docs: fix readme consistency#9
Merged
Merged
Conversation
- Unify RTX 3090 price to ~$900 across megakernel README (was $700 in comparison table and $900-1,000 in 'Why 3090' section) - Replace 'Ampere+, sm_86+' with 'RTX 3090 (sm_86) only' in hub and megakernel READMEs (sm_80 is Ampere, not sm_86) - Roadmap: add 'CLI for local AI' to Q2 2026 and 'Lucebox official launch' to Q3 2026 - Standardize all lucebox.com links to https://www.lucebox.com
Flip previous commit's www direction. Canonical URL is https://lucebox.com, so dflash README's www blog link is now rewritten to match the rest.
alessandropuppo
changed the title
alessandropuppo
changed the title
…admap item - Requirements now read 'Built and benchmarked on RTX 3090; portable to other Ampere+ (sm_86+) NVIDIA GPUs with minor tuning' instead of '3090 only'. Kernel has hardcoded 3090 constants (NUM_BLOCKS=82) but nothing about the approach is 3090-exclusive. - Roadmap: 'CLI for local AI' -> 'Lucebox OS for local AI machines' to match the Q3 'Lucebox official launch' positioning.
This was referenced May 13, 2026
davide221
pushed a commit
to dusterbloom/lucebox-hub
that referenced
this pull request
May 13, 2026
…e-dflash post-PR#9 PR Luce-Org#9 (TQ3 sparse-FA + FWHT contract + May 2026 CUDA opts) merged into Luce-Org/llama.cpp-dflash-ggml@luce-dflash at 567d42e. The previous submodule pin (ecb832b on dusterbloom's personal fork) is now obsolete: it diverges from luce-dflash by 18 commits behind / 317 ahead, so it would silently regress trunk work to gain the PR Luce-Org#9 feature set. Two changes: - .gitmodules: url -> Luce-Org/llama.cpp-dflash-ggml.git, branch -> luce-dflash - submodule pin: ecb832b -> 567d42e (clean fast-forward from current main pin c79573c) 567d42e is the merge commit of PR Luce-Org#9 into luce-dflash and is the single SHA that combines (a) PR Luce-Org#9's TQ3 work and (b) all trunk commits up to the merge. Validated end-to-end on RTX 3090: PR Luce-Org#9 build clean, byte-comparable to luce-dflash HEAD baseline, +11.2 % mean decode tok/s on Qwen3.6-27B Q4_K_M + TQ3 KV + SWA=2048 + budget=22 (51.10 vs 45.94 tok/s, mean AL 6.87 vs 6.35).
oliveagle
pushed a commit
to oliveagle/lucebox-hub
that referenced
this pull request
May 22, 2026
docs: fix readme consistency
oliveagle
pushed a commit
to oliveagle/lucebox-hub
that referenced
this pull request
May 22, 2026
…e-dflash post-PR#9 PR Luce-Org#9 (TQ3 sparse-FA + FWHT contract + May 2026 CUDA opts) merged into Luce-Org/llama.cpp-dflash-ggml@luce-dflash at 567d42e. The previous submodule pin (ecb832b on dusterbloom's personal fork) is now obsolete: it diverges from luce-dflash by 18 commits behind / 317 ahead, so it would silently regress trunk work to gain the PR Luce-Org#9 feature set. Two changes: - .gitmodules: url -> Luce-Org/llama.cpp-dflash-ggml.git, branch -> luce-dflash - submodule pin: ecb832b -> 567d42e (clean fast-forward from current main pin c79573c) 567d42e is the merge commit of PR Luce-Org#9 into luce-dflash and is the single SHA that combines (a) PR Luce-Org#9's TQ3 work and (b) all trunk commits up to the merge. Validated end-to-end on RTX 3090: PR Luce-Org#9 build clean, byte-comparable to luce-dflash HEAD baseline, +11.2 % mean decode tok/s on Qwen3.6-27B Q4_K_M + TQ3 KV + SWA=2048 + budget=22 (51.10 vs 45.94 tok/s, mean AL 6.87 vs 6.35).
easel
added a commit
to easel/lucebox-hub
that referenced
this pull request
May 29, 2026
…appers
- .github/workflows/{ci,docker,release-luce-bench}.yml: pin
actions/checkout, docker/{setup-buildx,login,metadata,bake}-action,
and astral-sh/setup-uv to immutable commit SHAs with `# vN` comments
so the supply chain is reproducible (#4).
- harness/src/harness/clients/_common.py: replace the external `timeout`
shell-out with `subprocess.run(..., timeout=N)`, return 124 on
TimeoutExpired to match GNU timeout's exit code (#5).
- scripts/build_image.sh: normalize REGISTRY to end in `/` instead of
silently producing `ghcr.io/luce-orglucebox-hub` when the trailing
slash is missing (#6).
- harness/src/harness/clients/pi.py: non-interactive launch now mirrors
run_pi.sh's validated invocation (--provider, --print, --mode json,
--tools, --no-session, --offline) and sets PI_CODING_AGENT_DIR /
PI_CODING_AGENT_SESSION_DIR / PI_OFFLINE (Luce-Org#7).
- docker-bake.hcl: sanitize `+` → `-` in VERSION before composing tags,
since `+` is not a valid Docker tag character (Luce-Org#8).
- harness/src/harness/clients/hermes.py: set HERMES_HOME + the rest of
run_hermes.sh's env wiring and call `chat --provider --model
--accept-hooks --yolo --max-turns --source --query` instead of a bare
positional prompt (Luce-Org#9, Luce-Org#10).
- harness/src/harness/clients/openclaw.py: apply the OpenClaw config
patch via `openclaw config patch --file` before the run, and call
`agent --local --json --model lucebox/<model> --session-id --timeout
--message` instead of a bare positional prompt (Luce-Org#11).
- pyproject.toml: drop the dead dflash/scripts/{prefix_cache,test_server,
tool_memory}.py ruff include pins (those paths were renamed during
the dflash→server rename and then deleted upstream) (Luce-Org#12).
- lefthook.yml: widen the shellcheck/bash-parse glob from `*.sh` to
`**/*.sh` so scripts under nested dirs (harness/clients/*.sh,
scripts/*.sh, server/scripts/*.sh) are linted on commit (Luce-Org#13).
Addresses cubic #4–Luce-Org#13 (P2) on PR Luce-Org#285. Luce-Org#14 was already addressed in
the previous commit alongside the LUCEBENCH_THINK default fix.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Small README polish pass, four fixes:
~$900.megakernel/README.mdcontradicted itself:~$700in the 3090 vs M5 Max table,~$900-1,000in the "Why an RTX 3090?" section. Now both say~$900.Ampere+, sm_86+, but sm_80 is Ampere (A100), and the megakernel has hardcoded 3090 constants. Reworded toBuilt and benchmarked on NVIDIA RTX 3090 (2020); portable to other Ampere+ (sm_86+) NVIDIA GPUs with minor tuning. Reflects reality: 3090 is the tested target, other Ampere+ cards need small tuning. (dflash/README.mduntouched, already lists supported cards.)Lucebox OS for local AI machinesto Q2 2026 andLucebox official launchto Q3 2026.https://lucebox.com(apex, no www). One blog link usedwww, the rest didn't. All canonical lucebox.com links now use the apex domain.No code changes. Docs-only.
Test plan
https://lucebox.com/...megakernel/README.md, confirm price shows$900everywhere and requirements read 'portable with minor tuning'dflash/README.md, confirm footer link goes tolucebox.com