PC requirements

SRTMiniServer requires Windows systems, 64 bits:

  • Windows 10
  • Windows Server 2008 and next

Q: Can I use SRTMiniServer in cloud like AWS or Google Cloud ?
A: YES!

About PC specifications (CPU, GPU, RAM)

We got many questions about PC specifications for SRTMiniServer

  • What is CPU I need for 5 feeds simulteniously ?
  • What is GPU I need ?

It strongly depends on the parameters of incoming streams.

For example: my notebook with Core-i5 uses 17% for 4 incoming streams of HD@25fps/2Mbs (see details below)

Before let me note SRTMiniServer does two expensive operations:

  • decode incoming feed (from h264 or h265 or mpeg2)
  • encode to NDI feed

Let's go over these points

Decode incoming feed

SRTMiniServer allows select "software decode" or "hardware decode"(means GPU). But we recommended use "software" in many cases.

Because:

  • "hardware" decoding is very sensitive to format. If shortly - NVidia decoder perfect for feed only from NVidia encoder.
  • hardware decoder usually has limit for simultaneous session (usually 2-3).

Resume: using CPU for decode is preferable.

Encode to NDI

NDI uses only CPU. No options.

Resume

  • GPU does not make sense

  • CPU is make sense. We recommended use Core i7 or Core i9 and 16 Gb RAM.


Tips and examples

Do stress test

Perfomance very sensitive to incoming feed: 4K@60fps and HD@25fps show different CPU usage.

So common recommendation is make stress-test for got real perfomance of your PC.

UPDATE: we prepare special tool for stress test ( here )

For stress-test you need (on other host):

  • ffmpeg
  • pre-recorded file (mp4)

If you plan use HD@25fps / 5Mbs, then prepare mp4 (h264/aac, HD@25 5Mbs) duration ~5 mins.

Use this command for send SRT stream to SRTMiniServer by ffmpeg:

ffmpeg -re -i hd_25_example.mp4 -vcodec copy -acodec copy -f mpegts "srt://192.168.1.53:9001?streamid=streamkey1ABC"

you need change IP and streamid for your value

You can run many commands (with different streamid) at same time for generate many feeds.For example I run 4 feeds to check my notebook for HD@25fps feeds:

my notebook with Core-i5 uses 17% for 4 streams of HD@25fps/2Mbs


example: I run few streams from macOS


Don't use Preview window long time

Built-in Preview window takes CPU for rendering, so no reason use it on slow PC long time. In our test we got increase CPU usage from 17% to 35% when open Preview window

Like most websites our site uses cookies to ensure that we give you the best experience.