§ )`ƒiê ã óŒ—dZddlZddlZddlmZejd„¦«Zedejdejdejd dfd „¦«ZdS) a3 Copyright (c) 2025 by FlashInfer team. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. éNé)Úflashinfer_apicóF—ddlm}|¦« ¦«S)Nr©Úgen_concat_mla_module)Újit.dsv3_optimizationsrÚbuild_and_loadrs úi/home/jaya/work/projects/VOICE-AGENT/VIET/agent-env/lib/python3.11/site-packages/flashinfer/concat_ops.pyÚget_concat_mla_moduler s/€à=Ð=Ð=Ð=Ð=Ð=à Ð Ñ "Ô "× 1Ò 1Ñ 3Ô 3Ð3óÚkÚk_nopeÚk_ropeÚreturncóL—t¦« |||¦«dS)aConcatenate k_nope and k_rope tensors for MLA attention. This function efficiently concatenates: - k_nope: per-head nope values - k_rope: shared rope values (broadcast to all heads) Key optimizations: - Warp-based processing with software pipelining - Vectorized memory access (int2 for nope, int for rope) - L2 prefetching for next row while processing current - Register reuse for rope values across all heads in a chunk Parameters ---------- k : torch.Tensor Output tensor, shape: ``[num_tokens, num_heads, nope_dim + rope_dim]``. Modified in-place. k_nope : torch.Tensor The nope part of k, shape: ``[num_tokens, num_heads, nope_dim]``. k_rope : torch.Tensor The rope part of k (shared), shape: ``[num_tokens, 1, rope_dim]``. This is broadcast to all heads. Example ------- >>> import torch >>> import flashinfer >>> num_tokens = 2048 >>> num_heads = 128 >>> nope_dim = 128 >>> rope_dim = 64 >>> k = torch.empty(num_tokens, num_heads, nope_dim + rope_dim, dtype=torch.bfloat16, device="cuda") >>> k_nope = torch.randn(num_tokens, num_heads, nope_dim, dtype=torch.bfloat16, device="cuda") >>> k_rope = torch.randn(num_tokens, 1, rope_dim, dtype=torch.bfloat16, device="cuda") >>> flashinfer.concat_ops.concat_mla_k(k, k_nope, k_rope) Note ---- This kernel is specifically optimized for: - ``num_heads = 128`` - ``nope_dim = 128`` - ``rope_dim = 64`` N)r Ú concat_mla_k)r rrs r rrs)€õfÑÔ×(Ò(¨¨F°FÑ;Ô;Ð;Ð;Ð;r ) Ú__doc__Ú functoolsÚtorchÚ api_loggingrÚcacher ÚTensorr©r r úrs©ðððð ÐÐÐà € € € à'Ð'Ð'Ð'Ð'Ð'ð „ð4ð4ñ„ð4ð ð2<Ø „|ð2<à ŒLð2<ð ŒLð2<ð ð 2<ð2<ð2<ñ„ð2<ð2<ð2