Skip to content
KernelIndex
Search⌘K
Sign in

NVIDIA A100

sm_80
Records held
178
Runs
933
Operations
22560 implementations

Records held on this GPU

Operation · workload
History
Record
Implementation
Since
GRPO loss, token-level importance samplingbf16 · [2, 1025, 128256]
23.7ms
PyTorch
2025-08-05
GRPO loss, token-level importance samplingbf16 · [4, 1025, 128256]
46.9ms
PyTorch
2025-08-04
GRPO loss, token-level importance samplingbf16 · [8, 1025, 128256]
94.5ms
PyTorch
2025-08-04
GRPO loss, token-level importance samplingbf16 · [16, 1025, 128256]
190.0ms
PyTorch
2025-08-04
GRPO loss, token-level importance samplingbf16 · [2, 1025, 128256]
42.3ms
Liger
2025-08-05
GRPO loss, token-level importance samplingbf16 · [4, 1025, 128256]
82.2ms
Liger
2025-08-05
GRPO loss, token-level importance samplingbf16 · [8, 1025, 128256]
163.4ms
Liger
2025-08-05
GRPO loss, token-level importance samplingbf16 · [16, 1025, 128256]
325.4ms
Liger
2025-08-05
GRPO loss, token-level importance samplingbf16 · [2, 1025, 128256]
1.81ms
Liger
2025-08-05
GRPO loss, token-level importance samplingbf16 · [4, 1025, 128256]
1.85ms
Liger
2025-08-05
GRPO loss, token-level importance samplingbf16 · [8, 1025, 128256]
1.89ms
Liger
2025-08-05
GRPO loss, token-level importance samplingbf16 · [16, 1025, 128256]
1.97ms
Liger
2025-08-05
Show all 40 records ›
Chunked distillation cosine lossbf16 · [1024]
13.8ms
Liger
2025-06-27
Chunked distillation cosine lossbf16 · [2048]
27.6ms
Liger
2025-06-27
Chunked distillation cosine lossbf16 · [4096]
54.8ms
Liger
2025-06-27
Chunked distillation cosine lossbf16 · [8192]
109.7ms
Liger
2025-06-27
Chunked distillation cosine lossbf16 · [1024]
14.7ms
Liger
2025-06-27
Chunked distillation cosine lossbf16 · [2048]
28.3ms
Liger
2025-06-27
Chunked distillation cosine lossbf16 · [4096]
56.0ms
Liger
2025-06-27
Chunked distillation cosine lossbf16 · [8192]
111.4ms
Liger
2025-06-27
Fused linear + CPO lossbf16 · [2, 1024, 4096]
14.5ms
Transformers
2024-11-14
Fused linear + CPO lossbf16 · [4, 1024, 4096]
26.1ms
Transformers
2024-11-14
Fused linear + CPO lossbf16 · [8, 1024, 4096]
51.9ms
Transformers
2024-11-14
Fused linear + CPO lossbf16 · [16, 1024, 4096]
105.0ms
Transformers
2024-11-14
Fused linear + CPO lossbf16 · [2, 1024, 4096]
33.3ms
Liger
2024-11-14
Fused linear + CPO lossbf16 · [4, 1024, 4096]
63.1ms
Liger
2024-11-14
Fused linear + CPO lossbf16 · [8, 1024, 4096]
125.5ms
Liger
2024-11-14
Fused linear + CPO lossbf16 · [16, 1024, 4096]
250.2ms
Liger
2024-11-14
Fused linear + SimPO lossbf16 · [2, 1024, 4096]
14.5ms
Transformers
2024-11-15
Fused linear + SimPO lossbf16 · [4, 1024, 4096]
28.8ms
Transformers
2024-11-15
Fused linear + SimPO lossbf16 · [8, 1024, 4096]
52.6ms
Transformers
2024-11-15
Fused linear + SimPO lossbf16 · [16, 1024, 4096]
104.4ms
Transformers
2024-11-15
Fused linear + SimPO lossbf16 · [2, 1024, 4096]
31.6ms
Liger
2024-11-15
Fused linear + SimPO lossbf16 · [4, 1024, 4096]
61.4ms
Liger
2024-11-15
Fused linear + SimPO lossbf16 · [8, 1024, 4096]
120.0ms
Liger
2024-11-15
Fused linear + SimPO lossbf16 · [16, 1024, 4096]
238.1ms
Liger
2024-11-15
Fused linear + ORPO lossbf16 · [2, 4096, 4096]
39.6ms
Transformers
2024-11-13
Fused linear + ORPO lossbf16 · [4, 4096, 4096]
80.0ms
Transformers
2024-11-13
Fused linear + ORPO lossbf16 · [8, 4096, 4096]
160.1ms
Transformers
2024-11-13
Fused linear + ORPO lossbf16 · [16, 4096, 4096]
321.5ms
Transformers
2024-11-13

138 more records · Open in the records ledger →

Coverage by operation family

Family
Operations
Runs
With source
elementwise
2
224
180
reduction
1
96
72
embedding
1
96
96
trimul
1
69
67
grpo
1
48
48
rope
1
48
48
conv
1
40
40
cross-entropy
2
40
40
rmsnorm
1
36
36
orpo
1
32
32
sort
1
28
25
gemm
1
27
27
scan
1
25
24
geglu
1
24
24
histogram
1
24
24
layernorm
1
20
20
cosine-similarity-loss
1
16
16
cpo
1
16
16
simpo
1
16
16
swiglu
1
8
8

Record activity

25Nov
10Dec
3Jan
2Feb
12Mar
2Apr
3May
Jun
Jul
Aug
Sep
Oct
Sources: GPU MODE KernelBot (2026-05-15) · June 9 Researcher Reciprocity License v1.0 · Liger-Kernel benchmarks (2025-08-05) · BSD-2-Clauselast observed 2026-05-15