Skip to content
KernelIndex
Search⌘K

FlashInfer-Bench baselines

Library · flashinfer-ai/flashinfer-trace · Apache-2.0
Records held
1,288as of 2026-10-01 23:33 UTC
Kernels
90
Runs
1,851
Is this you? Claim this project ›

A claim grants attribution and metadata maintenance, never the right to edit evidence. Evidence claims are reviewed by a maintainer.

Current records

Operation · workload
History
Record
Implementation
Hardware
Since
RMSNorm h7168bf16 · [7168] · batch 64
8.10µs
0.9% faster · was 8.17 µs
B200
2025-10-21
RMSNorm h7168bf16 · [7168] · batch 18
8.06µs
1.3% faster · was 8.17 µs
B200
2025-10-21
RMSNorm h7168bf16 · [7168] · batch 1
8.10µs
0.3% faster · was 8.13 µs
B200
2025-10-21
RMSNorm h7168bf16 · [7168] · batch 7
8.10µs
0.8% faster · was 8.16 µs
B200
2025-10-21
RMSNorm h4096bf16 · [4096] · batch 79
7.66µs
4.6% faster · was 8.02 µs
B200
2025-10-21
RMSNorm h4096bf16 · [4096] · batch 170
8.08µs
0.4% faster · was 8.12 µs
B200
2025-10-21
RMSNorm h1536bf16 · [1536] · batch 539
8.10µs
0.6% faster · was 8.15 µs
B200
2025-10-21
RMSNorm h1536bf16 · [1536] · batch 1
7.11µs
1.9% faster · was 7.25 µs
B200
2025-10-21
Fused add RMSNorm h7168bf16 · [7168] · batch 64
8.29µs
1.9% faster · was 8.45 µs
B200
2025-10-21
Fused add RMSNorm h7168bf16 · [7168] · batch 32
8.23µs
17.9% faster · was 10.0 µs
B200
2025-10-21
Show all 40 records ›
Fused add RMSNorm h7168bf16 · [7168] · batch 1
8.19µs
0.3% faster · was 8.21 µs
B200
2025-10-21
Fused add RMSNorm h7168bf16 · [7168] · batch 7
8.21µs
1.6% faster · was 8.35 µs
B200
2025-10-21
Fused add RMSNorm h5120bf16 · [5120] · batch 37
3.42µs
3.6% faster · was 3.55 µs
B200
2026-03-26
Fused add RMSNorm h5120bf16 · [5120] · batch 32
3.30µs
3.7% faster · was 3.42 µs
B200
2026-03-26
Fused add RMSNorm h5120bf16 · [5120] · batch 6
3.33µs
1.9% faster · was 3.39 µs
B200
2026-03-26
Fused add RMSNorm h5120bf16 · [5120] · batch 1174
12.7µs
B200
2026-03-24
Fused add RMSNorm h5120bf16 · [5120] · batch 1
3.58µs
B200
2026-03-24
10.2µs
0.9% faster · was 10.3 µs
B200
2026-04-06
51.5µs
0.6% faster · was 51.8 µs
B200
2026-04-06
73.8µs
1.5% faster · was 74.9 µs
B200
2026-04-06
34.8µs
1.8% faster · was 35.5 µs
B200
2026-04-06
53.3µs
0.0% faster · was 53.3 µs
B200
2026-04-06
3.75ms
95.2% faster · was 78.5 ms
B200
2025-10-21
16.5µs
69.3% faster · was 53.7 µs
B200
2025-10-21
28.5µs
81.8% faster · was 156.6 µs
B200
2025-10-21
1.83ms
94.9% faster · was 35.9 ms
B200
2025-10-21
2.54ms
94.9% faster · was 50.3 ms
B200
2025-10-21
237.5µs
94.3% faster · was 4.13 ms
B200
2025-10-21
18.4µs
79.1% faster · was 87.9 µs
B200
2025-10-21
20.8µs
87.6% faster · was 168.0 µs
B200
2025-10-21
26.8µs
73.5% faster · was 101.1 µs
B200
2025-10-21
81.3µs
92.9% faster · was 1.15 ms
B200
2025-10-21
18.4µs
79.8% faster · was 91.0 µs
B200
2025-10-21
26.6µs
68.5% faster · was 84.6 µs
B200
2025-10-21
16.5ms
95.2% faster · was 347.0 ms
B200
2025-10-21

1248 more records · Open in the records ledger →

Measured kernels

Implementation
Operation
Runtime
Best latency
Evidence
Availability
flashinfer / wrapper7f6051FlashInfer-Bench baselines
python
3.30µs
Reported
Apache-2.0 · source
flashinfer / wrapper57c111FlashInfer-Bench baselines
python
6.28µs
Reported
Apache-2.0 · source
flashinfer / wrapper846dc8FlashInfer-Bench baselines
python
6.44µs
Reported
Apache-2.0 · source
flashinfer / wrapper0af255FlashInfer-Bench baselines
python
6.86µs
Reported
Apache-2.0 · source
flashinfer / wrappera27dc7FlashInfer-Bench baselines
python
6.92µs
Reported
Apache-2.0 · source
flashinfer / wrapper2e27cdFlashInfer-Bench baselines
python
7.59µs
Reported
Apache-2.0 · source
flashinfer / wrapper74a870FlashInfer-Bench baselines
python
7.67µs
Reported
Apache-2.0 · source
flashinfer / wrapper5d67c6FlashInfer-Bench baselines
python
8.06µs
Reported
Apache-2.0 · source
flashinfer / wrapper0ff432FlashInfer-Bench baselines
python
8.07µs
Reported
Apache-2.0 · source
flashinfer / wrapper3f9411FlashInfer-Bench baselines
python
8.15µs
Reported
Apache-2.0 · source
Show all 40 kernels ›
flashinfer / wrapper5bddf1FlashInfer-Bench baselines
python
8.19µs
Reported
Apache-2.0 · source
flashinfer / wrapper78fd04FlashInfer-Bench baselines
python
8.19µs
Reproduction-ready
Apache-2.0 · source
flashinfer / wrappera9588fFlashInfer-Bench baselines
python
8.26µs
Reported
Apache-2.0 · source
flashinfer / wrapper44b034FlashInfer-Bench baselines
python
8.93µs
Reported
Apache-2.0 · source
torch_matmul_317103FlashInfer-Bench baselines
python
9.45µs
Reported
Apache-2.0 · source
flashinfer / wrapperacea60FlashInfer-Bench baselines
python
10.2µs
Reported
Apache-2.0 · source
flashinfer / wrapper68ff49FlashInfer-Bench baselines
python
10.2µs
Reported
Apache-2.0 · source
flashinfer / wrapperd90b98FlashInfer-Bench baselines
python
10.2µs
Reproduction-ready
Apache-2.0 · source
flashinfer / wrapperf9a07bFlashInfer-Bench baselines
python
10.2µs
Reported
Apache-2.0 · source
flashinfer / wrapper463e16FlashInfer-Bench baselines
python
10.2µs
Reproduction-ready
Apache-2.0 · source
flashinfer / wrapper33cb6fFlashInfer-Bench baselines
python
10.2µs
Reproduction-ready
Apache-2.0 · source
torch / matmul756deaFlashInfer-Bench baselines
python
10.7µs
Reported
Apache-2.0 · source
torch / matmul73342cFlashInfer-Bench baselines
python
10.8µs
Reported
Apache-2.0 · source
torch / matmul075b0dFlashInfer-Bench baselines
python
12.0µs
Reported
Apache-2.0 · source
flashinfer / wrapperece89aFlashInfer-Bench baselines
python
12.3µs
Reported
Apache-2.0 · source
deepgemm_fp8_fp4_gemm_nt_n2048_k2048FlashInfer-Bench baselines
python
13.5µs
Reported
Apache-2.0 · source
torch / matmul926adcFlashInfer-Bench baselines
python
13.8µs
Reported
Apache-2.0 · source
flashinfer / wrapperad9a00FlashInfer-Bench baselines
python
14.5µs
Reported
Apache-2.0 · source
flashinfer / wrapper03f7b0FlashInfer-Bench baselines
python
15.1µs
Reported
Apache-2.0 · source
torch / matmul0d13dfFlashInfer-Bench baselines
python
15.6µs
Reported
Apache-2.0 · source
torch / matmul67278eFlashInfer-Bench baselines
python
15.7µs
Reported
Apache-2.0 · source
torch / matmulff95bcFlashInfer-Bench baselines
python
15.9µs
Reported
Apache-2.0 · source
torch_matmul_254647FlashInfer-Bench baselines
python
16.1µs
Reported
Apache-2.0 · source
flashinfer / wrapperea3787FlashInfer-Bench baselines
python
16.3µs
Reported
Apache-2.0 · source
flashinfer / wrapper5222a7FlashInfer-Bench baselines
python
16.7µs
Reproduction-ready
Apache-2.0 · source
flashinfer / wrappera3c91fFlashInfer-Bench baselines
python
16.7µs
Reproduction-ready
Apache-2.0 · source
flashinfer / wrapper69a524FlashInfer-Bench baselines
python
16.8µs
Reproduction-ready
Apache-2.0 · source
deepgemm_fp8_fp4_gemm_nt_n4096_k4096FlashInfer-Bench baselines
python
17.7µs
Reported
Apache-2.0 · source
flashinfer / wrapper4c2606FlashInfer-Bench baselines
python
18.1µs
Reported
Apache-2.0 · source
torch / matmul3b6488FlashInfer-Bench baselines
python
18.5µs
Reported
Apache-2.0 · source

Showing 40 of 90; the full list is in the API dossier below.

Record activity

Nov
Dec
Jan
Feb
465Mar
458Apr
May
64Jun
Jul
Aug
Sep
Oct
Sources: FlashInfer-Bench (2026-06-06) · Apache-2.0last observed 2026-06-06JSON