Issues

• [Feature]: Let short requests interrupt a long single-engine response

  Niko1221/Strata#2821 · Filter by author Zcg2021 opened on Oct 16, 2026

• [Feature Request]: add missing llama.cpp server endpoints (/tokenize, /completion, /embedding, /metrics)

  Niko1221/Strata#2820 · Filter by author jeremyoha450 opened on Oct 16, 2026

• [Feature]: images on the Intel (SYCL) port through setup, with the CPU image encoder as on AMD

  Niko1221/Strata#2816 · Filter by author hotschmoe opened on Oct 16, 2026

• dequant launch: an illegal instruction was encountered

  Niko1221/Strata#2813 · Filter by author XeonG opened on Oct 16, 2026
  1

• [Feature]: SYCL: a two-kernel hyper-connection read (norm + down + reduce in one launch) for Intel Arc, opt-in

  Niko1221/Strata#2811 · Filter by author hotschmoe opened on Oct 16, 2026

• [Bug]: STRATA_HC_Q8=1 does nothing on the SYCL (Intel) build: the Q8_0 copies are loaded but never read

  Niko1221/Strata#2809 · Filter by author hotschmoe opened on Oct 16, 2026

• [Feature]: SYCL: read prompts faster on Battlemage by attending over the union of neighbouring positions' selections

  Niko1221/Strata#2807 · Filter by author hotschmoe opened on Oct 16, 2026

• [Feature]: SYCL: decode QSA selection and attention in fewer, shorter launches (opt-in, same bits)

  Niko1221/Strata#2805 · Filter by author hotschmoe opened on Oct 16, 2026

• [Feature]: SYCL: faster small-batch decode GEMVs for Q4_K, Q5_K, Q6_K and IQ4_XS on the Arc Pro B70 (opt-in)

  Niko1221/Strata#2803 · Filter by author hotschmoe opened on Oct 16, 2026

• [Feature]: SYCL: cheaper IQ weight decodes in the grouped expert kernels (opt-in, same bits)

  Niko1221/Strata#2801 · Filter by author hotschmoe opened on Oct 16, 2026

• [Bug]: SYCL, 2x Arc Pro B70 layer split: --pipeline-windows 2 fails the first verify window (UR_RESULT_ERROR_UNINITIALIZED)

  Niko1221/Strata#2800 · Filter by author hotschmoe opened on Oct 16, 2026

• [strata] the engine stopped unexpectedly (exit code -15). Unrecoverable native verification failure: verify: timed out at layer 26; its GPU waits were released and the GPU finished (#267)

  Niko1221/Strata#2798 · Filter by author VinterSolen opened on Oct 16, 2026

• [Linux, RX 7900 XTX gfx1100, 0.1.44.1, ntfs3] --ple-io direct's PLE reads run at ~950 IOPS (3.7 MiB/s) while the same file does 70,406 IOPS from another process; boot probe says "fine", and the #1407 I/O allowance cannot fire for small random reads

  Niko1221/Strata#2797 · Filter by author frankwang1130 opened on Oct 16, 2026
  1

• [Bug]: A </think> the model quotes inside a code block in its reasoning still ends the reasoning, and the turn's tool call is lost (follow-up to #1526)

  Niko1221/Strata#2794 · Filter by author KarlGrier opened on Oct 16, 2026

• tools/strata_bench.py: UnicodeDecodeError reading CMakeLists.txt on non-UTF-8 Windows (GBK/cp932), before any measurement

  Niko1221/Strata#2792 · Filter by author Lithivm opened on Oct 16, 2026

• Benchmark 48GB RAM + 2xP100 IQ3_XSS

  Niko1221/Strata#2789 · Filter by author Penguinehis opened on Oct 16, 2026
  2

• [Feature request] 16-bit KV cache option above 8K context (to A/B against int8)

  Niko1221/Strata#2787 · Filter by author mbaaku opened on Oct 16, 2026
  2

• 长上下文的情况下智能体完成任务后缓存异常清空

  Niko1221/Strata#2785 · Filter by author 1459945216 opened on Oct 16, 2026

• [Feature] Third-party qwen4exp GGUFs via tools/iq_pack.py — measured report (OrcaRouter Uncensored IQ3_XXS, YaRN to 1M)

  Niko1221/Strata#2783 · Filter by author emiltsoi opened on Oct 16, 2026

• sm_120 (RTX PRO 6000 Blackwell): mmvq_il_parity --emit-table over 7 runs, decode +1.3% in the engine; the table would also apply to the RTX 5070

  Niko1221/Strata#2782 · Filter by author brocktice opened on Oct 16, 2026

• [BUG] Engine crashes with 0xC0000005 in hipModuleLoad on first request — AMD RX 9070 XT (gfx1201) on Windows, Strata 0.1.44.1

  Niko1221/Strata#2781 · Filter by author jx839842046 opened on Oct 16, 2026
  3

• > # Community benchmark on NVIDIA GeForce RTX 5070 Ti

  Niko1221/Strata#2780 · Filter by author nghia1930mmm-cmd opened on Oct 16, 2026

• Batched prompt read stalls (#1118 family) + prefill copy_i32 illegal memory access, engine 0.1.44

  Niko1221/Strata#2778 · Filter by author zy15617 opened on Oct 16, 2026

• Community benchmark: RTX 5070 Ti, i5-13600KF, 64 GB RAM, IQ3_S @ 262K context

  Niko1221/Strata#2772 · Filter by author wx100fen opened on Oct 16, 2026
  1

• 2080 Ti 22G+ 3070 16G Ok ?

  Niko1221/Strata#2770 · Filter by author chensisi0730 opened on Oct 16, 2026
  1