Skip to content

Releases: Livepool-io/transcoder

v0.7.5

21 Jun 03:09
9cacf8e
Compare
Choose a tag to compare

This is the latest release, bringing livepool back to feature parity with go-livepeer.

Confluence

01 Mar 15:29
Compare
Choose a tag to compare

v0.5.22: Verification stability

31 Oct 14:51
Compare
Choose a tag to compare

Summary

This release includes an important MPEG-7 video signature capability fix, which is used for fast verification, for orchestrators/transcoders running on Windows. If you are running an orchestrator or transcoder on Windows you should upgrade to this release as soon as possible..

Full release notes

v0.5.21 - New regions, Verification

19 Oct 10:19
Compare
Choose a tag to compare

Livepool additions

  • Better region support for EU
  • New region support for US-west
  • Re-enable payments

Summary (from Go-Livepeer)

This release includes a new orchestrator/transcoder MPEG-7 video signature capability which is required for broadcasters to run transcoding verification. The MPEG-7 video signatures are used as perceptual hashes in a "fast verification" algorithm that is described in the Fast and Full Transcoding Verification design. If a CPU is used for transcoding, the video signature computation will occur on the CPU. If a GPU is used for transcoding (i.e. the -nvidia flag is used), the video signature computation will occur on the GPU. Orchestrators and transcoders should upgrade to this release as soon as possible to ensure that they will be able to continue serving broadcasters that run fast verification (the complete broadcaster implementation will be shipped at a later date).

Orchestrators and transcoders should also make note of the breaking change in this release that drops support for Nvidia GPUs from the Kepler series.

Thanks to everyone that submitted bug reports and assisted in testing!

Breaking Changes 🚨🚨

  • #2027 Nvidia GPUs belonging to the Kepler series (GeForce 600, 700 series) and older, are no longer supported by go-livepeer. Cuda 11 is needed for newer GPUs, which only supports the Maxwell series or newer.

Full list of changes

Transcoder Workflow Updates

27 May 01:29
Compare
Choose a tag to compare
  • The pool will now re-use the transcoder for all chunks of a stream as long as it performs well

  • Better handling of input video related errors

Running a Transcoder

Windows

Unzip the archive

From the unzipped folder run

./livepool.exe -ethAcctAddr <your_eth_address> -nvidia <nvidia GPU IDs e.g. 0,1>

Linux

Unpack the tarball

tar -zxvf livepeer-linux-amd64.tar.gz

Move the binary to executable path

mv livepeer-linux-amd64/livepool /usr/local/bin/

Start the transcoder

livepool -ethAcctAddr <your_eth_address> -nvidia <nvidia GPU IDs e.g. 0,1>

On success you should see

I0527 03:28:58.192433   19677 ot_rpc.go:132] Transcoder started

Max sessions

Max sessions is set to 3 by default. A single GPU can handle more than that but then it consumes more recources.

If you wish to increase this you can do so by adding the -maxSessions flag to your startup command

e.g. -maxSessions 10

Check out this link for more documentation: https://livepeer.org/docs/video-miners/how-to-guides/session-limits

v0.5.14 Multi-Region Support 🌍

26 Feb 20:44
Compare
Choose a tag to compare

** Changelog**

  • Multi-region support
    • Europe - Frankfurt
    • North America - Toronto
    • Asia - Signapore
  • Improved session handling and reduction of stuck sessions

Full list of changes to be found at https://github.com/livepeer/go-livepeer/releases/tag/v0.5.14

0.5.10 Release

05 Sep 20:07
Compare
Choose a tag to compare

This release includes breaking changes, all transcoders are advised to upgrade to this release.

Check out the go-livepeer v0.5.10 release notes for a full list of changes

v0.5.9

29 Jul 13:23
Compare
Choose a tag to compare

See go-livepeer 0.5.9 release notes https://github.com/livepeer/go-livepeer/releases

v0.5.7

03 May 16:16
Compare
Choose a tag to compare

Windows Users

This release fixes the broken v0.5.5 Windows binary and also includes a small update to the HTTP ingest response format for broadcasters.

There is a known issue with the GPU transcoding test that the Windows binary runs on start up, but transcoding works after start up. This will be fixed in a future release, but for now a Windows based orchestrator and/or transcoder should be started with the -testTranscoder=false flag.

Important Changes

  • Enabled filtergraph persistence in between segments during GPU transcoding. This update results in roughly a 30% performance gain when using a GPU to transcode a 1080p input into 720p, 576p, 480p and 360p outputs (#1444)
    With this update, the transcode job serialization introduced in v0.5.5 is no longer needed (as it was a temporary solution for sub-optimal GPU transcoding performance) so that code has been removed

  • Fixed an overflow bug that could cause an orchestrator to advertise a price different from the price that it expects payments to be based on (#1450)
    This resolves the invalid ticket recipientRand error that many orchestrators have been observing in their logs when transcoding streams

Running Livepool

  1. Download the release
  2. Unpack the archive
  3. (optional) windows: move the executable to your PATH , others: move the binary to /usr/local/bin
  4. Run the binary with livepool -ethAcctAddr <YOUR_ETH_ADDRESS> -nvidia <YOUR GPU PCIe IDs>

v0.5.5 Final release

20 Apr 12:44
Compare
Choose a tag to compare

Livepool Video Miner v0.5.5 ⛏️

This is the first release of the Livepool Video Miner, it includes executables for Linux, Darwin and Windows.

After downloading and unpacking the archive for your OS move the livepool binary into your $PATH or execute it from your current directory.

To start your livepool video miner you need to provide on which ethereum address you'd like to receive payments and which GPUs you want to use (currently only Nvidia GPUs are supported).

Eg. `livepool -ethAcctAddr <YOUR_ETHEREUM_ADDRESS> -nvidia 0

Will start the livepool video miner using the GPU in your main PCI-express slot. If you have multiple GPUs you can provide a comma-separated list. To get the PCI-e IDs for your GPUs you can run nvidia-smi