Category: Know How

System-on-a-chip or one-chip system

If the most important or even all components of a system are combined on a single chip, this is referred to as a system-on-a-chip. The most important component of such a system is the main processor – whether 8-bit or quad-core CPU, you can find all sorts on the market. Other processors, such as for graphics or audio processors for decoding and controlling audio data in smartphones, are often soldered on as well. The next component is the main memory, which is integrated via one or more memory chips. All internal components of the system are connected via the system bus (for data exchange between CPU, RAM, cache) and the peripheral bus (for USB, among other things).

Areas of application for SoCs

Single-chip systems are primarily found in the field of mobile devices and applications, i.e. smartphones and tablets, and in control and automation technology, i.e. washing machines or industrial automation. 

Advantages of a single-chip system

Thanks to their special design, single-chip systems not only save an enormous amount of space, so that they can be used almost anywhere, but also reduce costs. Furthermore, single-chip systems offer reduced power consumption.

Nvidia Tegra SoC integrates everything

As already mentioned, the TEGRA 2 released in 2018 is the first (and so far only) system in the spo-comm Mini-PC range to be based on an SoC chip. More precisely, on an Nvidia SoC chip. It integrates everything necessary for digital signage applications: an 8GB memory, a 32GB eMMC flash memory, the eponymous Nvidia Tegra X2 CPU, as well as a LAN and WLAN module. And all this in a housing measuring just 160 x 132 x 51 mm (W x D x H). The TEGRA 2 can not only play 4K@60 Hz content on two screens simultaneously. Thanks to the 256 integrated CUDA cores, it can also be used for deep learning applications and for real-time calculations and video editing in the automotive industry.

Configure your TEGRA 2!

The term ray tracing refers to an algorithm for calculating images. The aim is to represent the three-dimensional world in a two-dimensional image by simulating the physical behavior of light and light rays. The interplay of reflections, light and shadow creates a particularly realistic image for PC games or animated films.

Ray tracing vs. rasterization

Before the development of ray tracing, rasterization was already in use. However, this only calculates light and objects that are directly visible in the image (more on the difference between ray tracing and rasterization can be found on gamestar.de ). The special thing about ray tracing is that invisible objects or light sources are also analyzed and calculated, for example the sun in the sky, which is not visible in the image but is reflected in the puddle. Or light rays that are reflected by a window that is outside the visible range. 

How does ray tracing work?

So what exactly happens? The rays of light are followed by the viewer’s eye into the virtual space and up to an object. The surface of this object is then analyzed and, depending on its composition, the light beam is reflected, diffracted or absorbed. This allows the light beams to spread throughout the entire virtual space and capture all objects, regardless of whether they are visible or not. The end result is a very realistic picture for the viewer.

If you would like to learn more about the various ray tracing methods, you can find them clearly explained in a video from Nvidia.

Where is ray tracing used?

This rendering approach is not new and has been used in movies and advertising for some time. This is known as CGI (Computer Generated Imagery) – a technology that also manages to make artificial images look real.

The RTX graphics cards from Nvidia enable real-time ray tracing and are therefore suitable for PC games. However, rendering still takes quite a long time, which is why even the Turing graphics chips would be overwhelmed with complete game scenes. Nvidia therefore relies on hybrid rendering and mixes ray tracing with rasterization.

What hardware do I need for ray tracing?

The aforementioned Nvidia Turing graphics cards, i.e. the RTX 2000 models, were specially developed for ray tracing (hence the “R” in “RTX”). Thanks to our own ray tracing cores, you can calculate this in real time. With the appropriate new drivers, ray tracing is also possible with the Nvidia GeForce GTX 1000 graphics cards. However, good performance is not to be expected here. On the software side, Windows 10 with the DirectX 12 ray tracing API DXR version 1809 is required.

spo-comm industrial PC for ray tracing

If you are looking for a robust industrial PC for demanding graphics applications, our KUMO V is just right for you. It is equipped with an Nvidia GeForce RTX 2060 – with 6GB of integrated GDDR6 memory and 1920 CUDA cores. In addition to real-time ray tracing, the mini PC offers AI computing power and new shading technologies. No wishes remain unfulfilled.

Configure the KUMO V here

A hot swap is basically the replacement of PC peripherals while the system is running. The components to be replaced are not connected to the operating system or other software. This means that neither the power supply is interrupted nor the system restarted.  

Assuming a fan is defective while the Mini-PC is running, the damaged fan can be easily replaced with a new one thanks to hot swapping. Other components, such as PCIe or SATA drives in a RAID array, can also be swapped using hot swap.

Basis for hot swapping

For a component to be replaced using hot swapping, the associated plug connection must meet certain criteria. The plug contacts must have a leading design, which means that the contacts are connected and disconnected in a defined sequence, usually in three or more stages. This is realized by a common housing in which the contacts are positioned further forward and further back. The most common systems with leading plug contacts are SATA and USB (see Fig. 1).

The user should also ensure that the plugs are plugged in and unplugged as straight and slowly as possible.

Good to know: What is USB?

What are the advantages of hot swapping?

As already mentioned at the beginning, the advantage of hot swapping is that systems do not have to be restarted and there is no need to interrupt operations to replace broken components. Particularly in the area of servers or in industry, where PCs have to be in use 24/7, the rapid replacement of components using hot swap offers a decisive advantage, as even the slightest interruption to operation can mean an enormous loss.

Variants of hot swapping

• Hot Add: Describes the addition of new components that work without configuration in the running system.
• Hot Plug: The addition and removal of components during operation as long as no software accesses them, also known as
  plug and play
  .
• Hot switch: Switching between components, for example to a redundant power supply unit.

Hot swappable spo-comm Mini-PCs  

• RUGGED GTX1050 Ti
• RUGGED Ryzen
• RUGGED Q170
• MOVE QM87

What does “embedded” mean?

The term “embedded” means “embedded”. An embedded system is a computer that is integrated into a technical environment and performs tasks such as monitoring, control or data processing. These include, for example, microcomputers in washing machines, televisions, routers, refrigerators and cars. However, we are talking about somewhat larger systems here: Embedded PCs that are equipped with a corresponding CPU.

Where are embedded CPUs used?

Embedded CPUs are primarily used in professional environments, for example for industrial applications, in vehicles or in medical technology.

What are the advantages of an embedded CPU?

The CPUs are characterized by increased reliability, an extended temperature range and, above all, long component availability. Intel, for example, guarantees long-term availability of up to 15 years for its embedded processors. The advantage of this is that once an application is running, its use is guaranteed for years to come and the developers do not have to adapt the hardware and software again after a short time. Embedded systems are also often certified. If the CPUs are available for many years, there is no need for expensive recertification or recertification.

Another advantage is that embedded CPUs are extremely robust, as they are usually soldered processors with BGA (Ball Grid Array), which provide a high level of mechanical and thermal robustness. This means that embedded PCs can run in 24/7 continuous operation in harsh environments without any problems. Thanks to their compact design, embedded CPUs are perfect for mini PCs and also score points with their low power consumption.

What are the disadvantages of an embedded CPU?

With all the advantages, the question arises as to whether an embedded CPU has any negative characteristics at all. In fact, we can only mention the low performance here because, as with mobile CPUs, you can’t fit too much into a small space. However, a lot has happened here in recent years. And for the areas in which embedded CPUs are predominantly used, it is certainly sufficient.

Good to know: What are mobile CPUs?

Intel and AMD embedded CPUs

Both major processor manufacturers have various models on offer: Intel offers embedded CPUs in the Celeron, Pentium and Core i3 series. AMD offers the Ryzen Embedded and Epyc Embedded series.

The abbreviations BGA, PGA, LGA and CCGA denote housing forms of integrated circuits. However, they differ in the type of connections.

What are integrated circuits?

Ball Grid Array (BGA)

The abbreviation BGA stands for “Ball Grid Array “. In this type of housing, small solder balls form the connections, which are arranged in a square grid of columns and rows on the underside of the chip. This design means that considerably more connections can be accommodated, about twice as many as with PGA. The solder balls provide short connections and therefore enormous performance.

The advantages of BGA are the small space requirement, good heat dissipation and low impedance due to short connection paths to the PCB. The chips can also be desoldered from the circuit board without damaging them. This makes it possible to remove old beads (deballing) and fill them with new beads (reballing). The chip can then be soldered onto a new circuit board. As soldered processors are mechanically and thermally extremely robust, BGA is mainly used in embedded CPUs.

A major disadvantage is that the solder joints can only be checked by X-ray, as the connections are concealed and difficult to access. This also severely restricts repair options. Special equipment, a so-called reflow oven, is required for safe soldering. In addition, BGA chips can only be used sensibly on multilayer boards, which somewhat limits their application possibilities.

Pin Grid Array (PGA)

The so-called “Pin Grid Array ” ( PGA ) is mainly used in processors. While BGA uses soldering beads, the pin grid array – as the name suggests – uses small pins (i.e. “contact pins” or “legs”) as connections. These are also arranged in a square grid, but the number of connections and arrangement of the arrays varies, so that there are a large number of variants and therefore also different CPU sockets. The rows of pins can be arranged in parallel or staggered and are marked with numbers and letters.

There are different variants of PGA: 

• In the Ceramic Pin Grid Array (CPGA), the semiconductor chip is fixed to a thermally conductive ceramic carrier. It is used in the first generation Intel Pentium, Socket A variants of the AMD Athlon and the Duron family.
• In the Plastic Pin Grid Array (PPGA), the carrier for the semiconductor chip is made of plastic. This variant is somewhat cheaper, has better thermal properties and also improved electrical performance than ceramic. PPGA is mainly used in Pentium MMX processors and Celeron.
• The Staggered Pin Grid Array (SPGA) is characterized by staggered rows of connections. This variant is required for CPUs that have more than 200 connections, as the offset layout provides more space. It is used in the Pentium and later central processing units.
• In the flip-chip pin grid array (FCPGA), the integrated circuit is attached to the top of the carrier (“flip-chip” means “upside-down, turned chip”). This design is used, for example, in Pentium III and some Celeron processors.

As the pins are located on the CPU with PGA, the corresponding holes are on the motherboard socket so that the CPU can be installed without great pressure.

Land Grid Array (LGA)

The “Land Grid Array” (LGA) is the exact opposite of PGA, as the contact pins are located on the socket of the mainboard. The CPU has the same number of contact points with which a connection is established. Intel has been using LGA for the majority of its Celeron, Pentium, Core and Xeon CPUs for many years.

One advantage of LGA is the smaller size of the pins, which allows a larger number of pins on the same surface area. Secondly, they are not so easily damaged because they do not have pins that can be crushed. Compared to LGA, PGA sockets have the advantage that the mainboard cannot actually be damaged. It is also easier to repair pins on a PGA processor than on an LGA mainboard.

Ceramic Column Grid Array (CCGA)

Even if it has nothing to do with mini PCs, we would like to mention the Ceramic Column Grid Array (CCGA) here for the sake of completeness. CCGA housings are extremely reliable and are used in aerospace and military technology. The solder connections on the underside of the housing are column-shaped (hence the name “column”) and consist of highly leaded solder. The columns are arranged in a grid, similar to BGA. They cannot be used in the civilian market as they are not permitted under the RoHS directives due to the high lead content as a result of EU trade bans.

Good news for Linux fans: Free support for Nvidia Turing

Since the release of the Turing graphics cards in fall 2018, Linux developers from Nouveau, a collection of free Linux drivers for Nvidia graphics cards, have been working on reverse engineering suitable drivers. The problem is that the firmware is signed and can only be provided by Nvidia. However, the graphics card manufacturer has so far been unwilling to share its data with the Nouveau developers. This has now apparently changed and the Linux team is working intensively on drivers for hardware acceleration. Why are we interested in this? Quite a few spo-comm customers who rely on the KUMO series for high-end graphics applications work with Linux. And the latest model, the KUMO V with a dedicated Nvidia Turing graphics card, is already in the starting blocks and will be presented at ISE in February.

Discover the KUMO series

No more support: The end of Windows 7

As announced a long time ago, Microsoft released the last security updates for Windows 7 SP1 systems on January 14, 2020. As a result, hardware and software manufacturers will no longer support the popular operating system in the future. Anyone who continues to work with Windows 7 is exposing themselves and their applications to high security risks and should switch to Windows 10 or another operating system as soon as possible. If a changeover is not possible in a company environment, Microsoft’s Extended Security Update Program (ESU) offers an option. Further information can be found at heise.de.

The interesting alternative: Windows 10 IoT Enterprise

If you are looking for the right operating system and a simple, cost-effective software solution that can be perfectly adapted to your needs, Windows 10 IoT Enterprise may be the perfect solution. spo-comm now offers this in three variants: BASIC, ON and OFF. With the BASIC package, which costs no extra, customers receive an unedited image in the standard languages German and English. With the ON version, the display language is freely selectable and we update the update status quarterly so that the Mini-PC is ready for use immediately after delivery. If you want to make sure that your application runs smoothly and no updates get in the way, choose the OFF package. Here the updates are completely deactivated. The language is also freely selectable. In addition, programs such as Cortana, OneDrive and XBOX-Network, which are not required for the majority of industrial applications, are removed in the latter two variants.