-
Notifications
You must be signed in to change notification settings - Fork 1
All issues
Issue creation is restricted in this repository
Issues
is:issue state:open
is:issue state:open
Search results
oggopus: clamp the end-trim keep in int64 before converting to int (32-bit hardening)
area: decoderOpus/CELT/SILK decoderOpus/CELT/SILK decoderarea: oggopusOgg Opus container (RFC 7845)Ogg Opus container (RFC 7845)Status: Open.#89 In tphakala/go-opus;quality: CI regression gate comparing the encoder against ffmpeg (libopus) with objective metrics
enhancementNew feature or requestNew feature or requestStatus: Open.#87 In tphakala/go-opus;perf: close the mono-encode Go/C gap (profile + remaining levers)
area: celtinternal/celtinternal/celtarea: encoderOpus/CELT encoderOpus/CELT encoderperformanceSpeed, allocations, or SIMD. Never at the cost of bit-exactness.Speed, allocations, or SIMD. Never at the cost of bit-exactness.Status: Open.#86 In tphakala/go-opus;perf(opusenc): SIMD compute_frame_energy sum-of-squares (shares the tphakala/simd#267 blocker with #77)
area: encoderOpus/CELT encoderOpus/CELT encoderperformanceSpeed, allocations, or SIMD. Never at the cost of bit-exactness.Speed, allocations, or SIMD. Never at the cost of bit-exactness.Status: Open.#85 In tphakala/go-opus;oggopus: decide handling for channel mapping families 2 and 3 (ambisonics)
enhancementNew feature or requestNew feature or requestStatus: Open.#82 In tphakala/go-opus;celt: SIMD compute_band_energies sum-of-squares (blocked on tphakala/simd#267)
area: celtinternal/celtinternal/celtdeferredTracked and intentionally not being done nowTracked and intentionally not being done nowperformanceSpeed, allocations, or SIMD. Never at the cost of bit-exactness.Speed, allocations, or SIMD. Never at the cost of bit-exactness.Status: Open.#77 In tphakala/go-opus;- Status: Open.#69 In tphakala/go-opus;
celt: SIMD expRotation1 (strided Q15 Givens rotation chains)
area: celtinternal/celtinternal/celtdeferredTracked and intentionally not being done nowTracked and intentionally not being done nowperformanceSpeed, allocations, or SIMD. Never at the cost of bit-exactness.Speed, allocations, or SIMD. Never at the cost of bit-exactness.Status: Open.#44 In tphakala/go-opus;SIMD: fused PVQ pyramid search (op_pvq_search inner argmax), the top encode kernel
area: celtinternal/celtinternal/celtdeferredTracked and intentionally not being done nowTracked and intentionally not being done nowperformanceSpeed, allocations, or SIMD. Never at the cost of bit-exactness.Speed, allocations, or SIMD. Never at the cost of bit-exactness.Status: Open.#30 In tphakala/go-opus;SIMD part 2: the fused Opus-specific kernels (PVQ search, exp_rotation, butterflies, comb filter)
area: celtinternal/celtinternal/celtdeferredTracked and intentionally not being done nowTracked and intentionally not being done nowperformanceSpeed, allocations, or SIMD. Never at the cost of bit-exactness.Speed, allocations, or SIMD. Never at the cost of bit-exactness.Status: Open.#13 In tphakala/go-opus;Bounds-check elimination and inlining in the scalar hot loops (11% ceiling, measured)
area: celtinternal/celtinternal/celtperformanceSpeed, allocations, or SIMD. Never at the cost of bit-exactness.Speed, allocations, or SIMD. Never at the cost of bit-exactness.Status: Open.#6 In tphakala/go-opus;Performance profile: the "no-SIMD" C oracle is not scalar — clang auto-vectorizes it, and that is 72% of the 2x
performanceSpeed, allocations, or SIMD. Never at the cost of bit-exactness.Speed, allocations, or SIMD. Never at the cost of bit-exactness.Status: Open.#1 In tphakala/go-opus;