Issues

• TensorFold spill v2 builds: cached prompt tokens are never read (tensorfold:prompt_cache_cached_tokens), so Cost compare prices every prompt token as computed

  MiaAI-Lab/sparkDash#281 · Filter by author lastingchild opened on Oct 26, 2026
  1

• Adding windows standalone machines results in SSH error

  MiaAI-Lab/sparkDash#280 · Filter by author mapfde opened on Oct 25, 2026
  2

• Overview hides GPU-busy evidence on launcher discovery failure and mislabels an unready head as no model

  MiaAI-Lab/sparkDash#279 · Filter by author smkzw opened on Oct 25, 2026

• Token totals silently stop when a two-node engine moves its API to the node configured as worker

  MiaAI-Lab/sparkDash#271 · Filter by author eleata opened on Oct 23, 2026
  1

• RoCE panel: RDMA traffic during TP2 decode is not visible (netdev counters)

  MiaAI-Lab/sparkDash#270 · Filter by author Nipale-ai opened on Oct 22, 2026
  2

• [Feature Request] Make the GPU temperature warning and critical thresholds configurable in Settings

  MiaAI-Lab/sparkDash#233 · Filter by author jammsen opened on Oct 17, 2026
  6

• [Feature Request] Serve under a URL sub-path behind a reverse proxy (SPARKDASH_BASE_PATH)

  MiaAI-Lab/sparkDash#192 · Filter by author naeemakhtar110 opened on Oct 11, 2026
  7

• [Feature Request] Add versioning in the display IE tensor fold version, this repo version at compose time, etc.

  MiaAI-Lab/sparkDash#173 · Filter by author bangell opened on Oct 9, 2026
  9