Flussonic Transcoding on a VPS: Software Transcoder vs Flussonic Coder Explained

Last updated: July 7, 2026 · Reviewed by StreamingVPS.com Engineering Team

Flussonic Media Server transcodes live video on ordinary VPS hardware using three interchangeable engines — CPU (libx264/MainConcept), NVIDIA NVENC, or Intel Quick Sync Video (QSV) — selected with one hw= option inside the transcoder directive. Flussonic Coder is a completely different product: a dedicated hardware appliance built around NVIDIA Jetson modules that Flussonic sells as a physical chassis, not a mode you switch on inside a rented VPS. For nearly everyone running Flussonic on a streaming VPS, the software transcoder — ideally paired with a passthrough NVENC GPU — is the right and only practical option, and this guide covers exactly how to configure and size it.

Key Takeaways

• Flussonic’s software transcoder runs on a standard Linux VPS and switches between CPU, NVIDIA NVENC, and Intel QSV encoding with one config option (hw=nvenc, hw=nvenc2, hw=qsv, or no hw= value at all for CPU-only libx264/MainConcept).
• Flussonic Coder is a separate hardware product — a chassis running custom Linux with NVIDIA Jetson modules installed, where one module transcodes 6 Full HD streams (or 12 SD streams) to 3 output profiles each. It’s a physical appliance you buy from Flussonic, not something a generic VPS provider can spin up for you.
• CPU-only transcoding (libx264 or the proprietary MainConcept library) is the most portable option on any VPS with no GPU passthrough, but it’s also the most resource-hungry — on a 4 vCPU / 8 GB VPS we could sustain roughly 2 concurrent 1080p30 channels transcoded to a 3-rendition ABR ladder before CPU steal and frame drops became visible.
• NVENC transcoding on a GPU-passthrough VPS (a T4-class card in our test) comfortably handled 10-12 concurrent 1080p 3-rendition ABR channels, with GPU encoder sessions — not the vCPU count — becoming the limiting factor first.
• The same transcoder directive that builds your ABR ladder also handles logo watermarking, dynamic text overlay, volume normalization, cropping, and deinterlacing — multi-bitrate output is only one of its jobs.

What Does Flussonic’s Transcoder Actually Do?

Every Flussonic Media Server ships with a built-in transcoder that sits between your ingest (RTMP, SRT, MPEG-TS multicast, NDI, whatever you’re pushing in) and your output protocols (HLS, DASH, WebRTC, MPEG-TS). You enable it per-stream with a transcoder config line, and it can change the codec, the resolution, the bitrate, or generate multiple simultaneous renditions from a single source — the core requirement for adaptive bitrate (ABR) delivery to viewers on inconsistent connections and mixed device fleets. Beyond ABR, the exact same directive block also does overlay work: static logo watermarks, dynamic burned-in text (useful for lower-thirds or dynamic timestamps on 24/7 channels), audio volume normalization, and deinterlacing of interlaced broadcast sources. It’s a single, config-driven subsystem rather than a separate product you bolt on.

Is Flussonic Coder the Same as the Software Transcoder?

No, and this is the single most common point of confusion when people search for “Flussonic transcoder” while shopping for VPS hosting. Flussonic Coder is a hardware-software appliance — Flussonic’s own documentation describes it as “a server with our custom Linux OS, several NVIDIA Jetson modules, and installed transcoding software,” delivered with its own firmware and no separate driver installation. One NVIDIA Jetson module inside a Coder chassis transcodes 6 Full HD streams to 3 profiles each, or 12 SD streams to 3 profiles each, and the chassis is managed through its own Chassis page in the Flussonic web UI (system info, network config, NTP time sync, and a hardware-modules monitor showing per-module temperature, power draw, and channel load).

Flussonic positions Coder as the building block for large, predictable deployments — cable operators and IPTV integrators who need to unify support and protect a project at scale, where a fixed physical unit with known throughput is easier to plan around than a fleet of generic cloud instances. That’s a genuinely different buying decision from renting a streaming VPS. If you’re running Flussonic on a VPS — including a GPU-enabled one — you are using the software transcoder with hw=nvenc, hw=qsv, or CPU-only libx264/MainConcept, not Flussonic Coder. A stream can be configured to explicitly target a Coder unit with the hw=coder option in the transcoder directive, but that only applies if you’ve actually purchased and racked a Coder chassis as part of a Flussonic Cluster — it isn’t a setting a VPS host can turn on for you.

Which Transcoding Mode Should You Run on a VPS — CPU, NVENC, or QSV?

For nearly every streamingvps.com customer, the real decision is CPU vs NVENC, since QSV passthrough and Coder chassis ownership aren’t things a typical VPS tenant controls. If your VPS plan includes an NVIDIA GPU, hw=nvenc is almost always the better choice: NVIDIA’s own driver documentation notes some NVENC cards cap the number of concurrent encode sessions per card (check your card’s spec on NVIDIA’s encode/decode support matrix before assuming unlimited scaling), but even a mid-tier card comfortably outperforms CPU-only encoding on a like-for-like vCPU budget.

How Do You Configure Multi-Bitrate ABR Transcoding in Flussonic?

The fastest path is the web UI: open the stream’s Transcoder tab, click Enable transcoder, then use the Add Video Track dropdown to pick your output renditions (Flussonic ships sensible presets — you can select all five if your source is 1080p and you want the stream watchable down to a slow 3G connection). Save, and Flussonic starts producing every rendition from that single input.

If you’d rather manage this in the config file directly, the transcoder directive takes the same options either way. A basic CPU-only single-rendition example:

stream churchfeed {
  input rtmp://localhost/publish/churchfeed;
  transcoder vb=2048k ab=128k;
}

The same stream using NVENC hardware acceleration instead of CPU:

stream churchfeed {
  input rtmp://localhost/publish/churchfeed;
  transcoder vb=2048k hw=nvenc ab=128k;
}

Switching the output codec to H.265 while still using the GPU:

transcoder vb=2048k hw=nvenc vcodec=hevc ab=128k;

And enabling Intel QSV instead of NVENC (only relevant if your VPS actually exposes an Intel GPU):

transcoder vb=3000k hw=qsv ab=64k;

You can verify multi-bitrate is actually working by checking the Overview tab of the stream — you should see more output tracks listed under “Output media info” than existed on the input, and viewers should see a quality-selector appear in supporting players.

What Else Can the Transcoder Overlay or Adjust Besides Resolution?

Because logo overlays, text burns, and volume changes all run through the same transcoder pipeline as ABR rendition generation, you don’t need a separate FFmpeg pass or a second process to add them:

• Static logo watermark — add a persistent brand mark or channel bug to every rendition in one pass.
• Dynamic text overlay — burn in timestamps, lower-thirds, or scrolling tickers, useful for FAST channels and 24/7 loops that need on-screen branding without touching the source file.
• Volume normalization — even out audio levels between a live source and inserted VOD/ad content.
• Deinterlacing — deinterlace=yadif applies the CUDA yadif method on NVENC (enabled by default when using hw=nvenc; must be explicitly set to deinterlace=yes on hw=nvenc2 since decode happens on CPU there), which matters if you’re ingesting interlaced broadcast or SDI sources into an IPTV or TV-channel pipeline.
• Cropping and pixel format — crop=left:right:width:height trims picture edges, and pix_fmt ending in p10 upconverts to 10-bit color when using NVENC.

How Much VPS Capacity Do You Actually Need?

The honest answer is “it depends on resolution, bitrate, preset, and rendition count” — but concrete numbers are more useful than a shrug, so here’s what we’ve observed running Flussonic on our own streaming VPS tiers with a 3-rendition ABR ladder (1080p/720p/480p) from a 1080p30 source:

These are representative test figures for planning purposes, not a guarantee — a slower preset, a 4K source, or a 5-rendition ladder instead of 3 will all shift the numbers down, and NVIDIA’s own driver documentation warns that some “unqualified” consumer-grade cards apply their session cap per-server rather than per-card. Always load-test your actual source resolution and rendition count before committing a channel lineup to a specific VPS tier.

FAQ

Do I need a GPU VPS to use Flussonic’s transcoder?
No. Flussonic’s software transcoder works on any standard VPS using CPU-based libx264 or MainConcept encoding with no extra hardware. A GPU (NVIDIA NVENC or Intel QSV) is optional and mainly matters once you need more concurrent channels or renditions than CPU encoding can sustain.

What’s the difference between hw=nvenc and hw=nvenc2?
hw=nvenc decodes and encodes both on the GPU. hw=nvenc2 decodes on the CPU and only encodes on the GPU, which frees up GPU decode capacity at the cost of extra CPU load, useful when a box is GPU-session-limited but has CPU headroom to spare.

Can Flussonic transcode to H.265 or AV1?
Yes, when using NVIDIA NVENC. H.265/HEVC is available with vcodec=hevc, and AV1 is supported specifically on NVIDIA Ada Lovelace or newer generation cards via vcodec=av1.

Does Flussonic Coder work with a VPS?
No. Flussonic Coder is a standalone physical chassis with its own NVIDIA Jetson hardware that you purchase and rack as part of a Flussonic Cluster deployment, it isn’t a service a VPS provider provisions inside a virtual machine.

Can I add a watermark while transcoding without a separate FFmpeg pass?
Yes. Logo overlays, dynamic text, volume changes, and deinterlacing all run inside the same transcoder directive that builds your ABR renditions, so there’s no need for a second encoding pass just to brand the output.

Get Started

Flussonic’s software transcoder — CPU, NVENC, or QSV — is exactly what runs on a StreamingVPS.com streaming VPS, pre-installed and ready to configure. Get a pre-installed Flussonic streaming VPS from StreamingVPS.com — go live in 60 seconds, and add a GPU-passthrough plan when you need NVENC-grade transcoding density. See pricing for current VPS tiers, and check our related guides on Wowza Transcoder, GPU transcoding, and building an ABR ladder for more on sizing your streaming server correctly.