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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 3,584 | cast | 0.0031 | 128 | 7 | 16,260 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 3,584 | cast | 0.0031 | 128 | 7 | 18,050 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 86,528 | cast | 0.00355 | 128 | 169 | 222,850 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 86,528 | mul | 0.00349 | 128 | 169 | 190,240 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,096 | other | 0.00848 | 256 | 4 | 216,740 | 36 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 1,024 | residual | 0.00243 | 128 | 2 | 18,300 | 40 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 3,584 | cast | 0.0031 | 128 | 7 | 16,510 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 3,584 | cast | 0.00307 | 128 | 7 | 18,050 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 93,696 | cast | 0.00355 | 128 | 183 | 231,040 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 93,696 | mul | 0.00349 | 128 | 183 | 204,060 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 8,192 | other | 0.00781 | 512 | 4 | 227,740 | 32 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 1,024 | residual | 0.00256 | 128 | 2 | 18,820 | 40 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 3,584 | cast | 0.0031 | 128 | 7 | 16,770 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 3,584 | cast | 0.00307 | 128 | 7 | 18,050 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 100,352 | cast | 0.00355 | 128 | 196 | 239,230 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 100,352 | mul | 0.00346 | 128 | 196 | 218,400 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 2,048 | other | 0.00973 | 128 | 4 | 249,060 | 46 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 1,024 | residual | 0.0025 | 128 | 2 | 19,330 | 40 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 3,584 | cast | 0.00307 | 128 | 7 | 17,020 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,096 | cast | 0.00307 | 128 | 8 | 18,050 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 108,032 | cast | 0.00362 | 128 | 211 | 247,420 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 108,032 | mul | 0.00349 | 128 | 211 | 233,250 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 8,192 | other | 0.00816 | 512 | 4 | 256,420 | 32 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 1,024 | residual | 0.00253 | 128 | 2 | 19,840 | 40 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,096 | cast | 0.00314 | 128 | 8 | 17,280 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,096 | cast | 0.00307 | 128 | 8 | 18,050 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 115,200 | cast | 0.00358 | 128 | 225 | 255,620 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 115,200 | mul | 0.00355 | 128 | 225 | 248,610 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,096 | other | 0.00854 | 256 | 4 | 274,340 | 36 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 1,024 | residual | 0.00253 | 128 | 2 | 20,350 | 40 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,096 | cast | 0.0031 | 128 | 8 | 17,540 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,096 | cast | 0.0031 | 128 | 8 | 18,050 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 123,392 | cast | 0.00358 | 128 | 241 | 263,810 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 123,392 | mul | 0.00362 | 128 | 241 | 264,480 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 8,192 | other | 0.00829 | 512 | 4 | 287,140 | 32 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 1,024 | residual | 0.00256 | 128 | 2 | 20,860 | 40 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,096 | cast | 0.0031 | 128 | 8 | 17,790 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,096 | cast | 0.00304 | 128 | 8 | 18,050 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 131,072 | cast | 0.00368 | 128 | 256 | 272,000 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 131,072 | mul | 0.00362 | 128 | 256 | 280,860 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 2,048 | other | 0.01142 | 128 | 4 | 310,500 | 46 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 1,024 | residual | 0.00234 | 128 | 2 | 21,380 | 40 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,096 | cast | 0.00314 | 128 | 8 | 18,050 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,608 | cast | 0.00307 | 128 | 9 | 22,140 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | cast | 0.00387 | 128 | 273 | 415,360 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | mul | 0.00352 | 128 | 273 | 297,760 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 10,240 | other | 0.00854 | 512 | 5 | 319,900 | 32 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.00234 | 128 | 3 | 21,890 | 40 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,608 | cast | 0.0031 | 128 | 9 | 18,300 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,608 | cast | 0.00307 | 128 | 9 | 22,140 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | cast | 0.00381 | 128 | 289 | 427,650 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | mul | 0.00371 | 128 | 289 | 315,170 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.00896 | 256 | 5 | 339,620 | 36 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.00234 | 128 | 3 | 22,400 | 40 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,608 | cast | 0.00314 | 128 | 9 | 18,560 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,608 | cast | 0.0031 | 128 | 9 | 22,140 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | cast | 0.00387 | 128 | 307 | 439,940 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | mul | 0.00371 | 128 | 307 | 333,090 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 10,240 | other | 0.00858 | 512 | 5 | 354,720 | 32 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.00237 | 128 | 3 | 22,910 | 40 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,608 | cast | 0.00307 | 128 | 9 | 18,820 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,608 | cast | 0.00307 | 128 | 9 | 22,140 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | cast | 0.00387 | 128 | 324 | 452,220 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | mul | 0.00378 | 128 | 324 | 351,520 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 2,560 | other | 0.01027 | 128 | 5 | 380,130 | 46 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.00234 | 128 | 3 | 23,420 | 40 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 4,608 | cast | 0.0031 | 128 | 9 | 19,070 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.00307 | 128 | 10 | 22,140 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | cast | 0.00384 | 128 | 343 | 464,510 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | mul | 0.00374 | 128 | 343 | 370,460 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 10,240 | other | 0.00906 | 512 | 5 | 391,580 | 32 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.00237 | 128 | 3 | 23,940 | 40 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.0031 | 128 | 10 | 19,330 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.0031 | 128 | 10 | 22,140 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | cast | 0.0039 | 128 | 361 | 476,800 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | mul | 0.00384 | 128 | 361 | 389,920 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.00938 | 256 | 5 | 413,600 | 36 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.00237 | 128 | 3 | 24,450 | 40 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.0031 | 128 | 10 | 19,580 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.00307 | 128 | 10 | 22,140 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | cast | 0.0039 | 128 | 381 | 489,090 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | mul | 0.00394 | 128 | 381 | 409,890 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 10,240 | other | 0.00912 | 512 | 5 | 430,500 | 32 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.0024 | 128 | 3 | 24,960 | 40 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.00314 | 128 | 10 | 19,840 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.0032 | 128 | 10 | 22,910 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | cast | 0.00387 | 128 | 400 | 501,380 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | mul | 0.00384 | 128 | 400 | 430,370 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 2,560 | other | 0.01242 | 128 | 5 | 457,950 | 46 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.00243 | 128 | 3 | 25,470 | 40 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.0031 | 128 | 10 | 20,100 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.00304 | 128 | 11 | 22,140 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | cast | 0.0039 | 128 | 421 | 513,660 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | mul | 0.0039 | 128 | 421 | 451,360 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.00938 | 512 | 6 | 471,460 | 32 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.0025 | 128 | 3 | 25,980 | 40 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.0031 | 128 | 11 | 20,350 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 0.00301 | 128 | 11 | 22,140 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | cast | 0.00403 | 128 | 441 | 525,950 | 16 |
Qwen3.5-9B_prefill | RTX2080Ti | Qwen3.5-9B | 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 | 262,144 | mul | 0.00403 | 128 | 441 | 472,860 | 16 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.