Anatomy of a motherboard
Gavin Bonshor breaks down just what makes your motherboard tick, in this primer on the computing world’s most underlooked piece of tech
Let’s face it, much of the buzz surrounding product launches of new CPU architectures revolve around core count, thread count, clock speed, turbo clock speed, supported memory speeds – you name it. However, if the CPU is the engine of a system, then the motherboard is the manifold without which nothing would function.
With so many different key components that make up a computer system, the motherboard is the central hub that binds and connects the processor to the graphics card, memory, storage, and so on. Not only that, but it also connects a whole host of controllers from the motherboard chipset to the CPU, which essentially manifests into input and output, such as USB ports, audio jacks, M.2 slots, networking controllers like Ethernet and Wi-Fi, and even Thunderbolt 3 on premium models. At the launch of every new chipset, motherboard vendors release swathes of models, from the more budget-focused to the highly premium that are decked out with more features than you can shake a stick at.
The core fundamental aspects of a motherboard consist of much the same across all models, whether that’s an AMD or Intel-based board, with each motherboard vendor implementing its own aesthetic and feature sets based on support from both the processor and the motherboard chipset.
In this issue, we’re going to take a journey through how a motherboard “works,” what components make up a conventional consumer-level desktop motherboard, and why each component is important.
The modern motherboard: a basic diagram
One of our favorite quips within this industry is that “no motherboard is created equally,” and although true to a certain degree, in this instance it doesn’t have much swing. This is because each desktop motherboard has the same core characteristics that makes it a motherboard, and that remains the same across each platform regardless of processor, brand, or chipset model. Let’s have a look at a typical motherboard and its most commonly associated and distinctive features.
Each motherboard will include a different feature set, and as a result will include different headers – some more and some less – especially when compared with the premium Asus Prime X299-Deluxe motherboard we’ve illustrated. This includes things like M.2 slots, with some models opting for one, and some even managing to include three or four. Parts including Power/Reset buttons and LED debuggers are more commonly found on premium models, and are omitted from a lot of budget-focused models primarily due to cost.
Now let’s take a dive into some of the more intricate motherboard characteristics and commonly
used terminology, and what it means.
Form factor: size matters
When it comes to motherboards and deciding which one is the right one for the job, the form factor is one of the most important aspects for a multitude of reasons. Firstly the form factor relates to the size of the motherboard, with the most common form factors consisting of ATX, micro-ATX and mini-ITX. Each includes its own set of characteristics, as the smaller the motherboard, the less that can be placed onto it. Technically speaking, a mini-ITX system will have a smaller desktop footprint than that of an ATX-sized PC, and will restrict things like memory capacity and expansion slot (PCIe) support. The dimensions of the three most common form factors are as follows:
• ATX – 305 x 244mm
• Micro-ATX – 244 x 244mm
• Mini-ITX – 170 x 170mm
Expansion slots: the TL:DR on PCIe
Related to sizing, each motherboard includes an array of expansion slots, with PCIe x16 slots or full-length slots for devices such as graphics cards, and PCIe x1 slots for things like network controllers, SATA controllers, and premium sound cards. Each motherboard chipset and CPU has a limited amount of PCIe and high-speed input and output (HSIO) lanes, which determine how many slots can be implemented and how they can be used. A common example of this is multi graphics card support, which will require two or more fulllength PCIe slots, with AMD Crossfire two-way requiring a minimum of PCIe x4 to operate, and Nvidia SLI demanding at least PCIe x8. On more recent models, there is support for the PCIe 4.0 standard (here’s looking at you AMD Ryzen 3000), which is the latest and greatest, but there is still little in the way of components on the market to fully utilise the extra bandwidth available. Most motherboards on the market include PCIe 3.0, although it is backwards compatible, meaning a PCIe 3.0 card could be placed into a PCIe 4.0 slot, but it would just operate at PCIe 3.0 speeds. Some older models and the cheapest of the cheap sometimes include PCI slots, which is an older standard and virtually non-existent in today’s ever-evolving industry.
CPU: socket LGA and PGA
One of, if not the most important trait of a motherboard is the CPU socket, as this is where the PC’s brain and engine rests. For consumer models, there are two primary socket types: LGA, or land grid array; and PGA, also known as pin grid array. AMD has opted for a PGA socket that, when reverting to basics, means the pins are on the CPU, with pin slots in the socket; for example socket AM4. For as long as we’ve been using computers, Intel has used an LGA design, which means the pins are in the socket.
Regardless of whether or not the pins are in the CPU socket or on the CPU itself, these are very fragile, and one slight knock or slip can result in dead hardware. To make the installation of CPUs easier, Intel LGA sockets use a mechanism and latch to create compression force when installing it, with a notch to show which way the processor is installed. For AMD, there is a similar notch or arrow in one of the corners of the CPU, and it uses a handle to release and hold the CPU in place. It’s worth noting that most manufacturers don’t include installation socket damage in their warranty cover, so great care must always be taken when installing the CPU.
What is the chipset?
The second most important and functional part of a motherboard is the chipset. There are various chipset types, each with its own set of support, including PCIe, SATA, USB, and on newer chipsets, integrated networking PHY or a physical layer. The term chipset is a far-reaching term and in itself, acts as the adhesive and communication hub between the CPU and the motherboard’s componentry, including slots, and input and output.
While the CPU is by far the most
influential common denominator in regards to overall system performance, it drives the PCIe x16 slot and – using the AMD Ryzen 3950X processor as an example – includes integrated PCIe 4.0 lanes. The latest performance chipset from AMD, the X570 chipset, includes PCIe 4.0 lanes within it, which enables users to benefit from PCIe Gen4 M.2 SSDs for trailblazing read and write performance. The most recent chipset from AMD is the B550 chipset, which is a watereddown and budget version of the X570 chipset, but instead of PCIe 4.0, it uses PCIe 3.0 in the chipset.
This is important to know. Generally speaking, the top full-length PCIe slot will always be driven by the PCIe lanes from the CPU, and on premium models, the second full-length slot too. This also includes the top M.2 connector, which is generally powered by the CPU too. Expansion slots, such as additional full-length PCIe slots outside of the top or top two, and PCIe x1 slots are all controlled from the chipset. The chipset also controls a motherboard’s core SATA ports and additional M.2 slots.
Chipset support: budget, performance and HEDT
Another important thing to note is that some processors will only work with certain motherboard chipsets, even including those that feature the same socket type, for example Intel’s LGA115x platforms. There are budget chipsets that generally begin with the letter A, B, and H, followed by the series, such as B360 for Intel’s 9th-generation Coffee Lake processors, or A520 which is designed for AMD’s Ryzen 3000 series and Athlon processors. Going even further into chipsets, there are the more premium variants, such as Intel’s latest Z490 chipset for its 10th-generation Comet Lake, and AMD’s latest X570 chipset, which generally speaking are both the flagship consumer chipsets for performance and premium features. These models for the most part include certain traits, such as four memory slots with support for dual-channel memory, premium integrated audio, and more USB support than most users will ever need.
The crème de la crème of motherboard chipsets are the high-end desktop models, or as they are commonly known, HEDT. These chipsets include Intel’s X299 and AMD’s TRX40, which offer superior memory and connectivity support, and are commonly used by content creators who demand high core-count processors for intensive workloads, such as video rendering, image manipulation, or for just pure E-peen. These typically include support for quad-channel RAM and support for higher capacity memory, and usually include up to eight memory slots, compared to four on even the most high-end consumerbased model.
Choosing the right chipset is important, and it goes hand in hand with CPU selection, as both need to support each other. Some chipsets even include active cooling solutions as they can run very hot, especially AMD’s X570 chipset, which generally comes with a cooling fan integrated into the chipset heat sink. When buying a new motherboard, always check which chipset supports which CPU, as installing the wrong processor into the wrong motherboard can not only cause damage during installation, but can effectively kill hardware very quickly if not paired up properly.
The power delivery: VRMs and what they do
In recent years, a lot of focus on selecting a motherboard has been around the power delivery, or VRM
as it’s commonly known. The term VRM stands for voltage regulation module, and a basic definition of what it does includes controlling the power to the CPU as well, as the system on a chip–or SoC – which is commonly used by processors with integrated graphics. It also converts the power down from the 12V rail on the motherboard – through the 12V ATX power inputs – to a workable DC voltage, as the VRM also acts as a buck, which converts DC to DC power. This conversion is important as it enables the CPU to operate with sustainable voltages, which, if it wasn’t converted, would blow the processor pretty much instantly.
A basic example of how a VRM works is as follows. It starts by pulling power from the 12V rail from the power supply into the power delivery or VRM. The VRM then converts this 12V power into a lower voltage so the processor can use it, and it also regulates it so that it stays well within the operating voltage that the CPU requires to work. Even with overclocking the processor, CPU core voltage is typically kept under 1.4V, which in itself is around an 88% reduction in the voltage drawn from the power supply compared to what the CPU is actually using.
Another variable to consider when looking at a motherboard’s VRM is its capabilities in relation to maximum supported current. This is especially important when overclocking the processor, as some power deliveries aren’t designed for thunderous power loads, although motherboards in recent times are more than well-enough equipped. The configuration of the power delivery is also important, as one of the biggest grumbles is assuming a higher number of power stages or phases mean it has a better VRM. This is a gray area with some vendors in regards to marketing and advertisement, as a six-phase CPU VRM with six 90A power stages can deliver a maximum current of up to 540 amps, where an eight-phase CPU VRM with eight 60A power stages equates to a maximum current of 480 amps.
Some motherboard manufacturers use different types of configuration, with some offering direct phase designs that work in a one-to-one method, with one channel operating one phase from the PWM (pulse-width modulation) controller, which drives and controls the power delivery. Some opt for the use of doublers, which work with two phases operating from one channel. Asus has switched to “teaming” its phases in pairs, which is designed to deliver higher levels of burst current while maintaining efficiency, in regards to computational and thermal performance.
The benefit of more phases is generally associated with better reliability and can have a positive effect on operating temperatures, as usually the VRM is cooled by a simple heat sink. Some enthusiast grade models incorporate active designs with small cooling fans, with the most premium including VRM water blocks, such as the Asus ROG Formula series.
Networking & audio
For networking, some boards use a single RJ45 connector, while others use two for superior connectivity. The general shift in recent months has gone from 1 Gigabit to 2.5 Gigabit Ethernet ports, which motherboard vendors are looking to use, as the new go-to technology as the industry quickly moves far and beyond standard Gigabit. This also includes Wi-Fi, with a big
transition occurring this year from Wi-Fi 5 (802.11ac) to the better Wi-Fi 6 (802.11ax) standard. This is good for gamers who demand a solid connection during online gameplay but don’t necessarily have the scope or capability to use wired Ethernet, which is almost always the best connection standard for performance and network stability.
Other areas to consider include onboard audio, which usually comes with different inputs and outputs depending on the grade of audio codec that each model is using. More premium codecs, such as the Realtek ALC1220, feature five 3.5mm audio jacks and a single S/PDIF optical output, while lesser codecs like the Realtek ALC892/887 can consist of just three 3.5 mm audio jacks. This is something to factor in when purchasing, or you could use a PCIe lane or USB port and opt for a higher quality audio interface altogether at an additional cost. Some models also include PS/2 ports, which allow the use of older and more legacy peripherals, including keyboards and mice.
PCI Express x16, x4, x1, and PCIe 3.0/4.0
The term PCIe stands for PCI Express, with PCI itself abbreviated from peripheral component interconnect, which is an interface standard. This enables users to connect extra components to a system, including graphics cards. This also includes aftermarket sound cards, additional networking controllers, and even RAID cards for storagesavvy power users.
At present, there are two primary interface standards for this type of connector: PCIe 3.0 and PCIe 4.0. The maximum theoretical bandwidth of PCIe 3.0 equates to 32GB/s, with a frequency of 8.0GHz, while PCIe 4.0 doubles these figures to 64GBps and 16.0GHz. The benefits of PCIe 4.0 at the current time are limited, not by the technology itself, but by the shortage of devices that are PCIe 4.0 enabled to make use of this extra bandwidth.
One important point to note is that in multi graphics card configurations, such as AMD CrossFire and Nvidia SLI setups, to run more than one Nvidia GeForce graphics card in SLI, a full-length PCIe 3.0 slot that operates at a minimum of x8 is required, and even some chipsets such as AMD’s B550 just don’t have the capability to do this. For AMD CrossFire setups, a PCIe 3.0 x4 full-length slot is needed, which is usually operated from the chipset.
In regards to size, PCIe 3.0 and PCIe 4.0 x16, x8, and x4 slots are generally always either full-length or half-length on some models, with x1 slots the smallest of all. The type of slot needed for the type of expansion card usually comes down to what interface the card supports.
The only area that currently shows a real-world benefit comes through storage, and even genuine PCIe 4.0 drives are hard to source. One example is M.2 SSDs with PCIe Gen4 controllers and one of the most commonly used NVMe controllers from Phison, the E16 to be exact. This is identical to the Phison E13 controller bar PCIe 4.0 supports, which does give it an advantage over PCIe 3.0 M.2 SSDs, but not enough to saturate the bandwidth to its fullest capabilities. Phison has announced and previewed its E18 controller based on a 12 nm FFC process, which looks to further enhance the PCIe 4.0 interface.
Storage: M.2 to U.2, with some SATA on the side
Storage is an equally important part when considering your new motherboard, whether that’s a gaming system, something more for conventional office use, or even a megalithic rendering PC for content creation. For consumer desktop motherboards, there are three main types to consider, which boils down to two things: support from the CPU, and support from the chipset.
The SATA or Serial ATA connector has been the standard interface for storage for many years, which is a rectangle-shaped connector that is usually found in two flavours: right-angled and straight-angled format. This is the most commonly used storage interface, with support in regards to RAID arrays dependent on the motherboard chipset used. Intel typically supports RAID 0,1,5, and 10 arrays, while AMD only offers support for RAID 0, 1, and 10 officially on its desktop models.
The most common type of premium storage found on motherboards is M.2, with different types, such as the conventional SATA-based M.2 SSDs, or the intergalactic NVMe super-speed drives that operate from PCIe 3.0/4.0 x4/x2 slots. When upgrading a PC, one of the areas where users can see the biggest possible gain in real-world scenarios is going from a SATA 3.5-inch hard drive to a SATA 2.5-inch SSD; the difference in loading times is mind-blowing.
The M.2 interface further improves on this, but with a little bit of diminishing returns due to bandwidth.
But in regards to transfer speeds, PCIe 4.0 x4 M.2 drives do it best, with crazy speeds, bearing
in mind the limitation is the read and write speeds on the slowest drive in the transfer itself.
Another PCIe-based storage interface is U.2, which is similar to M.2 in many ways and shares similar support for PCIe, meaning more bandwidth and ultimately more performance. The biggest difference between M.2 and U.2 – aside from the shape of the connector, U.2 being a square shape and M.2 being small and flat – is that U.2 simply supports higher capacities of SSD, which is typically found on more professional-focused motherboards.
That being said, M.2 drive capacities are starting to increase, and U.2 doesn’t seem to have the buzz it once had. The M.2 form factor drives are also considerably cheaper, so it makes selecting one of the many M.2 drives on the market a no-brainer.
Now you’ve met the motherboard – what are the options?
Despite there being various differences in Intel and AMD’s CPU architecture and design, each corresponding socket release yields a new chipset, and thus offers a completely different feature set. The current generation of motherboards include a variety of different characteristics, including USB 3.2 G2 connectivity, with Thunderbolt 3 offering the most comprehensive and exclusive of all the connectors, though this is usually found on more premium models.
Finishing off our little biography is a table of current motherboard chipsets from Intel and AMD (within the last two years), which CPU architectures are supported, and what market they are aimed at. This journalist regularly overviews chipsets at launch, as well as in-depth motherboard reviews over at AnandTech (www. anandtech.com), so for more definitive guides on each chipset, feel free to check them out.
A chipset is not only part of the overall engine, but it’s the gateway between the processor and all of the board-integral componentry. One of the most important things to consider when building a new PC or graphics card-laden system is ensuring the processor and the chipset provide adequate features to fit the bill.
Leaving the more prosumer chipsets out of the equation. There are notable differences between the HEDT chipsets, such as Intel’s X299, and its desktop Z490 chipset. The biggest difference is that both AMD’s and Intel’s current HEDT platforms support quadchannel memory, while the non-HEDT enthusiast, budget, performance, and mainstream chipsets include support for dual-channel DDR4.