Issues

• [BUG] SM120 dense HSTU smem overflow

  bug
  NVIDIA/recsys-examples#492 · Filter by author Clebrate opened on Sep 21, 2026

• Split key_value_table.py into multiple files to improve maintainability.

  NVIDIA/recsys-examples#483 · Filter by author jiashuy opened on Sep 10, 2026

• [PERF][RFC] Explore a shared-memory AOTI + FlexKV integration for Triton

  NVIDIA/recsys-examples#475 · Filter by author shijieliu opened on Sep 1, 2026

• [FEA] Add KV-cache-aware Triton orchestration and move offload/reap off the inference critical path`

  NVIDIA/recsys-examples#474 · Filter by author shijieliu opened on Sep 1, 2026

• [FEA] Add CUDA Graph replay to the HSTU KV-cache AOTI compute path`

  NVIDIA/recsys-examples#473 · Filter by author shijieliu opened on Sep 1, 2026

• [BUG] FlexKV SSD-to-host transfer bandwidth is lower than expected

  bug
  NVIDIA/recsys-examples#472 · Filter by author geoffreyQiu opened on Sep 1, 2026
  2

  

• [FEA] Overlap SSD-to-host and host-to-device KV-cache transfers in FlexKV

  NVIDIA/recsys-examples#471 · Filter by author geoffreyQiu opened on Sep 1, 2026

• [FEA] Support configurable table-fusion indexing and NVE layer types for exportable inference embeddings

  NVIDIA/recsys-examples#470 · Filter by author geoffreyQiu opened on Sep 1, 2026

• Development Roadmap(July/Aug/Sep)

  NVIDIA/recsys-examples#462 · Filter by author shijieliu opened on Aug 10, 2026

• [FEA] Optimize online request scheduling for HSTU inference

  NVIDIA/recsys-examples#461 · Filter by author shijieliu opened on Aug 10, 2026

• [FEA] Add an online serving benchmark for HSTU inference

  NVIDIA/recsys-examples#460 · Filter by author shijieliu opened on Aug 10, 2026

• [FEA] Resolve key-to-index mapping in DynamicEmb row-wise input_dist

  NVIDIA/recsys-examples#459 · Filter by author shijieliu opened on Aug 10, 2026

• [FEA] Add FP8 KV-cache support for HSTU inference

  feature
  NVIDIA/recsys-examples#457 · Filter by author shijieliu opened on Aug 10, 2026

• [FEA] Support transformer backbone in HSTU inference

  feature
  NVIDIA/recsys-examples#456 · Filter by author shijieliu opened on Aug 10, 2026

• [FEA] Support target-aware masks in FA

  feature
  NVIDIA/recsys-examples#455 · Filter by author shijieliu opened on Aug 10, 2026

• [FEA][Dynamicemb] Support DRAM as an internal storage tier and PS as an external Parameter Store tier

  NVIDIA/recsys-examples#453 · Filter by author shijieliu opened on Aug 10, 2026

• [FEA] Add Rubin support for HSTU training

  NVIDIA/recsys-examples#452 · Filter by author shijieliu opened on Aug 10, 2026

• [QST] Timeline for "native page tables in decode attention" for SID-GR — variable-length online throughput gap

  question
  NVIDIA/recsys-examples#445 · Filter by author randomseed713 opened on Jul 31, 2026
  1

• [QST] Is this level of run-to-run variance expected when training HSTU retrieval on ML-20M?

  question
  NVIDIA/recsys-examples#427 · Filter by author ngocbh opened on Jun 21, 2026
  1

• [FEA] Recsys-FlexKV: support SSD and Layerwise

  NVIDIA/recsys-examples#422 · Filter by author Clebrate opened on Jun 16, 2026

• [FEA] online training example

  NVIDIA/recsys-examples#420 · Filter by author shijieliu opened on Jun 16, 2026

• [QST] nsys profile fails to generate profiling file on L20 with version 2603

  question
  NVIDIA/recsys-examples#397 · Filter by author theflyfish opened on May 19, 2026
  1

  

• [QST] Control over gradient communication (AllReduce) and optimizer steps for decoupled DynamicEmb

  questionTBD
  NVIDIA/recsys-examples#393 · Filter by author jlinswenmogu opened on May 15, 2026
  3

  

• [FEA] fused hier output dist

  NVIDIA/recsys-examples#384 · Filter by author shijieliu opened on May 11, 2026
  1

• [FEA] remove unnecessary keys selection and fuse selection to keys and scores

  NVIDIA/recsys-examples#357 · Filter by author jiashuy opened on Apr 13, 2026