source stringclasses 13
values | gpu stringclasses 4
values | model stringclasses 13
values | kernel_family stringclasses 7
values | kernel_name stringclasses 328
values | dtype stringclasses 2
values | held_out bool 2
classes | M float64 1 16.4k ⌀ | N float64 8 248k ⌀ | K float64 2.05k 29.6k ⌀ | bs float64 1 1 ⌀ | seq float64 1 16.4k ⌀ | n_heads float64 8 64 ⌀ | head_dim float64 64 256 ⌀ | kv_heads float64 1 64 ⌀ | numel float64 64 531M ⌀ | op_type stringclasses 177
values | gpu_time_duration_ms float64 0 703 | launch_block_size float64 16 512 | launch_grid_size float64 1 1.04M | dram_bytes_sum float64 1.27 94.3B ⌀ | launch_registers_per_thread float64 16 255 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | reduce | void at::reduce_kernel<512, 1, at::ReduceOp<float, at::MeanOps<float, float, float, float>, unsigned int, float, 4, 4>>(T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 262,144 | rmsnorm | 0.01059 | 512 | 128 | 4,250,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctorOnSelf_add<float>, std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 512 | residual | 0.00339 | 128 | 1 | 5,250 | 30 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::rsqrt_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 512 | rmsnorm | 0.00358 | 128 | 1 | 6,910 | 22 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | 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 | true | null | null | null | 1 | 128 | null | null | null | 2,097,152 | mul | 0.01021 | 128 | 4,096 | 4,210,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | cast | void at::vectorized_elementwise_kernel<4, at::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | cast | 0.00707 | 128 | 1,024 | 4,200,000 | 28 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | mul | 0.01085 | 128 | 2,048 | 2,130,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::<unnamed>::silu_kernel(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 6)]::operator ()() const::[lambda(c10::BFloat16) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 2,097,152 | silu | 0.01357 | 128 | 3,584 | 7,350,000 | 28 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>, std::array<char *, 3>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 2,097,152 | mul | 0.01443 | 128 | 3,584 | 14,690,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<c10::BFloat16>, std::array<char *, 3>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | residual | 0.00698 | 128 | 1,024 | 4,200,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | cast | void at::unrolled_elementwise_kernel<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, 2>, 4, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned in... | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | cast | 0.01251 | 128 | 2,048 | 2,130,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, void at::<unnamed>::pow_tensor_scalar_kernel_impl<float, float>(at::TensorIteratorBase &, T2)::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | rmsnorm | 0.00749 | 128 | 1,024 | 4,200,000 | 28 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | reduce | void at::reduce_kernel<512, 1, at::ReduceOp<float, at::MeanOps<float, float, float, float>, unsigned int, float, 4, 4>>(T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 262,144 | rmsnorm | 0.01059 | 512 | 128 | 4,250,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctorOnSelf_add<float>, std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 512 | residual | 0.00342 | 128 | 1 | 5,250 | 30 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::rsqrt_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 512 | rmsnorm | 0.00358 | 128 | 1 | 6,780 | 22 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | 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 | true | null | null | null | 1 | 128 | null | null | null | 2,097,152 | mul | 0.01037 | 128 | 4,096 | 4,210,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | cast | void at::vectorized_elementwise_kernel<4, at::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | cast | 0.00694 | 128 | 1,024 | 4,200,000 | 28 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | mul | 0.01098 | 128 | 2,048 | 2,130,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | mul | 0.01184 | 128 | 2,048 | 2,150,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::neg_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 9)]::operator ()() const::[lambda(c10::BFloat16) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | neg | 0.00787 | 128 | 1,024 | 1,070,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | copy | void at::<unnamed>::CatArrayBatchedCopy<at::<unnamed>::OpaqueType<2>, unsigned int, 4, 64, 64>(T1 *, at::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::<unnamed>::TensorSizeStride<T2, 4>, int, T2) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | other | 0.01779 | 512 | 432 | 2,100,000 | 23 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | mul | 0.01098 | 128 | 2,048 | 2,150,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::CUDAFunctor_add<c10::BFloat16>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | residual | 0.01213 | 128 | 2,048 | 4,210,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 131,072 | mul | 0.00739 | 128 | 256 | 314,620 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::neg_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 9)]::operator ()() const::[lambda(c10::BFloat16) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 65,536 | neg | 0.00701 | 128 | 128 | 148,860 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | copy | void at::<unnamed>::CatArrayBatchedCopy<at::<unnamed>::OpaqueType<2>, unsigned int, 4, 64, 64>(T1 *, at::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::<unnamed>::TensorSizeStride<T2, 4>, int, T2) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | other | 0.00986 | 512 | 432 | 269,060 | 23 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 131,072 | mul | 0.00723 | 128 | 256 | 314,620 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::CUDAFunctor_add<c10::BFloat16>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 131,072 | residual | 0.00739 | 128 | 256 | 544,130 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | copy | void at::<unnamed>::CatArrayBatchedCopy<at::<unnamed>::OpaqueType<2>, unsigned int, 4, 64, 64>(T1 *, at::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::<unnamed>::TensorSizeStride<T2, 4>, int, T2) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | other | 0.00864 | 512 | 432 | 269,310 | 23 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | copy | void at::<unnamed>::CatArrayBatchedCopy<at::<unnamed>::OpaqueType<2>, unsigned int, 4, 64, 64>(T1 *, at::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::<unnamed>::TensorSizeStride<T2, 4>, int, T2) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | other | 0.00867 | 512 | 432 | 269,180 | 23 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<c10::BFloat16>, std::array<char *, 3>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | residual | 0.0071 | 128 | 1,024 | 4,200,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | cast | void at::unrolled_elementwise_kernel<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, 2>, 4, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned in... | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | cast | 0.01286 | 128 | 2,048 | 2,130,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, void at::<unnamed>::pow_tensor_scalar_kernel_impl<float, float>(at::TensorIteratorBase &, T2)::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | rmsnorm | 0.00774 | 128 | 1,024 | 4,200,000 | 28 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | reduce | void at::reduce_kernel<512, 1, at::ReduceOp<float, at::MeanOps<float, float, float, float>, unsigned int, float, 4, 4>>(T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 262,144 | rmsnorm | 0.01062 | 512 | 128 | 4,250,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctorOnSelf_add<float>, std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 512 | residual | 0.00333 | 128 | 1 | 5,250 | 30 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::rsqrt_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 512 | rmsnorm | 0.00352 | 128 | 1 | 6,910 | 22 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | 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 | true | null | null | null | 1 | 128 | null | null | null | 2,097,152 | mul | 0.0103 | 128 | 4,096 | 4,210,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | cast | void at::vectorized_elementwise_kernel<4, at::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | cast | 0.00707 | 128 | 1,024 | 4,200,000 | 28 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | mul | 0.01091 | 128 | 2,048 | 2,130,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::<unnamed>::silu_kernel(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 6)]::operator ()() const::[lambda(c10::BFloat16) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 2,097,152 | silu | 0.01376 | 128 | 3,584 | 7,350,000 | 28 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>, std::array<char *, 3>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 2,097,152 | mul | 0.01392 | 128 | 3,584 | 14,690,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<c10::BFloat16>, std::array<char *, 3>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | residual | 0.0071 | 128 | 1,024 | 4,200,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | cast | void at::unrolled_elementwise_kernel<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, 2>, 4, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned in... | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | cast | 0.0128 | 128 | 2,048 | 2,130,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, void at::<unnamed>::pow_tensor_scalar_kernel_impl<float, float>(at::TensorIteratorBase &, T2)::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | rmsnorm | 0.00752 | 128 | 1,024 | 4,200,000 | 28 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | reduce | void at::reduce_kernel<512, 1, at::ReduceOp<float, at::MeanOps<float, float, float, float>, unsigned int, float, 4, 4>>(T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 262,144 | rmsnorm | 0.01056 | 512 | 128 | 4,250,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctorOnSelf_add<float>, std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 512 | residual | 0.00339 | 128 | 1 | 5,250 | 30 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::rsqrt_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 512 | rmsnorm | 0.00352 | 128 | 1 | 6,780 | 22 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | 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 | true | null | null | null | 1 | 128 | null | null | null | 2,097,152 | mul | 0.01024 | 128 | 4,096 | 4,210,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | cast | void at::vectorized_elementwise_kernel<4, at::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | cast | 0.00717 | 128 | 1,024 | 4,200,000 | 28 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | mul | 0.01104 | 128 | 2,048 | 2,130,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | mul | 0.01187 | 128 | 2,048 | 2,150,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::neg_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 9)]::operator ()() const::[lambda(c10::BFloat16) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | neg | 0.00816 | 128 | 1,024 | 1,070,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | copy | void at::<unnamed>::CatArrayBatchedCopy<at::<unnamed>::OpaqueType<2>, unsigned int, 4, 64, 64>(T1 *, at::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::<unnamed>::TensorSizeStride<T2, 4>, int, T2) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | other | 0.01766 | 512 | 432 | 2,100,000 | 23 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | mul | 0.01091 | 128 | 2,048 | 2,150,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::CUDAFunctor_add<c10::BFloat16>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | residual | 0.01213 | 128 | 2,048 | 4,210,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 131,072 | mul | 0.00723 | 128 | 256 | 314,620 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::neg_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 9)]::operator ()() const::[lambda(c10::BFloat16) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 65,536 | neg | 0.0071 | 128 | 128 | 148,860 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | copy | void at::<unnamed>::CatArrayBatchedCopy<at::<unnamed>::OpaqueType<2>, unsigned int, 4, 64, 64>(T1 *, at::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::<unnamed>::TensorSizeStride<T2, 4>, int, T2) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | other | 0.00979 | 512 | 432 | 269,180 | 23 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 131,072 | mul | 0.00717 | 128 | 256 | 314,750 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::CUDAFunctor_add<c10::BFloat16>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 131,072 | residual | 0.00746 | 128 | 256 | 544,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | copy | void at::<unnamed>::CatArrayBatchedCopy<at::<unnamed>::OpaqueType<2>, unsigned int, 4, 64, 64>(T1 *, at::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::<unnamed>::TensorSizeStride<T2, 4>, int, T2) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | other | 0.0087 | 512 | 432 | 269,060 | 23 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | copy | void at::<unnamed>::CatArrayBatchedCopy<at::<unnamed>::OpaqueType<2>, unsigned int, 4, 64, 64>(T1 *, at::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::<unnamed>::TensorSizeStride<T2, 4>, int, T2) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | other | 0.00858 | 512 | 432 | 269,060 | 23 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<c10::BFloat16>, std::array<char *, 3>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | residual | 0.00701 | 128 | 1,024 | 4,200,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | cast | void at::unrolled_elementwise_kernel<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, 2>, 4, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned in... | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | cast | 0.01254 | 128 | 2,048 | 2,130,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, void at::<unnamed>::pow_tensor_scalar_kernel_impl<float, float>(at::TensorIteratorBase &, T2)::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | rmsnorm | 0.00768 | 128 | 1,024 | 4,200,000 | 28 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | reduce | void at::reduce_kernel<512, 1, at::ReduceOp<float, at::MeanOps<float, float, float, float>, unsigned int, float, 4, 4>>(T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 262,144 | rmsnorm | 0.01062 | 512 | 128 | 4,250,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctorOnSelf_add<float>, std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 512 | residual | 0.00333 | 128 | 1 | 5,250 | 30 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::rsqrt_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 512 | rmsnorm | 0.00368 | 128 | 1 | 6,780 | 22 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | 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 | true | null | null | null | 1 | 128 | null | null | null | 2,097,152 | mul | 0.01024 | 128 | 4,096 | 4,210,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | cast | void at::vectorized_elementwise_kernel<4, at::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | cast | 0.00714 | 128 | 1,024 | 4,200,000 | 28 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | mul | 0.01104 | 128 | 2,048 | 2,130,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::<unnamed>::silu_kernel(at::TensorIteratorBase &)::[lambda() (instance 1)]::operator ()() const::[lambda() (instance 6)]::operator ()() const::[lambda(c10::BFloat16) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 2,097,152 | silu | 0.01366 | 128 | 3,584 | 7,350,000 | 28 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>, std::array<char *, 3>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 2,097,152 | mul | 0.01408 | 128 | 3,584 | 14,690,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<c10::BFloat16>, std::array<char *, 3>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | residual | 0.00698 | 128 | 1,024 | 4,200,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | cast | void at::unrolled_elementwise_kernel<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, 2>, 4, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned in... | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | cast | 0.01296 | 128 | 2,048 | 2,130,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, void at::<unnamed>::pow_tensor_scalar_kernel_impl<float, float>(at::TensorIteratorBase &, T2)::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | rmsnorm | 0.00755 | 128 | 1,024 | 4,200,000 | 28 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | reduce | void at::reduce_kernel<512, 1, at::ReduceOp<float, at::MeanOps<float, float, float, float>, unsigned int, float, 4, 4>>(T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 262,144 | rmsnorm | 0.01075 | 512 | 128 | 4,250,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctorOnSelf_add<float>, std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 512 | residual | 0.00349 | 128 | 1 | 5,250 | 30 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::rsqrt_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 512 | rmsnorm | 0.00355 | 128 | 1 | 6,780 | 22 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | 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 | true | null | null | null | 1 | 128 | null | null | null | 2,097,152 | mul | 0.0103 | 128 | 4,096 | 4,210,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | cast | void at::vectorized_elementwise_kernel<4, at::bfloat16_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | cast | 0.00701 | 128 | 1,024 | 4,200,000 | 28 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | mul | 0.01094 | 128 | 2,048 | 2,130,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | mul | 0.01162 | 128 | 2,048 | 2,150,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::neg_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 9)]::operator ()() const::[lambda(c10::BFloat16) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | neg | 0.00819 | 128 | 1,024 | 1,070,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | copy | void at::<unnamed>::CatArrayBatchedCopy<at::<unnamed>::OpaqueType<2>, unsigned int, 4, 64, 64>(T1 *, at::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::<unnamed>::TensorSizeStride<T2, 4>, int, T2) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | other | 0.01773 | 512 | 432 | 2,100,000 | 23 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | mul | 0.01107 | 128 | 2,048 | 2,150,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::CUDAFunctor_add<c10::BFloat16>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | residual | 0.01216 | 128 | 2,048 | 4,210,000 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 131,072 | mul | 0.00723 | 128 | 256 | 314,750 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::neg_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 9)]::operator ()() const::[lambda(c10::BFloat16) (instance 1)]>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 65,536 | neg | 0.0073 | 128 | 128 | 148,860 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | copy | void at::<unnamed>::CatArrayBatchedCopy<at::<unnamed>::OpaqueType<2>, unsigned int, 4, 64, 64>(T1 *, at::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::<unnamed>::TensorSizeStride<T2, 4>, int, T2) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | other | 0.0097 | 512 | 432 | 269,180 | 23 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::BinaryFunctor<c10::BFloat16, c10::BFloat16, c10::BFloat16, binary_internal::MulFunctor<float>>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 131,072 | mul | 0.00717 | 128 | 256 | 314,620 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::elementwise_kernel<128, 4, void at::gpu_kernel_impl_nocast<at::CUDAFunctor_add<c10::BFloat16>>(at::TensorIteratorBase &, const T1 &)::[lambda(int) (instance 1)]>(int, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 131,072 | residual | 0.00736 | 128 | 256 | 544,130 | 16 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | copy | void at::<unnamed>::CatArrayBatchedCopy<at::<unnamed>::OpaqueType<2>, unsigned int, 4, 64, 64>(T1 *, at::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::<unnamed>::TensorSizeStride<T2, 4>, int, T2) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | other | 0.00864 | 512 | 432 | 269,180 | 23 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | copy | void at::<unnamed>::CatArrayBatchedCopy<at::<unnamed>::OpaqueType<2>, unsigned int, 4, 64, 64>(T1 *, at::<unnamed>::CatArrInputTensorMetadata<T1, T2, T4, T5>, at::<unnamed>::TensorSizeStride<T2, 4>, int, T2) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | other | 0.00858 | 512 | 432 | 269,310 | 23 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctor_add<c10::BFloat16>, std::array<char *, 3>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | residual | 0.00704 | 128 | 1,024 | 4,200,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | cast | void at::unrolled_elementwise_kernel<at::direct_copy_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 3)]::operator ()() const::[lambda() (instance 7)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, 2>, 4, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned in... | bf16 | true | null | null | null | 1 | 128 | null | null | null | 1,048,576 | cast | 0.0128 | 128 | 2,048 | 2,130,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, void at::<unnamed>::pow_tensor_scalar_kernel_impl<float, float>(at::TensorIteratorBase &, T2)::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 524,288 | rmsnorm | 0.00762 | 128 | 1,024 | 4,200,000 | 28 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | reduce | void at::reduce_kernel<512, 1, at::ReduceOp<float, at::MeanOps<float, float, float, float>, unsigned int, float, 4, 4>>(T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 262,144 | rmsnorm | 0.01082 | 512 | 128 | 4,250,000 | 32 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::CUDAFunctorOnSelf_add<float>, std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 512 | residual | 0.00336 | 128 | 1 | 5,250 | 30 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | elementwise | void at::vectorized_elementwise_kernel<4, at::rsqrt_kernel_cuda(at::TensorIteratorBase &)::[lambda() (instance 2)]::operator ()() const::[lambda() (instance 2)]::operator ()() const::[lambda(float) (instance 1)], std::array<char *, 2>>(int, T2, T3) | bf16 | true | null | null | null | 1 | 128 | null | null | null | 512 | rmsnorm | 0.00358 | 128 | 1 | 6,910 | 22 |
Llama-3.1-70B-Instruct_prefill | A100 | Llama-70B | 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 | true | null | null | null | 1 | 128 | null | null | null | 2,097,152 | mul | 0.01021 | 128 | 4,096 | 4,210,000 | 16 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.