Issues

• [DEV] Support indexed Mamba-2 inference on NVIDIA and MetaX

  InfiniTensor/InfiniCore#1571 · Filter by author big-hip opened on Sep 20, 2026

• [DEV] 为 InfiniLM-v0.2.9c 增加可复用的 GGUF 线性层与 BlockFP8 支持

  InfiniTensor/InfiniCore#1569 · Filter by author xindongliu594 opened on Sep 20, 2026

• [DEV] Complete NVIDIA runtime support for Qwen greedy MTP

  InfiniTensor/InfiniCore#1565 · Filter by author big-hip opened on Sep 19, 2026

• [BUG] ContextImpl 默认在 0 号卡建 Runtime，进程使用其他卡时会多占一张卡

  InfiniTensor/InfiniCore#1553 · Filter by author JoeZhang-0x000 opened on Sep 7, 2026

• [DEV] enable InfiniOps dispatch for ALI devices

  InfiniTensor/InfiniCore#1541 · Filter by author spike-zhu opened on Aug 27, 2026

• [BUG] retain torch_python for ATen builds with moore gpu

  InfiniTensor/InfiniCore#1539 · Filter by author spike-zhu opened on Aug 27, 2026

• [DEV] Cross-device copy_from

  InfiniTensor/InfiniCore#1529 · Filter by author PanZezhong1725 opened on Aug 25, 2026

• [DEV] upsample supports strides

  InfiniTensor/InfiniCore#1528 · Filter by author PanZezhong1725 opened on Aug 25, 2026

• [DEV] timestep embedding

  InfiniTensor/InfiniCore#1527 · Filter by author PanZezhong1725 opened on Aug 25, 2026

• [DEV] MiniMax-H3 算子、运行时支持

  InfiniTensor/InfiniCore#1526 · Filter by author PanZezhong1725 opened on Aug 25, 2026

• [BUG] Ascend C compilation fails because paged attention uses reserved block_idx

  InfiniTensor/InfiniCore#1519 · Filter by author ETOwang opened on Aug 22, 2026

• [DEV] add MUSA include path for InfiniOps adapters for moore gpu

  InfiniTensor/InfiniCore#1516 · Filter by author spike-zhu opened on Aug 21, 2026

• [BUG] 昇腾平台图模式开启输出异常

  InfiniTensor/InfiniCore#1460 · Filter by author Jingbo-gao opened on Aug 4, 2026

• [DEV] 模型适配：Qwen3.6-35B-A3B 推理所需优化修改

  InfiniTensor/InfiniCore#1450 · Filter by author qinyiqun opened on Jul 31, 2026

• [DEV] InfiniCCL添加InitRank，Send/Recv，Broadcast接口

  InfiniTensor/InfiniCore#1403 · Filter by author PanZezhong1725 opened on Jul 21, 2026

• [DEV] 为 mul_scalar / moe_sum / prepare_moe_input / rwkv5_wkv / mamba_selective_scan / dequantize_gptq 补齐 Metax 后端

  InfiniTensor/InfiniCore#1395 · Filter by author Lfan-ke opened on Jul 13, 2026
  1

• [DEV] nsa_paged_attention / nsa_compress_paged_cache 的 Metax 后端支持

  InfiniTensor/InfiniCore#1393 · Filter by author Lfan-ke opened on Jul 13, 2026
  1

• [DEV] 为 InfiniRT 添加 moore、metax 等 bridge 基建

  InfiniTensor/InfiniCore#1386 · Filter by author spike-zhu opened on Jul 10, 2026

• [DEV] Qwen3.5模型算子国产平台迁移

  InfiniTensor/InfiniCore#1375 · Filter by author PanZezhong1725 opened on Jul 8, 2026

• [DEV] adjust InfiniCore cuda_like runtime to InfiniRT

  InfiniTensor/InfiniCore#1367 · Filter by author spike-zhu opened on Jul 5, 2026

• [DEV] Support Marlin AWQ/GPTQ repack and CUDA graph-safe Add on NVIDIA

  InfiniTensor/InfiniCore#1336 · Filter by author qinyiqun opened on Jun 25, 2026

• [DEV] NVIDIA添加moe_wna16_marlin_gemm算子

  InfiniTensor/InfiniCore#1319 · Filter by author xgqdut2016 opened on Jun 18, 2026

• [DEV] adjust InfiniCore cpu runtime to InfiniRT

  InfiniTensor/InfiniCore#1311 · Filter by author spike-zhu opened on Jun 18, 2026

• [DEV] Add MoE inference operators and EP communication primitives

  InfiniTensor/InfiniCore#1308 · Filter by author qinyiqun opened on Jun 18, 2026

• [DEV] moore mate xmake adjust

  InfiniTensor/InfiniCore#1296 · Filter by author spike-zhu opened on Jun 16, 2026