source stringclasses 13
values | gpu stringclasses 4
values | model stringclasses 13
values | kernel_family stringclasses 7
values | kernel_name stringclasses 328
values | dtype stringclasses 2
values | held_out bool 2
classes | M float64 1 16.4k ⌀ | N float64 8 248k ⌀ | K float64 2.05k 29.6k ⌀ | bs float64 1 1 ⌀ | seq float64 1 16.4k ⌀ | n_heads float64 8 64 ⌀ | head_dim float64 64 256 ⌀ | kv_heads float64 1 64 ⌀ | numel float64 64 531M ⌀ | op_type stringclasses 177
values | gpu_time_duration_ms float64 0 703 | launch_block_size float64 16 512 | launch_grid_size float64 1 1.04M | dram_bytes_sum float64 1.27 94.3B ⌀ | launch_registers_per_thread float64 16 255 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01034 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 14,336 | null | null | null | null | null | null | null | 0.01034 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01027 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00522 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00534 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01043 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 14,336 | null | null | null | null | null | null | null | 0.0103 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.0103 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00534 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00528 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01034 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 14,336 | null | null | null | null | null | null | null | 0.0103 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01043 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00531 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00534 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01037 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 14,336 | null | null | null | null | null | null | null | 0.01053 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01037 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00522 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00534 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.0103 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 14,336 | null | null | null | null | null | null | null | 0.01034 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.0103 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00531 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00541 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01024 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 14,336 | null | null | null | null | null | null | null | 0.0103 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01034 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00525 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00525 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01043 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 14,336 | null | null | null | null | null | null | null | 0.01034 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01053 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00534 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00541 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.0103 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 14,336 | null | null | null | null | null | null | null | 0.0104 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01024 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00547 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00541 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01043 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 14,336 | null | null | null | null | null | null | null | 0.0105 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01059 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00531 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00538 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01043 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 14,336 | null | null | null | null | null | null | null | 0.01059 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01059 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00534 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00541 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01034 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 14,336 | null | null | null | null | null | null | null | 0.01046 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.0103 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00538 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00547 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01053 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 14,336 | null | null | null | null | null | null | null | 0.01059 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01046 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00544 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00534 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.0104 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 14,336 | null | null | null | null | null | null | null | 0.01043 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01053 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00534 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 1,024 | 4,096 | null | null | null | null | null | null | null | 0.00541 | 512 | 256 | 1,060,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 4,096 | null | null | null | null | null | null | null | 0.01059 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | splitk_reduce | void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 0, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ... | bf16 | false | 128 | 4,096 | 14,336 | null | null | null | null | null | null | null | 0.0104 | 512 | 512 | 3,150,000 | 32 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01843 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01952 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01949 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.0191 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01914 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01904 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01968 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01955 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.0192 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.0191 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.0191 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.0199 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01965 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01917 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01946 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01923 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01968 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01952 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01936 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01923 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01917 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01965 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01942 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01914 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01914 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.0192 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01955 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01971 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.0192 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.0191 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.01914 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | flash_attn | void pytorch_flash::flash_fwd_kernel<Flash_fwd_kernel_traits<128, 128, 64, 4, 0, 0, cutlass::bfloat16_t, Flash_kernel_traits<128, 128, 64, 4, cutlass::bfloat16_t>>, 0, 1, 0, 0, 1, 1, 0, 0>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 32 | 128 | 8 | null | flash_attn | 0.0191 | 128 | 32 | 1,650,000 | 255 |
Llama-3.1-8B-Instruct_prefill | A100 | Llama-8B | copy | void at::vectorized_gather_kernel<16, long>(char *, char *, T2 *, int, long, long, long, long, bool) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 262,144 | other | 0.00938 | 256 | 256 | 1,060,000 | 32 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.