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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00861 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01338 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00838 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01299 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00906 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01283 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00854 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01357 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00845 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01277 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00851 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01344 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00861 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01414 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00877 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01312 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00867 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01344 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00848 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01392 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.0089 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01283 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.0088 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01286 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00877 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01267 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00867 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01296 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00861 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01267 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00851 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01274 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00867 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01274 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.0089 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01264 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00874 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01299 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00877 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01325 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00854 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01312 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00848 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01286 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00842 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01277 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00864 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01267 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00877 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01312 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00848 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.0127 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00864 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01264 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00861 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (bool)1, (bool)0, (bool)0, (bool)0, (bool)1, (bool)0, (bool)1, (bool)0>(pytorch_flash::Flash_fwd_... | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.01277 | 128 | 64 | 2,180,000 | 255 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | flash_attn | void pytorch_flash::flash_fwd_splitkv_combine_kernel<Flash_fwd_kernel_traits<(int)256, (int)64, (int)64, (int)4, (bool)0, (bool)0, cutlass::bfloat16_t, Flash_kernel_traits<(int)256, (int)64, (int)64, (int)4, cutlass::bfloat16_t>>, (int)4, (int)1, (bool)1>(pytorch_flash::Flash_fwd_params) | bf16 | false | null | null | null | 1 | 128 | 16 | 224 | 8 | null | flash_attn | 0.00899 | 128 | 512 | 4,240,000 | 54 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::elementwise_kernel<(int)128, (int)4, void at::native::gpu_kernel_impl_nocast<at::native::neg_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 9)]::operator ()() const::[lambda(c10::BFloat16) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[l... | bf16 | false | null | null | null | 1 | 128 | null | null | null | 131,072 | neg | 0.00611 | 128 | 256 | 285,180 | 18 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | copy | void at::native::<unnamed>::CatArrayBatchedCopy<at::native::<unnamed>::OpaqueType<(unsigned int)2>, unsigned int, (int)4, (int)64, (int)64>(T1 *, at::native::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::native::<unnamed>::TensorSizeStride<T2, (unsigned int)4>, int, T2) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 917,504 | other | 0.00726 | 512 | 528 | 531,200 | 25 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::elementwise_kernel<(int)128, (int)4, void at::native::gpu_kernel_impl_nocast<at::native::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, at::native::binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 458,752 | mul | 0.00646 | 128 | 512 | 616,960 | 18 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::elementwise_kernel<(int)128, (int)4, void at::native::gpu_kernel_impl_nocast<at::native::CUDAFunctor_add<c10::BFloat16>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 458,752 | residual | 0.00653 | 128 | 512 | 1,080,000 | 18 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | copy | void at::native::<unnamed>::CatArrayBatchedCopy<at::native::<unnamed>::OpaqueType<(unsigned int)2>, unsigned int, (int)4, (int)64, (int)64>(T1 *, at::native::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::native::<unnamed>::TensorSizeStride<T2, (unsigned int)4>, int, T2) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 917,504 | other | 0.00704 | 512 | 528 | 531,200 | 25 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | copy | void at::native::<unnamed>::CatArrayBatchedCopy<at::native::<unnamed>::OpaqueType<(unsigned int)2>, unsigned int, (int)4, (int)64, (int)64>(T1 *, at::native::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::native::<unnamed>::TensorSizeStride<T2, (unsigned int)4>, int, T2) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 917,504 | other | 0.00726 | 512 | 528 | 531,460 | 25 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | cast | void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int... | bf16 | false | null | null | null | 1 | 128 | null | null | null | 458,752 | cast | 0.00688 | 128 | 896 | 950,270 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)4, void at::native::<unnamed>::pow_tensor_scalar_kernel_impl<float, float>(at::TensorIteratorBase &, T2)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 229,376 | rmsnorm | 0.00426 | 128 | 448 | 1,840,000 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | reduce | void at::native::reduce_kernel<(int)512, (int)1, at::native::ReduceOp<float, at::native::MeanOps<float, float, float, float>, unsigned int, float, (int)4, (int)4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 16,384 | rmsnorm | 0.01734 | 512 | 8 | 1,900,000 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)4, at::native::CUDAFunctorOnSelf_add<float>, std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 512 | residual | 0.00397 | 128 | 1 | 5,890 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)4, at::native::rsqrt_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 512 | rmsnorm | 0.00384 | 128 | 1 | 7,170 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 917,504 | mul | 0.00614 | 128 | 1,792 | 1,850,000 | 18 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | cast | void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int... | bf16 | false | null | null | null | 1 | 128 | null | null | null | 3,584 | cast | 0.0055 | 128 | 7 | 39,680 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)4, at::native::CUDAFunctorOnSelf_add<float>, std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 2,048 | residual | 0.00403 | 128 | 4 | 19,710 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 917,504 | mul | 0.00598 | 128 | 1,792 | 1,860,000 | 18 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | cast | void at::native::vectorized_elementwise_kernel<(int)8, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 229,376 | cast | 0.00467 | 128 | 448 | 1,840,000 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)8, at::native::CUDAFunctor_add<c10::BFloat16>, std::array<char *, (unsigned long)3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 229,376 | residual | 0.00429 | 128 | 448 | 1,840,000 | 34 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | cast | void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int... | bf16 | false | null | null | null | 1 | 128 | null | null | null | 458,752 | cast | 0.00614 | 128 | 896 | 950,270 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)4, void at::native::<unnamed>::pow_tensor_scalar_kernel_impl<float, float>(at::TensorIteratorBase &, T2)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 229,376 | rmsnorm | 0.00496 | 128 | 448 | 1,840,000 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | reduce | void at::native::reduce_kernel<(int)512, (int)1, at::native::ReduceOp<float, at::native::MeanOps<float, float, float, float>, unsigned int, float, (int)4, (int)4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 16,384 | rmsnorm | 0.01718 | 512 | 8 | 1,900,000 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)4, at::native::CUDAFunctorOnSelf_add<float>, std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 512 | residual | 0.00384 | 128 | 1 | 5,890 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)4, at::native::rsqrt_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 512 | rmsnorm | 0.0039 | 128 | 1 | 7,170 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 917,504 | mul | 0.00602 | 128 | 1,792 | 1,850,000 | 18 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | cast | void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int... | bf16 | false | null | null | null | 1 | 128 | null | null | null | 3,584 | cast | 0.00579 | 128 | 7 | 39,940 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)4, at::native::CUDAFunctorOnSelf_add<float>, std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 2,048 | residual | 0.00374 | 128 | 4 | 19,710 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 917,504 | mul | 0.00624 | 128 | 1,792 | 1,860,000 | 18 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | cast | void at::native::vectorized_elementwise_kernel<(int)8, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 229,376 | cast | 0.00422 | 128 | 448 | 1,840,000 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)8, at::native::GeluCUDAKernelImpl(at::TensorIteratorBase &, at::native::GeluType)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 4)]::operator ()() const::[lambda(c10::BFloat16) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 917,504 | other | 0.00653 | 128 | 1,792 | 3,680,000 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)8, at::native::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, at::native::binary_internal::MulFunctor<float>>, std::array<char *, (unsigned long)3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 917,504 | mul | 0.00688 | 128 | 1,792 | 7,350,000 | 40 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | cast | void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int... | bf16 | false | null | null | null | 1 | 128 | null | null | null | 458,752 | cast | 0.0064 | 128 | 896 | 950,270 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)4, void at::native::<unnamed>::pow_tensor_scalar_kernel_impl<float, float>(at::TensorIteratorBase &, T2)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 229,376 | rmsnorm | 0.0049 | 128 | 448 | 1,840,000 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | reduce | void at::native::reduce_kernel<(int)512, (int)1, at::native::ReduceOp<float, at::native::MeanOps<float, float, float, float>, unsigned int, float, (int)4, (int)4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 16,384 | rmsnorm | 0.01709 | 512 | 8 | 1,900,000 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)4, at::native::CUDAFunctorOnSelf_add<float>, std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 512 | residual | 0.00397 | 128 | 1 | 5,890 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)4, at::native::rsqrt_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 512 | rmsnorm | 0.0039 | 128 | 1 | 7,170 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 917,504 | mul | 0.00618 | 128 | 1,792 | 1,850,000 | 18 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | cast | void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int... | bf16 | false | null | null | null | 1 | 128 | null | null | null | 3,584 | cast | 0.00611 | 128 | 7 | 39,940 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)4, at::native::CUDAFunctorOnSelf_add<float>, std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 2,048 | residual | 0.0039 | 128 | 4 | 19,710 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::elementwise_kernel<(int)128, (int)2, void at::native::gpu_kernel_impl_nocast<at::native::BinaryFunctor<float, float, float, at::native::binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 917,504 | mul | 0.00605 | 128 | 1,792 | 1,860,000 | 18 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | cast | void at::native::vectorized_elementwise_kernel<(int)8, at::native::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 229,376 | cast | 0.00502 | 128 | 448 | 1,840,000 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)8, at::native::CUDAFunctor_add<c10::BFloat16>, std::array<char *, (unsigned long)3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 229,376 | residual | 0.00438 | 128 | 448 | 1,840,000 | 34 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | cast | void at::native::unrolled_elementwise_kernel<at::native::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>, (int)4, TrivialOffsetCalculator<(int)1, unsigned int... | bf16 | false | null | null | null | 1 | 128 | null | null | null | 458,752 | cast | 0.00666 | 128 | 896 | 950,270 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | elementwise | void at::native::vectorized_elementwise_kernel<(int)4, void at::native::<unnamed>::pow_tensor_scalar_kernel_impl<float, float>(at::TensorIteratorBase &, T2)::[lambda(float) (instance 1)], std::array<char *, (unsigned long)2>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 229,376 | rmsnorm | 0.00512 | 128 | 448 | 1,840,000 | 32 |
gemma-2-9b-it_prefill | H100 | Gemma-9B | reduce | void at::native::reduce_kernel<(int)512, (int)1, at::native::ReduceOp<float, at::native::MeanOps<float, float, float, float>, unsigned int, float, (int)4, (int)4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 16,384 | rmsnorm | 0.01706 | 512 | 8 | 1,900,000 | 32 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.