Skip to content
KernelIndex
Search⌘K
Sign in

FP8 Mamba2 out projection

51 eligible runs
quantization

FP8-quantized output projection in Mamba2 mixer that projects from intermediate size (2688) back to hidden dimension (2688). Uses BlockWise1x128 scaling for activations and BlockWise128x128 scaling for weights.

Fastest reported
27.2µsmean · 81% of SOL
RIAC as wellLicense unknown · unknown

Reported evidence · last observed 2026-08-22. Reported by source; not independently reproduced. No public source. License unknown.

No public sourceRun detail →

Current records

Workloadsuite of 16 cases · mean latencyk = 2688 · n = 2688 · block = 128 · k_blocks = 21 · n_blocks = 2116 cases
m
dtype
128fp32/fp8_e4m3256fp32/fp8_e4m3512fp32/fp8_e4m3640fp32/fp8_e4m31024fp32/fp8_e4m31152fp32/fp8_e4m31792fp32/fp8_e4m32560fp32/fp8_e4m32816fp32/fp8_e4m34096fp32/fp8_e4m34352fp32/fp8_e4m34608fp32/fp8_e4m35632fp32/fp8_e4m36656fp32/fp8_e4m37168fp32/fp8_e4m37680fp32/fp8_e4m3
scrolls · 16 cases total
Aggregatessuite of 16 cases · mean latency
HardwareBest knownImplementationRuns
NVIDIA B200 · env 127.2 µsRIAC as well44NVIDIA B200 · env 227.5 µsdoubleAI7

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 2026-08-24Record history →
#
Implementation
Latency
vs #1
Trust
Observed
1
RIAC as well
27.2µs81% SOL
1.00×
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 →
2
doubleAI
27.6µs81% 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 →
3
Hyra
27.9µs80% SOL
1.02×
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 →
4
ac4k
31.5µs77% SOL
1.16×
Reported · license unknown · no source
2026-08-02

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

no sourcelicense unknownno install recipeRun detail →
5
Amir M. Mir | SF Tensor
31.6µs77% SOL
1.16×
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 →
6
Amir M. Mir | SF Tensor
33.2µs76% SOL
1.22×
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 →
7
Amir M. Mir | SF Tensor
34.7µs75% SOL
1.27×
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 →
8
Amir M. Mir | SF Tensor
35.0µs74% 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 →
9
Geometric
35.7µs73% SOL
1.31×
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
Geometric
35.8µs73% SOL
1.32×
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 →
11
Geometric
36.2µs73% SOL
1.33×
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 →
12
Geometric
37.3µs72% SOL
1.37×
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 →
13
Geometric
37.3µs72% SOL
1.37×
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 →
14
Geometric
37.5µs72% SOL
1.38×
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 →
15
Geometric
37.6µs72% SOL
1.38×
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 →
16
Geometric
37.6µs72% SOL
1.38×
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 →
17
Geometric
37.7µs72% SOL
1.39×
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 →
18
Geometric
37.9µs72% SOL
1.39×
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 →
19
Geometric
37.9µs72% SOL
1.39×
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 →
20
Geometric
38.0µs72% SOL
1.40×
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 →
21
Geometric
38.1µs72% SOL
1.40×
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 →
22
Geometric
38.1µs72% SOL
1.40×
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 →
23
Geometric
38.1µs72% SOL
1.40×
Reported · license unknown · no source
2026-08-17

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

no sourcelicense unknownno install recipeRun detail →
24
Geometric
38.2µs72% SOL
1.40×
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 →
25
Geometric
38.4µs71% SOL
1.41×
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 →
26
Geometric
38.5µs71% SOL
1.42×
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 →
27
Geometric
38.6µs71% SOL
1.42×
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
Cursor
39.1µs71% SOL
1.44×
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 →
29
Geometric
39.2µs71% SOL
1.44×
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 →
30
Geometric
39.4µs71% SOL
1.45×
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 →
31
Geometric
39.6µs71% SOL
1.45×
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 →
32
Geometric
39.7µs71% SOL
1.46×
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 →
33
Recursive
41.0µs70% SOL
1.51×
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 →
34
Geometric
56.6µs62% SOL
2.08×
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 →
35
Geometric
64.6µs58% SOL
2.38×
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 →
36
Geometric
70.0µs56% SOL
2.57×
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 →
37
Geometric
70.2µs56% SOL
2.58×
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 →
38
Geometric
93.0µs48% SOL
3.42×
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 →
39
Amir M. Mir | SF Tensor
98.4µs47% SOL
3.62×
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 →
40
Geometric
117.8µs42% SOL
4.33×
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 →
41
Geometric
123.4µs41% SOL
4.54×
Reported · license unknown · no source
2026-07-26

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

no sourcelicense unknownno install recipeRun detail →
42
Geometric
124.0µs40% SOL
4.56×
Reported · license unknown · no source
2026-07-26

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

no sourcelicense unknownno install recipeRun detail →
43
Geometric
124.1µs40% SOL
4.56×
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 →
44
I am CCC
929.4µs8% SOL
34.18×
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 →

Implementations

Implementation
Runtime
Best latency
Evidence
Availability
Hyra
unknown
27.9µs
1.02×
Reported
license unknown · no source
ac4k
unknown
31.5µs
1.16×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
33.2µs
1.22×
Reported
license unknown · no source
doubleAI
unknown
27.5µs
1.01×
Reported
license unknown · no source
RIAC as well
unknown
27.2µs
1.00×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
31.6µs
1.16×
Reported
license unknown · no source
Geometric
unknown
35.7µs
1.31×
Reported
license unknown · no source
Cursor
unknown
39.0µs
1.43×
Reported
license unknown · no source
Show all 51 implementations ›
Recursive
unknown
40.1µs
1.47×
Reported
license unknown · no source
I am CCC
unknown
57.6µs
2.12×
Reported
license unknown · no source
doubleAI
unknown
27.6µs
1.02×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
34.7µs
1.27×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
35.0µs
1.29×
Reported
license unknown · no source
Geometric
unknown
35.8µs
1.32×
Reported
license unknown · no source
Geometric
unknown
36.2µs
1.33×
Reported
license unknown · no source
Geometric
unknown
37.3µs
1.37×
Reported
license unknown · no source
Geometric
unknown
37.3µs
1.37×
Reported
license unknown · no source
Geometric
unknown
37.5µs
1.38×
Reported
license unknown · no source
Geometric
unknown
37.6µs
1.38×
Reported
license unknown · no source
Geometric
unknown
37.6µs
1.38×
Reported
license unknown · no source
Geometric
unknown
37.7µs
1.39×
Reported
license unknown · no source
Geometric
unknown
37.9µs
1.39×
Reported
license unknown · no source
Geometric
unknown
38.0µs
1.40×
Reported
license unknown · no source
Geometric
unknown
37.9µs
1.39×
Reported
license unknown · no source
Geometric
unknown
38.1µs
1.40×
Reported
license unknown · no source
Geometric
unknown
38.1µs
1.40×
Reported
license unknown · no source
Geometric
unknown
38.1µs
1.40×
Reported
license unknown · no source
Geometric
unknown
38.2µs
1.40×
Reported
license unknown · no source
Geometric
unknown
38.4µs
1.41×
Reported
license unknown · no source
Geometric
unknown
39.2µs
1.44×
Reported
license unknown · no source
Geometric
unknown
38.5µs
1.42×
Reported
license unknown · no source
Cursor
unknown
39.1µs
1.44×
Reported
license unknown · no source
Geometric
unknown
38.6µs
1.42×
Reported
license unknown · no source
Geometric
unknown
39.6µs
1.45×
Reported
license unknown · no source
Geometric
unknown
39.4µs
1.45×
Reported
license unknown · no source
Geometric
unknown
39.7µs
1.46×
Reported
license unknown · no source
Recursive
unknown
41.0µs
1.51×
Reported
license unknown · no source
Geometric
unknown
56.6µs
2.08×
Reported
license unknown · no source
Geometric
unknown
64.6µs
2.38×
Reported
license unknown · no source
I am CCC
unknown
64.5µs
2.37×
Reported
license unknown · no source
Geometric
unknown
70.2µs
2.58×
Reported
license unknown · no source
Geometric
unknown
70.0µs
2.57×
Reported
license unknown · no source
Geometric
unknown
93.0µs
3.42×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
96.0µs
3.53×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
98.4µs
3.62×
Reported
license unknown · no source
Geometric
unknown
117.8µs
4.33×
Reported
license unknown · no source
Geometric
unknown
123.4µs
4.54×
Reported
license unknown · no source
Geometric
unknown
124.0µs
4.56×
Reported
license unknown · no source
Geometric
unknown
124.1µs
4.56×
Reported
license unknown · no source
Amir M. Mir | SF Tensor
unknown
848.1µs
31.19×
Reported
license unknown · no source
I am CCC
unknown
929.4µs
34.18×
Reported
license unknown · no source

Semantics

Inputs and outputs
xfp8_e4m3 [m, k]
wfp8_e4m3 [n, k]
scale_xfp32 [m, k_blocks]
scale_wfp32 [n_blocks, k_blocks]
outputbf16 [m, n]
Axes and behavior
kconstant = 2688
mvariable
nconstant = 2688
blockconstant = 128
k_blocksk_blocks = k // block
n_blocksn_blocks = n // block
determinismunspecified
constraintsNo mutation or aliasing
Identity
alias022_fp8_mamba2_out_projectionmodelnvidia-nemotron-3-nano-30b-a3b-fp8sha256eca32c08238a…
Sources: NVIDIA SOL-ExecBench (2026-08-24)last observed 2026-08-24How records are decidedJSON