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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00573 | 128 | 254 | 329,600 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.00554 | 128 | 254 | 280,700 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<256, 2, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 5,120 | other | 0.01155 | 256 | 5 | 320,380 | 34 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 1,536 | residual | 0.004 | 128 | 3 | 25,340 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 5,120 | cast | 0.00515 | 128 | 10 | 20,100 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 5,632 | cast | 0.00534 | 128 | 11 | 22,530 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00576 | 128 | 274 | 341,890 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.00566 | 128 | 274 | 301,440 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<512, 1, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 12,288 | other | 0.01142 | 512 | 6 | 330,750 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 1,536 | residual | 0.0039 | 128 | 3 | 26,110 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 5,632 | cast | 0.00506 | 128 | 11 | 20,480 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 5,632 | cast | 0.00512 | 128 | 11 | 22,530 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00595 | 128 | 294 | 354,180 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.00582 | 128 | 294 | 322,940 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<128, 4, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 3,072 | other | 0.01258 | 128 | 6 | 371,330 | 48 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 1,536 | residual | 0.00397 | 128 | 3 | 26,750 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 5,632 | cast | 0.00509 | 128 | 11 | 20,860 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 5,632 | cast | 0.00534 | 128 | 11 | 22,400 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00576 | 128 | 316 | 366,460 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.00563 | 128 | 316 | 345,220 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<512, 1, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 12,288 | other | 0.01174 | 512 | 6 | 373,760 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 1,536 | residual | 0.00403 | 128 | 3 | 27,520 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 5,632 | cast | 0.00515 | 128 | 11 | 21,250 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 6,144 | cast | 0.00518 | 128 | 12 | 22,530 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00576 | 128 | 338 | 378,750 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.0057 | 128 | 338 | 368,130 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<256, 2, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 6,144 | other | 0.01238 | 256 | 6 | 406,400 | 34 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 1,536 | residual | 0.00394 | 128 | 3 | 28,290 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 6,144 | cast | 0.00518 | 128 | 12 | 21,630 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 6,144 | cast | 0.00522 | 128 | 12 | 22,400 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00592 | 128 | 361 | 391,040 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.00576 | 128 | 361 | 392,060 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<512, 1, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 12,288 | other | 0.01213 | 512 | 6 | 419,970 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 1,536 | residual | 0.0041 | 128 | 3 | 29,180 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 6,144 | cast | 0.00506 | 128 | 12 | 22,020 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 6,144 | cast | 0.00512 | 128 | 12 | 22,530 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00589 | 128 | 384 | 403,330 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.00576 | 128 | 384 | 416,510 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<128, 4, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 3,072 | other | 0.01366 | 128 | 6 | 463,490 | 48 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 1,536 | residual | 0.00387 | 128 | 3 | 29,820 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 6,144 | cast | 0.00509 | 128 | 12 | 22,400 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 6,656 | cast | 0.00506 | 128 | 13 | 28,540 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00576 | 128 | 409 | 618,370 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.00582 | 128 | 409 | 441,980 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<512, 1, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 14,336 | other | 0.01235 | 512 | 7 | 469,120 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 2,048 | residual | 0.00368 | 128 | 4 | 30,590 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 6,656 | cast | 0.00512 | 128 | 13 | 22,780 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 6,656 | cast | 0.00515 | 128 | 13 | 28,540 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00582 | 128 | 434 | 636,800 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.00563 | 128 | 434 | 467,970 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<256, 2, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 7,168 | other | 0.0127 | 256 | 7 | 504,700 | 34 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 2,048 | residual | 0.00365 | 128 | 4 | 31,360 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 6,656 | cast | 0.00509 | 128 | 13 | 23,170 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 7,168 | cast | 0.00522 | 128 | 14 | 28,540 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00595 | 128 | 460 | 655,230 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.00576 | 128 | 460 | 494,850 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<512, 1, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 14,336 | other | 0.01283 | 512 | 7 | 521,220 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 2,048 | residual | 0.00368 | 128 | 4 | 32,130 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 7,168 | cast | 0.00538 | 128 | 14 | 23,550 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 7,168 | cast | 0.00518 | 128 | 14 | 28,540 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00589 | 128 | 486 | 673,660 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.00582 | 128 | 486 | 522,500 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<128, 4, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 3,584 | other | 0.01398 | 128 | 7 | 567,940 | 48 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 2,048 | residual | 0.00381 | 128 | 4 | 33,150 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 7,168 | cast | 0.00502 | 128 | 14 | 23,940 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 7,168 | cast | 0.00509 | 128 | 14 | 28,540 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00586 | 128 | 514 | 692,100 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.00579 | 128 | 514 | 550,910 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<512, 1, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 14,336 | other | 0.01309 | 512 | 7 | 576,510 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 2,048 | residual | 0.00374 | 128 | 4 | 33,660 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 7,168 | cast | 0.00515 | 128 | 14 | 24,320 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 7,680 | cast | 0.00522 | 128 | 15 | 28,540 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00592 | 128 | 542 | 710,530 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.00586 | 128 | 542 | 580,100 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<256, 2, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 8,192 | other | 0.01344 | 256 | 8 | 615,300 | 34 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 2,048 | residual | 0.00365 | 128 | 4 | 34,430 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 7,680 | cast | 0.00528 | 128 | 15 | 24,700 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 7,680 | cast | 0.00522 | 128 | 15 | 28,540 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00592 | 128 | 571 | 728,960 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.00589 | 128 | 571 | 610,180 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<512, 1, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 16,384 | other | 0.01331 | 512 | 8 | 635,010 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 2,048 | residual | 0.00368 | 128 | 4 | 35,200 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 7,680 | cast | 0.00506 | 128 | 15 | 25,090 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 7,680 | cast | 0.00512 | 128 | 15 | 28,540 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00598 | 128 | 600 | 747,390 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.00592 | 128 | 600 | 640,770 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<128, 4, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 4,096 | other | 0.01472 | 128 | 8 | 684,670 | 48 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 2,048 | residual | 0.00397 | 128 | 4 | 35,970 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 7,680 | cast | 0.00518 | 128 | 15 | 25,470 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 8,192 | cast | 0.00515 | 128 | 16 | 28,540 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00592 | 128 | 631 | 765,820 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.00598 | 128 | 631 | 672,260 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<512, 1, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 16,384 | other | 0.0137 | 512 | 8 | 696,320 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 2,048 | residual | 0.00378 | 128 | 4 | 36,740 | 32 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 8,192 | cast | 0.00525 | 128 | 16 | 25,860 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 8,192 | cast | 0.00515 | 128 | 16 | 28,670 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | cast | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | cast | 0.00598 | 128 | 662 | 784,260 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::elementwise_kernel<128, 2, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<float, float, float, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 218,112 | mul | 0.00589 | 128 | 662 | 704,640 | 16 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | reduce | void at::reduce_kernel<256, 2, at::ReduceOp<float, at::func_wrapper_t<float, at::sum_functor<float, float, float>::operator ()(at::TensorIterator &)::[lambda(float, float) (instance 1)]>, unsigned int, float, 4, 4>>(T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 8,192 | other | 0.01408 | 256 | 8 | 738,300 | 34 |
Qwen3.5-27B_prefill | A100 | Qwen3.5-27B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<float>, std::array<char *, 3>>(int, T2, T3) | bf16 | false | null | null | null | 1 | 128 | null | null | null | 2,048 | residual | 0.00381 | 128 | 4 | 37,630 | 32 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.