-
Notifications
You must be signed in to change notification settings - Fork 112
All issues
Issue creation is restricted in this repository
Issues
is:issue state:open
is:issue state:open
Search results
[Bug]: qvac-fabric MoE expert cache asserts on Vulkan for caches > ~2.5 GB (bank tensor exceeds 1 GiB suballocation block) — --fit auto-sizes it to 10% of experts, so defaults crash on large MoE models
bugSomething isn't workingSomething isn't workingNLPllm and embedllm and embedStatus: Open.#4435 In tetherto/qvac;Weekend Mobile Tests (LLM) — 2026-09-13
weekly-mobile-reportAutomated weekend mobile test reportsAutomated weekend mobile test reportsStatus: Open.#4434 In tetherto/qvac;[Bug]: SDK config rejects
threads, so the LLM addon is stuck with fabric's default thread count (4 of 8 P-cores on Arrow Lake, 15 vs 22 tok/s)bugSomething isn't workingSomething isn't workingStatus: Open.#4433 In tetherto/qvac;[Bug]: llm-llamacpp never runs qvac-fabric --fit (sharded path skips it, single-file path aborts); >VRAM MoE model loads 78 GB onto a 24 GB GPU and runs at 2.5 tok/s
bugSomething isn't workingSomething isn't workingNLPllm and embedllm and embedStatus: Open.#4432 In tetherto/qvac;[Feature]: LLM modelConfig strict schema rejects threads / override-tensor / n-cpu-moe / fit-target / batch-size that the addon already supports (4.7 vs 22 tok/s on a 103 GB MoE)
enhancementNew feature or requestNew feature or requestStatus: Open.#4431 In tetherto/qvac;- Status: Open.#4394 In tetherto/qvac;
- Status: Open.#4337 In tetherto/qvac;
- Status: Open.#4292 In tetherto/qvac;
- Status: Open.#4290 In tetherto/qvac;
[Bug]: qvac-fabric-llm.cpp no CPU kernel in this release
bugSomething isn't workingSomething isn't workingStatus: Open.#4273 In tetherto/qvac;- Status: Open.#4151 In tetherto/qvac;
- Status: Open.#4131 In tetherto/qvac;