Extremely slow x265 software rendering with low CPU usage since v0.8.0

[xJayMorex]

Bug Information

Bug Description

When using version v0.8.0 or v0.8.1 for software encoding, CPU usage tends to be around 40% and rendering power around 35% compared to v0.7.3 (last version before huge rehaul).

Bug Steps

1. Try software encoding with v0.8.0 or v0.8.1 and compare it to v0.7.3 performance

Additional Information

Tried bind volume mounting the latest Debian build of handbrake-cli into v0.8.1 along with all required shlibs (119 to be exact, fun times) and the performance was exactly the same, so it's not the custom built binary at fault here, but the repoless base itself.

Application Information

HandBrake Web Version

• v0.8.0+

Docker Compose

services:
  handbrake-server:
    image: ghcr.io/thenickoftime/handbrake-web-server:latest
    container_name: handbrake-web-server
    user: 1000:1000 # edit to run as user (uuid:guid) with permissions to access your media. 0:0 to run as root (not recommended).
    ports:
      - 9999:9999
    volumes:
      - /data/handbrake/data:/data
      - /data/handbrake/video:/video # ensure this path is the same across all containers

  handbrake-worker:
    image: ghcr.io/thenickoftime/handbrake-web-worker:latest
    container_name: handbrake-web-worker
    user: 1000:1000 # edit to run as user (uuid:guid) with permissions to access your media. 0:0 to run as root (not recommended).
    environment:
      - WORKER_ID=handbrake-worker # give your worker a unique name
      - SERVER_URL=handbrake-server # change if setting up a standalone worker, prefix with http(s):// if necessary
      - SERVER_PORT=9999 # change if using a reverse proxy or the port is otherwise different than above
    volumes:
      - /data/handbrake/video:/video # ensure this path is the same across all containers
    depends_on:
      - handbrake-server

Host Machine

• OS: Devuan 6 VM running on Proxmox VE 9.1.2
• CPU: Intel Xeon E5-2658A V3
• GPU: N/A

Log Information

4.41fps (v0.8.1) vs 11.75fps (v0.7.3) - using 4 cores

[barndoor101]

Ive got this as well, just spun up a new instance, chucked 24 cores at it, and only seeing 10% cpu usage when encoding.

[barndoor101]

Just did some testing, spun up some worker only nodes, and it seems like they will only use 3-4 cores max.

[xsidexpt]

I am on the same boat here, sub 40% usage on a 8 core 9800x3d on version 0.8.1 (latest as of this comment)

[xsidexpt]

Updating my previous reply as I have found a way around this slow speed.
On my particular deployment I have the workers separate from the coordinator and both workers are windows machines running the containers via WSL.
Adding either to the compose file fixed the low cpu usage entirely
"priviliged: true"
or
" cap_add:
- SYS_ADMIN
- SYS_NICE"