling for input activations and BlockWise128x128 scaling for weights.

Fastest reported
288.8µsmean · 76% 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 latencyhidden_size = 3584 · hidden_blocks = 28 · intermediate_size = 18944 · intermediate_blocks = 14816 cases
m
dtype
384fp32/fp8_e4m3512fp32/fp8_e4m3768fp32/fp8_e4m31152fp32/fp8_e4m31280fp32/fp8_e4m31792fp32/fp8_e4m32048fp32/fp8_e4m32304fp32/fp8_e4m32816fp32/fp8_e4m33072fp32/fp8_e4m33584fp32/fp8_e4m33840fp32/fp8_e4m35120fp32/fp8_e4m35632fp32/fp8_e4m36144fp32/fp8_e4m38704fp32/fp8_e4m3
scrolls · 16 cases total
Aggregatessuite of 16 cases · mean latency
HardwareBest knownImplementationRuns
NVIDIA B200 · env 1288.8 µsRIAC as well44NVIDIA B200 · env 2379.2 µsdoubleAI6

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 · 44 results · last observed 2026-08-23Record history →
#
Implementation
Latency
vs #1
Trust
Observed
1
RIAC as well
288.8µs76% 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
Amir M. Mir | SF Tensor
333.4µs69% SOL
1.15×
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 →
3
ac4k
349.3µs67% SOL
1.21×
Reported · license unknown · no source
2026-07-31

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

no sourcelicense unknownno install recipeRun detail →
4
Amir M. Mir | SF Tensor
378.0µs64% SOL
1.31×
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 →
5
doubleAI
407.0µs61% SOL
1.41×
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 →
6
Geometric
413.8µs60% SOL
1.43×
Reported · license unknown · no source
2026-08-11

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

no sourcelicense unknownno install recipeRun detail →
7
Geometric
416.1µs60% SOL
1.44×
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 →
8
Geometric
416.7µs60% SOL
1.44×
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 →
9
Geometric
417.7µs60% SOL
1.45×
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 →
10
Geometric
418.6µs60% SOL
1.45×
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 →
11
Geometric
425.2µs59% SOL
1.47×
Reported · license unknown · no source
2026-08-14

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

no sourcelicense unknownno install recipeRun detail →
12
Amir M. Mir | SF Tensor
427.7µs59% SOL
1.48×
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
Samuel Mausberg
430.1µs59% SOL
1.49×
Reported · license unknown · no source
2026-08-14

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

no sourcelicense unknownno install recipeRun detail →
14
Geometric
433.0µs58% SOL
1.50×
Reported · license unknown · no source
2026-08-09

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

no sourcelicense unknownno install recipeRun detail →
15
Q H
433.5µs58% SOL
1.50×
Reported · license unknown · no source
2026-07-31

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

no sourcelicense unknownno install recipeRun detail →
16
Q H
434.6µs58% SOL
1.50×
Reported · license unknown · no source
2026-07-31

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

no sourcelicense unknownno install recipeRun detail →
17
Amir M. Mir | SF Tensor
435.5µs58% SOL
1.51×
Reported · license unknown · no source
2026-08-03

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

no sourcelicense unknownno install recipeRun detail →
18
Samuel Mausberg
438.5µs58% SOL
1.52×
Reported · license unknown · no source
2026-08-13

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

no sourcelicense unknownno install recipeRun detail →
19
Lucky Phoenix
440.2µs58% SOL
1.52×
Reported · license unknown · no source
2026-07-19

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

no sourcelicense unknownno install recipeRun detail →
20
Cursor
441.4µs58% SOL
1.53×
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
Samuel Mausberg
446.5µs57% SOL
1.55×
Reported · license unknown · no source
2026-08-13

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

no sourcelicense unknownno install recipeRun detail →
22
Amir M. Mir | SF Tensor
455.7µs57% SOL
1.58×
Reported · license unknown · no source
2026-07-19

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

no sourcelicense unknownno install recipeRun detail →
23
Hyra
474.4µs55% SOL
1.64×
Reported · license unknown · no source
2026-07-18

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

no sourcelicense unknownno install recipeRun detail →
24
Amir M. Mir | SF Tensor
483.2µs54% SOL
1.67×
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 →
25
Geometric
489.7µs54% SOL
1.70×
Reported · license unknown · no source
2026-08-09

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

no sourcelicense unknownno install recipeRun detail →
26
Geometric
502.7µs53% SOL
1.74×
Reported · license unknown · no source
2026-08-09

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

no sourcelicense unknownno install recipeRun detail →
27
Geometric
506.4µs52% SOL
1.75×
Reported · license unknown · no source
2026-08-09

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

no sourcelicense unknownno install recipeRun detail →
28
Geometric
511.8µs52% SOL
1.77×
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 →
29
Geometric
520.6µs51% SOL
1.80×
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 →
30
Geometric
539.8µs50% SOL
1.87×
Reported · license unknown · no source
2026-08-03

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

no sourcelicense unknownno install recipeRun detail →
31
Geometric
546.0µs50% SOL
1.89×
Reported · license unknown · no source
2026-08-05

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

no sourcelicense unknownno install recipeRun detail →
32
Geometric
560.4µs49% SOL
1.94×
Reported · license unknown · no source
2026-08-03

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

no sourcelicense unknownno install recipeRun detail →
33
Geometric
561.0µs49% SOL
1.94×
Reported · license unknown · no source
2026-08-03

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

no sourcelicense unknownno install recipeRun detail →
34
Lucky Phoenix
572.7µs48% SOL
1.98×
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 →
35
Samuel Mausberg
700.8µs41% SOL
2.43×
Reported · license unknown · no source
2026-08-13

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

no sourcelicense unknownno install recipeRun detail →
36
Recursive
820.3µs37% SOL
2.84×
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 →
37
Amir M. Mir | SF Tensor
860.0µs35% SOL
2.98×
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 →
38
Geometric
992.9µs32% SOL
3.44×
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 →
39
Geometric
1.05ms30% SOL
3.65×
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 →
40
Geometric
1.33ms25% SOL
4.60×
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 →
41
IcySky
1.37ms24% SOL
4.76×
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 →
42
Geometric
1.52ms22% SOL
5.27×
Reported · license unknown · no source
2026-08-09

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

no sourcelicense unknownno install recipeRun detail →
43
Geometric
1.61ms21% SOL
5.56×
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 →
44
Geometric
2.29ms15% SOL
7.94×
Reported · license unknown · no source
2026-08-09

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

no sourcelicense unknownno install recipeRun detail →

Implementations

Implementation
Runtime
Best latency
Evidence
Availability
Geometric
unknown
413.8µs
1.43×
Reported
license unknown · no source
Geometric
unknown
425.2µs
1.47×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
427.7µs
1.48×
Reported
license unknown · no source
doubleAI
unknown
407.0µs
1.41×
Reported
license unknown · no source
RIAC as well
unknown
288.8µs
1.00×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
333.4µs
1.15×
Reported
license unknown · no source
ac4k
unknown
349.3µs
1.21×
Reported
license unknown · no source
doubleAI
unknown
379.2µs
1.31×
Reported
license unknown · no source
Show all 50 implementations ›
Samuel Mausberg
unknown
430.1µs
1.49×
Reported
license unknown · no source
Q H
unknown
433.5µs
1.50×
Reported
license unknown · no source
Cursor
unknown
441.4µs
1.53×
Reported
license unknown · no source
Lucky Phoenix
unknown
440.2µs
1.52×
Reported
license unknown · no source
Hyra
unknown
474.4µs
1.64×
Reported
license unknown · no source
Recursive
unknown
551.9µs
1.91×
Reported
license unknown · no source
IcySky
unknown
979.3µs
3.39×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
378.0µs
1.31×
Reported
license unknown · no source
Geometric
unknown
416.1µs
1.44×
Reported
license unknown · no source
Geometric
unknown
416.7µs
1.44×
Reported
license unknown · no source
Geometric
unknown
417.7µs
1.45×
Reported
license unknown · no source
Geometric
unknown
418.6µs
1.45×
Reported
license unknown · no source
Q H
unknown
434.6µs
1.50×
Reported
license unknown · no source
Geometric
unknown
433.0µs
1.50×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
435.5µs
1.51×
Reported
license unknown · no source
Samuel Mausberg
unknown
438.5µs
1.52×
Reported
license unknown · no source
Cursor
unknown
443.3µs
1.53×
Reported
license unknown · no source
Samuel Mausberg
unknown
446.5µs
1.55×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
455.7µs
1.58×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
483.2µs
1.67×
Reported
license unknown · no source
Geometric
unknown
489.7µs
1.70×
Reported
license unknown · no source
Geometric
unknown
502.7µs
1.74×
Reported
license unknown · no source
Geometric
unknown
506.4µs
1.75×
Reported
license unknown · no source
Geometric
unknown
511.8µs
1.77×
Reported
license unknown · no source
Geometric
unknown
520.6µs
1.80×
Reported
license unknown · no source
Geometric
unknown
539.8µs
1.87×
Reported
license unknown · no source
Geometric
unknown
546.0µs
1.89×
Reported
license unknown · no source
Geometric
unknown
561.0µs
1.94×
Reported
license unknown · no source
Geometric
unknown
560.4µs
1.94×
Reported
license unknown · no source
Lucky Phoenix
unknown
572.7µs
1.98×
Reported
license unknown · no source
Samuel Mausberg
unknown
700.8µs
2.43×
Reported
license unknown · no source
Recursive
unknown
820.3µs
2.84×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
808.5µs
2.80×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
860.0µs
2.98×
Reported
license unknown · no source
Geometric
unknown
992.9µs
3.44×
Reported
license unknown · no source
Geometric
unknown
1.05ms
3.65×
Reported
license unknown · no source
Geometric
unknown
1.33ms
4.60×
Reported
license unknown · no source
IcySky
unknown
1.37ms
4.76×
Reported
license unknown · no source
Geometric
unknown
1.52ms
5.27×
Reported
license unknown · no source
Geometric
unknown
1.61ms
5.56×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
1.90ms
6.59×
Reported
license unknown · no source
Geometric
unknown
2.29ms
7.94×
Reported
license unknown · no source

Semantics

Inputs and outputs
xfp8_e4m3 [m, hidden_size]
scale_xfp32 [m, hidden_blocks]
gate_proj_weightfp8_e4m3 [intermediate_size, hidden_size]
scale_gatefp32 [intermediate_blocks, hidden_blocks]
up_proj_weightfp8_e4m3 [intermediate_size, hidden_size]
scale_upfp32 [intermediate_blocks, hidden_blocks]
outputbf16 [m, intermediate_size]
Axes and behavior
mvariable
hidden_sizeconstant = 3584
hidden_blockshidden_blocks = hidden_size // 128
intermediate_sizeconstant = 18944
intermediate_blocksintermediate_blocks = intermediate_size // 128
determinismunspecified
constraintsNo mutation or aliasing
Identity
alias003_fp8_mlp_gate_up_projectionmodelqwen3-vl-30b-a3b-instruct-fp8sha25659867c7a75ea…
Sources: NVIDIA SOL-ExecBench (2026-08-23)last observed 2026-08-23How records are decidedJSON