STEMMER IMAGING Modular Embedded
24 Nov 2022 | Reading time: 2 min
More speed thanks to TCP offload
Interview with Andreas Rittinger - Software Development - STEMMER IMAGING
A powerful hardware offering combined with a variety of state-of-the-art cameras, leading software tools and individual service packages: That's our new STEMMER IMAGING Modular Embedded ecosystem. The integrated TCP offload technology for image capture offloads the processor while enabling maximum performance to be achieved with low power consumption.
In this article, we would like to introduce you to TCP offload and its advantages in more detail. To do this, we spoke to Andreas Rittinger.
Andreas Rittinger, Team Manager Software Development at STEMMER IMAGING
Andreas, can you briefly describe TCP offload??
TCP offload means that TCP network packets are unpacked from the network card and assembled into larger segments.
This task is usually done by the CPU but can be done to a large extent on hardware with modern network set-ups. Another common term for TCP offload is Receive Segment Coalescing (RSC).
What are the advantages of this technology, especially for embedded systems?
To understand the advantages, you need to compare TCP offload with the UDP based GigE Vision Streaming protocol. Originally, UDP was chosen because it is a simpler and therefore is a less computationally intensive protocol compared to TCP. In contrast to TCP, UDP is connectionless. This makes the transmission of individual packets much less reliable. Since GigE Vision is usually used in dedicated networks, and often just point to point, this disadvantage is bearable considering the advantage of a lower computational load.
In modern network cards, TCP offload can more than compensate for the additional workload caused by the TCP protocol and is thus superior to UDP. This advantage becomes even more obvious at higher data rates with NBASE-T capable devices.
Embedded systems in particular benefit from the lower processing hardware requirements when using TCP Offload compared to UDP. This mainly affects CPU load, but also memory load as fewer packets have to be copied.
What is special about TCP Offload in STEMMER IMAGING Modular Embedded?
Our TCP offload implementation is designed as a GigE Vision standard extension and not a proprietary solution. We have implemented this design together with Allied Vision.
Can TCP offload be applied to other vision systems?
TCP offload technology works independently of the platform (embedded or not) and therefore is available on all vision systems.
However, on Windows systems, a firewall usually prevents efficient implementation in user space.
Are there applications for which this technology is particularly suitable?
In principle, all applications can benefit from a lower CPU load during image acquisition. The computing power saved is then available to the application. Computationally intensive applications can be stabilized in this way or, in extreme cases, become possible in the first place if hardware scaling was not possible.
If the reduced computing power is not needed elsewhere, you can at least be happy about the reduced energy used.
In addition, all applications where the integrity of the individual images is of particular importance can benefit, as this can be implemented much more easily with TCP.
A brief look into the future: What is the next stage in the further development of TCP offload technology?
TCP offload is based on features available in standard network cards. Thus, further technical development essentially depends on their development. In the future, there will of course be optimizations, for example in the camera firmware.
Currently, however, the focus is on promoting this technology to expand its reach. It is important for us that TCP offload becomes a GigE Vision standard feature and offers an alternative or addition to the established UDP-based protocol.
We are therefore actively involved in the standardization process to ensure that this technology is adopted by as many camera manufacturers as possible in the market.