Chore debug #14

2026-07-06T09:46:41-05:00

iamdoubz commented

2026-07-06 09:46:41 -05:00

No description provided.

iamdoubz added 28 commits 2026-07-06 09:46:41 -05:00

feat(hardware): AccelPath resolver + honest GPU availability (P0) e9567b823a

whisper.cpp bakes in one GPU backend at compile time, so a BackendId alone
can't say how a vendor is served. Add AccelPath + resolve_accel(_with) as the
single source of truth over (compiled features x runtime readiness), and derive
BackendInfo.available from it in the DXGI + NPU paths. Closes the gap where a
present-but-unsupported GPU was marked available and best() routed to a no-op
GPU. Pure resolve_accel_with is unit-tested across all vendor/feature combos.

feat(transcription): drive engine choice from AccelPath resolver (P0) 7fe0d87e55

load_transcriber now resolves the accel path instead of assuming a GPU
BackendId has a working offload. Only requests whisper GPU offload for
WhisperCuda/WhisperVulkan; NPU goes to the ONNX engine; everything else decodes
on CPU and reports the backend actually used.

build(deps): enable ort DirectML EP for the non-Vulkan GPU path (P2) c4658efe28

feat(paths): DirectML runtime dir/dll/ready helpers (P2) 2e8bb730a3

feat(transcription): generalize OnnxNpuTranscriber -> OnnxTranscriber with DirectML EP (P2) 0f93f5947e

load() now selects OpenVINO (NPU) or DirectML (AMD/Intel GPU) by BackendId over
the same ONNX artifacts; ensure_runtime_env takes the target runtime DLL. Adds
an opt-in directml_transcribes_speech spike for on-GPU validation.

refactor(transcription): re-export OnnxTranscriber; note ONNX tier covers NPU+DirectML (P2) eb87bf4edb

docs(transcription): update OnnxTranscriber name in doc comments (P2) 9ae02cabfe

feat(hardware): make directml_ready() real (npu feature + staged runtime) (P2) 0f2940e52b

feat(commands): route AMD/Intel to DirectML ONNX engine + download_directml_package (P2) b69c91a598

load_transcriber dispatches OnnxOpenVino/OnnxDirectML through one OnnxTranscriber
block. download_and_extract_runtime is parameterized (sha + readiness) so the new
stage_directml_runtime reuses it; hardware_status reports a directml package field.

feat(commands): register download_directml_package command (P2) ab5aa4e986

feat(hardware): directml_would_help() gate for the Settings package card (P2) 281ccf0b18

True only when staging DirectML would light up a present GPU this build can't
otherwise accelerate: npu compiled, no Vulkan baked in, and an AMD/Intel GPU
present (or NVIDIA without CUDA).

feat(commands): expose directml.applicable in hardware_status (P2) ceb7e59164

feat(api): DirectML package status + downloadDirectmlPackage binding (P2) 1b743dad67

feat(settings): DirectML GPU acceleration package card (P2) 1b74e0b320

Mirrors the NPU package card, shown only when hardware_status marks DirectML
applicable; downloads model+runtime via download_directml_package and reloads
hardware to flip to Ready.

build: bake short git commit hash into WA_GIT_HASH for the About page 035abb93ea

feat(commands): app_info (version+commit) and open_url for the About page 0317649b48

feat(commands): register app_info + open_url 42c2aedfd4

feat(api): AppInfo type + appInfo/openUrl bindings a51d7a777b

fix(settings): wrap nav tabs so the close button isn't cut off; add About page 31b3cfd5a7

The fixed-width settings panel overflowed horizontally because the tab nav
didn't wrap, pushing the X close button off the edge and adding a horizontal
scrollbar. nav now flex-wraps. Adds an About section: version, build commit,
and a link to the source repo (opened via open_url).

chore(release): bump version to 0.1.5 0d62416127

fix(theme): dark-mode native controls (Templates dropdown) 4442f809d1

Set color-scheme on the theme roots so native controls render in the active
theme, and give the header template <select> an explicit elevated background +
themed <option> colors — an author-styled select's popup otherwise defaults to
white in dark mode.

fix(theme): style textarea like inputs so the AI system prompt matches dark mode 8038918e1d

build(deps): swap zip for sevenz-rust2 (npu-gated) to unpack 7z runtime bundles 27ba4f9b03

The NPU/DirectML runtime bundles are now LZMA2+BCJ 7z archives; zip was only used
for the old .zip bundle. sevenz-rust2 pinned to 0.7.0 (newer needs rustc>MSRV).

feat(runtime): download NPU/DirectML runtimes as 7z from the repo raw URLs e6f19168f8

extract_7z_flat replaces extract_zip_flat; NPU + DirectML now default to
git.dou.bet raw/branch/main/runtime/{openvino,directml}.7z with pinned SHA-256s
(DirectML no longer requires WA_DIRECTML_RUNTIME_URL). Adds an extraction test
against the real bundle (proves LZMA2+BCJ decode + flatten).

docs(readme): add Quick start (install from MSI) 0ffe608e03

docs(readme): bump version references to 0.1.5 d34409a6ed

chore(runtime): add NPU/DirectML 7z bundles via Git LFS fa1c5ccc44

fix(runtime): use gitea /media/ URLs so Git LFS bundles resolve (not the pointer) 224aaf9994

iamdoubz merged commit 724ce4ba3b into main

2026-07-06 09:46:49 -05:00

iamdoubz deleted branch chore_debug

2026-07-06 09:46:49 -05:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: iamdoubz/WhispAssist#14