What is a Chipset? — Utmel
The chipset is a set of integrated circuit «chips» and is sold as a product. It is responsible for connecting the core microprocessor of the computer and other parts of the machine.
Catalog
Ⅰ Introduction
In the field of computers, the term «chipset» is usually a set of integrated circuit chips on a computer motherboard or expansion card. The manufacturer of chipsets can be independent of the manufacturer of the motherboard. For example, the PC motherboard chipset includes NVIDIA’s NFORCE chipset and the Viental KT880 of the VIA’s KT880, which is for AMD processors, or Intel chipsets.
Intel x58 North Bridge Chipset
Chipset is the core component of the motherboard. If the central processor (CPU) is the heart of the entire computer system, the chipset will be the torso of the entire body. The manufacturer of the design chipset in the computer is Core Logic. For the motherboard, the chipset almost determines the function of this motherboard, which in turn affects the performance of the entire computer system. The chipset is the soul of the motherboard. The advantages and disadvantages of the chipset performance determine the quality of the motherboard performance and the level of level. This is because the current CPU has a wide variety, and the features are different. If the chipset cannot work with the CPU, it will seriously affect the overall performance of the computer.
In general, the name of the chipset is the name of the North Bridge chip, such as the North Bridge chip of the AMDA75 chipset is a 75 Hudson D3 chipset, but also a North Bridge chip that supports the FX 8150 processor A75 series. The mainstream is: A75, A55, 990FX, 990X, 970, 870, 880G, 890GX, 890FX, 790GX, 785G, 780G, 770, 760G, E-APU, etc. The reason is that the Northbridge chip is the most important part of the dominant role in the motherboard chipset.
Ⅱ Function of Chipset
The motherboard chipset almost determines all the features of the motherboard.
The type of CPU, the system bus frequency, memory type, capacity and performance, and the graphics card slot specification are determined by the North Bridge chip in the chipset. The type and number of the expansion slot, the type, and the number of expansion interfaces (such as USB2.0 / 1.1, IEEE1394, serial port, parallel port, notebook VGA output interface), etc., is determined by the bridge of the chipset. Some chipsets are also determined by the display performance and audio playback performance of the computer system due to incorporating 3D accelerated display (integrated display chips), AC’97 sound decoding.
AMD chipset
The desktop chipset requires strong performance, good compatibility, interchangeability, and scalability. In the earliest notebook design, there is no separate notebook chipset, which uses the same chipset as the desktop. With the development of the technology, the appearance of the notebook’s special CPU has the adaptive notebook dedicated chipset.
The notebook chipset requires lower energy consumption, good stability, but the comprehensive performance and extension ability are also the lowest of the three.
The comprehensive performance and stability of the server/workstation chipset are the highest. Some products even require the full-year full load. It is also the highest in the supported memory capacity. It can support up to more than ten GB or even tens of GB of memory capacity. Its data transmission speed and data security requirements are the highest, so the storage device also uses the SCSI interface instead of the IDE interface, and the multiplier uses RAID to improve performance and ensure data security.
Ⅲ Chipset development
So far, manufacturers who can produce chipsets have Intel (USA), VIA (China Taiwan), SIS (Taiwan, China), AMD (USA), NVIDIA (USA), ATI (Canada), ServerWorks (USA), IBM (US), HP (US) is not many, in which Intel and NVIDIA and the VIA chipset are most common. On the Intel platform of the desktop, Intel’s own chipset has the largest market share, and the product line is complete.
Other chipset manufacturers are VIA, SIS, ULI. And latest joined manufacturers ATI and NVIDIA have added only a small market share. In addition to NVIDIA, other manufacturers outside NVIDIA are mainly in the low-end and integrated fields. NVIDIA only has medium and high-end products.
Samsung’s memory chipset
On the AMD platform, AMD itself usually played a pioneering role with a small market share. VIA previously occupying the largest market share of the AMD platform chipset. With its NFORCE2, NFORCE3, and the current NFORCE4 series chipset, it becomes the best chipset product of the AMD platform, which has won many market share from VIA, which has become the largest market for market occupation on the AMD platform. While SIS and ULI still play a supporting role, mainly in the middle, low-end and integrated fields.
In terms of notebooks, the Intel platform has absolute advantages. Intel’s own notebook chipsets also account for the largest market. Other manufacturers can only play aid angle and design products for AMD platforms with minimal market share.
On the server/workstation, the Intel platform is absolutely advantageous, and Intel’s own server/workstation chipset products occupy the most market share. In terms of high-end multi-channel servers based on Intel architectures, IBM and HP have absolute advantages, such as the IBM XA32 and HP F8. They are very excellent high-end multiple server chipset products. However, they are only applied to the company’s server products and are not too big. Because the market share of AMD server/workstation is small, it is mainly to adopt AMD home chipset products.
AMD chipset
The technology of chipsets is developing quickly these years: from ISA, PCI, AGP to PCI-Express, from ATA to SATA, ULTRA DMA technology, dual-channel memory technology, high-speed front-end bus, etc… Each new technology progress brings increased computence performance. In 2004, chipset technology will face major changes. The most striking is the PCI Express bus technology, which will replace PCI and AGP, greatly improve equipment bandwidth, resulting in a revolution of computer technology.
On the other hand, the chipset technology is also developing toward high integration, such as the internal memory controller. It greatly reduces the difficulty of designing products in chipset manufacturers. The current chipset products have been integrated audio, network, SATA, RAID, and other functions, which greatly reduce the cost of the user.
Ⅳ Chipset drive
Chipset drive is the motherboard driver. The driver is a special program that can communicate with the computer and the device. That is to say, the hardware interface can only control the work of the hardware device. If a device’s drive program failed to install it correctly, and it will not work properly. Therefore, the driver is the «hardware soul», «master of hardware», and «bridge between hardware and systems».
1 South-North Bridge Optimized Correction Procedure
2 motherboard integrated graphics card, sound card, network card, drive
3 hard drive
4 other motherboard related drivers
The motherboard chipset drive plays a crucial appointment in this communication.
It is mainly used to turn on the built-in functionality and characteristics of the motherboard chipset. And the motherboard driver is typically the motherboard identification that manages various peripheral device drivers or patches.
What motherboard chipset and socket do I need?
When you purchase through links on our site, we may earn an affiliate commission. Here’s how it works.
(Image credit: Future)
Motherboards and motherboard sockets are a subject that can make even the most enthusiastic PC gamer curl over with boredom. Still, it’s something you have to get right at the very start of a new build.
The array of motherboard choices out there can be bewildering; there’s one for every use, niche and sensibility, and these days they come in custom colors and patterns. The most important differences are in function, though, not aesthetics. Tiny, seemingly insignificant differences in model number hide huge disparities in performance potential and hardware compatibility.
So before you go wading through them all, there are a couple of things to narrow down. You have to understand the kind of socket you need, and know what form-factor you’re going to be building in. And deciding whether your rig is all about gaming, tweaking overclocked hardware, or productivity.
Motherboards don’t affect performance much themselves, but quality components enable your CPU to reach its full potential. That shiny new Ryzen 9 5900X can’t be overclocked without a motherboard that’ll let it, and an Intel Core i9 12900K Alder Lake chip simply wont fit into motherboards older than the Z690, with its LGA 1700 socket.
Once you’re done here, check out our best gaming motherboards feature to find the one for you.
Think about your build style and size
The key to navigating the motherboard maze is mapping where you want to take your system. It starts with size. How small does your computer need to be? When it comes to motherboards, bigger is usually better, roughly up to full-sized ATX.
Go with the biggest board your case can comfortably accommodate; don’t let the novelty of a small board tempt you unless absolutely necessary, or if novelty is part of the mission plan.
Why? Smaller boards cost more, provide fewer features and just aren’t as stable as big ones. Unless there’s a specific reason to go Mini-ITX, it’s better to avoid them for gaming.
Larger boards are easier to work with, provide better voltage regulation, and offer niceties like room for serious graphics cards, more slots for M.2 drives, and extra RAM capacity. You also avoid the skinned knuckles and high blood pressure inherent in every tight build.
For example, Mini-ITX boards that feature M.2 slots frequently put them on the backside of the motherboard, so you’ll need to disassemble your system to reach them or purchase an enclosure that has a cut-out specifically for this purpose.
The ‘bigger is better’ rule erodes for the largest motherboards, as prices for E-ATX and ATX-XL boards and the cases they require skyrocket.
Enclosure prices can more than double moving from mid- to full-sized towers, adding significantly to a system’s bottom line. Remember to include the hidden expense when buying and building beyond ATX.
The next step is listing all the things you need from a system. What kind of drives are you hooking up? Are you using Ethernet or wi-fi? Are you running more than one graphics card? How big is the CPU cooler? Any motherboard worthy of consideration should accommodate it all with to room to grow. It’s easy to be seduced into a high-priced boutique board only to find out the RAM slots are too close to the CPU socket for your cooler, or it has one less USB Type-C port than you need. When it comes to motherboards, features and stability trump performance claims.
Early in the build process, spend a little time thinking about the look and character of your upcoming PC. Do you want a flashy, LED-soaked screamer or a sleek and silent sleeper? Feel like building a show-it-all full-sized plexiglass tower, or an impossibly tiny silver desktop ingot? Once you have a blueprint in mind, you can start making hardware choices.
There are four common motherboard sizes which fit different sized cases:
Mini-ITX: The smallest, designed for very compact builds and cases. Typically has only a single PCIe slot, meaning you can only fit one graphics card.
MicroATX: Smaller than your standard motherboard, but with more features and expansion options than Mini-ITX boards.
ATX: The standard motherboard size that fits in most PC cases, with multiple PCIe slots, M.2 slots for SSDs, lots of SATA connectors, and other features. For any mid-tower or full tower case, this is the typical choice.
E-ATX: Extended motherboards are most commonly used for high-end CPUs, like Intel’s Extreme series. They’re more expensive and usually come packed with high-end features. Often requires a large full tower case to fit.
Our guide to the best PC cases has choices big and small, and tells you what size motherboards will fit inside.
Intel or AMD
If you’re still deciding on a processor, head on over to our best CPU for gaming guide where we break down the choices for every budget and interest.
While your pick here will influence the types of motherboard you can use, don’t worry about your choice locking you out of the latest features as it may have in years past. While fans of AMD or Intel may argue over processor supremacy, the motherboard situation is largely equal these days.
By now, you should have also selected the best graphics card for your needs. For gaming rigs, this is the single most important performance component, and the rest of the system should be built around running this GPU at the native resolution of your gaming monitor.
Motherboard chipsets explained
Chipsets provide the control logic required to make the components of a system work together, from CPUs to storage, and are responsible for the type and number of connectors available on a PC both inside and out.
Motherboards are identified by the chipsets they are based on, and these names change when major new CPU revisions are released, which happens frequently. That means that at any given time, a few generations of motherboards share the marketplace, adding to the confusion.
To help make sense of this, here are the current chipsets and the sockets they support.
Keep in mind that these are the standard specs for many, although you’ll want to double check for each individual manufacturer’s design, as they do vary.
Intel motherboard options
Intel 600 series
LGA 1700
Made to house 12th and 13th Gen Intel chips—Alder Lake and Raptor Lake respectively—Intel’s latest chipset instalments come in the form of Z690, H670, B660, and H610 motherboards. All of these allow some form of memory overclocking capability, and are all DDR5 compatible, though the H610 only comes with 1 memory channel—it and the B660 also have only 4 DMI 4.0 lanes, which is half that of the former two’s 8 lanes.
All chipsets in this series work with PCIe 5.0 for double the bandwidth of the previous generation, though the latter of the two has only one x16 PCIe slot, whereas the former also have two x8 slots.
If you plan on going for an enthusiast gaming build, the Z690 is the one for the CPU overclockers out there.
Intel 500 series
LGA 1200
Intel’s 10th Gen Ice Lake, and 11th Gen Tiger Lake CPUs, on the other hand, require LGA 1200 sockets. These include Z590, H570, B560, and H510 chipsets, with the Z590 made for CPU overclocking—you see the pattern emerging? The H510 doesn’t allow memory overclocking, either. Keep in mind that this series are PCIe 4.0 beasties, so they’re not touting the current 5.0 standard.
Intel 400 series
LGA 1151v2
Home to Intel’s 8th generation Coffee Lake and 9th generation Coffee Lake Refresh processors, the 300 series is getting on a bit now. The Z370 and Z390 chipsets, which support overclocking and Nvidia SLI, were most popular with enthusiasts with the mainstream h470 and B360 chipsets rounding out the gamer’s corner of these earlier motherboards. Intel also brought out its enthusiast X299 chipset, which is for its more extreme CPUs.
The X299 chipset uses the larger LGA 2066 socket, which accommodates the extra cores and other features found in the Extreme CPUs.
It was regarded as pricey in its day, but included plenty of lanes for M.2 SSDs, multiple GPUs, faster DDR4 support, more ports, and more.
AMD motherboard options
AM4 socket
AMD is a bit better about socket compatibility and motherboard longevity than Intel. Processor compatibility spans multiple product generations, so older Ryzen CPUs should work in later motherboards without a problem, although you’ll need to check the manual and update the BIOS to run newer processors on older chipsets.
Ryzen’s AM4 and the Threadripper’s TR4 sockets have been active for a long while now, and the 3rd gen Ryzen CPUs of 2019 even works on the 1st gen AM4 motherboards with a BIOS update.
AMD will be introducing it’s AM5 socket later this year though.
The current enthusiast chipsets for AMD’s sockets include the X370 and X470 for Ryzen, and Threadripper’s X399, TRX40 and the WRX80 for the Threadripper Pro.
Threadripper also sports 64 PCIe lanes and the core counts to back them up, making just about any combination of add-in cards and drives possible.
As with the X299 platform, however, for gaming purposes you’re almost always better off sticking with AM4. It’s not that Threadripper can’t play games, but it’s not faster than Ryzen and the extra cores aren’t put to use in any current games.
Budget choices
At the other end of the price spectrum, AMD offers its own inexpensive AM4 chipsets in the form of B550, B450, B350, A320, and A300 based motherboards. These offer fewer ports than the flagship X570 and X470, but the B550, B450 and B350 support overclocking and run the same default memory speeds as the high-end X370 and X470, making them a compelling alternative for Ryzen users looking to save a few bucks but keep some tweaking options open. A320/A300 boards meanwhile are pure budget offerings and lack many extras, and in most cases don’t cost much less than B350. We’d generally avoid A320/A300 unless you’re on the tightest of budgets.
If you don’t need absolutely top-of-the-line single-core CPU performance to satisfy your enthusiast urges, and if you’re using an older graphics card (basically Vega 64 / GTX 1080 or lower), you’ll find life with AMD pleasant and inherently more flexible.
Cheap and fun are a winning formula.
Learning about PCIe lanes
High speed I/O inside a motherboard is limited to the number of PCIe lanes available to the chipset and CPU. The main PCIe slots you plug graphics cards into support x16 speeds, though if you use multiple cards, that can drop the slot to x8 or even x4, depending on the motherboard and the CPU. The good news is you likely won’t need to worry about these complexities at all, unless you plan on an Nvidia SLI configuration with 3-4 graphics cards.
Standard Intel desktop processors have 16 lanes for one or two x16 PCIe slots, with more provided by the chipset. This seems like plenty until you remember one graphics card uses up to 16 lanes by itself.
Fortunately, modern, high-end chipsets like Intel’s Z690 provide enough to cover various configurations like dual GPUs or exotic M.2 configurations and a huge jump compared to the handful available during the Haswell era. Do note that the chipset to CPU link is only equivalent to four PCIe lanes, however, so it’s a potential bottleneck if you run multiple high-speed devices off the chipset lanes.
Intel’s X299 with Core i9 CPUs will give you 44 lanes on Skylake-X CPUs like the i9 7900X and newer i9 9900X and above, along with another 24 from the chipset. (Earlier i7 7800X and i7 7820X are limited to 28 PCIe lanes from the CPU.) Keep in mind that while the lanes may be available from the chipset, some boards only provide the physical connectors for a single x16 slot, running any additional full-length slots at x8 or slower.
AMD splits the difference this generation, with 24 total lanes on Ryzen (four for the chipset) and an amazing 128 PCIe 4.0 lanes on the Threadripper pro. Most AM4 CPUs provide four PCIe lanes for a dedicated M.2 slot, a nice advantage for the mainstream platform.
The main takeaway here is that the performance drop from x16 to x8 is unnoticeable in gaming and most workloads. If you really want a deep dive into PCIe, check out this video.
That covers the basics. Plan your build in terms of size as well as aesthetic, and then choose the chipset that will best fit the needs of your CPU and PC.
And now that we’ve covered the basics, you’re ready to go shopping.
Sign up to get the best content of the week, and great gaming deals, as picked by the editors.
Contact me with news and offers from other Future brandsReceive email from us on behalf of our trusted partners or sponsors
Screw sports, Katie would rather watch Intel, AMD and Nvidia go at it. Having been obsessed with computers and graphics for three long decades, she took Game Art and Design up to Masters level at uni, and has been demystifying tech and science—rather sarcastically—for two years since. She can be found admiring AI advancements, scrambling for scintillating Raspberry Pi projects, preaching cybersecurity awareness, sighing over semiconductors, and gawping at the latest GPU upgrades. She’s been heading the PCG Steam Deck content hike, while waiting patiently for her chance to upload her consciousness into the cloud.
Chipset (chipset) — what is it in the motherboard and what is it for, which is better, how to find out the chipset on a computer or laptop
The first computers occupied several rooms, now large companies have one main goal — to increase power reduce dimensions.
And the main element of this chain is the chipset.
What is a chipset?
Chipset (chipset, literally — a set of chips) is architecture of microcircuits, which are designed specifically to work with other components. The main purpose is to perform one or more functions.
The chipsets were most widely used on motherboards of personal computers or laptops. In these elements, it works as a connecting element , which is designed to ensure the high-quality operation of various memory systems or subsystems. The chipset also affects the CPU and input or output operations.
Also, similar elements are found in modern flagship phones . They are necessary to create proper capacities.
To put it simply, the chipset is chipset , which controls and organizes the operation and communication of various elements. It also determines the power of the elements in different performance modes.
Chipset functions
The main function is connection and optimization of operation of different components. The most common chipsets are found in the computer environment.
On the system board, he distributes responsibilities and manages the components. For example, the main chip interacts with RAM, input or output functions, adapters, and external device controllers. Connection in progress via bus system .
The main functions of the main chip include:
- Control and addition of subsystem elements. It regulates the allowed processors, memory sockets, video cards, expansion sockets, and motherboard ports.
- The frequency of the buses and the capacity of the system through which the interaction takes place.
- Monitoring the characteristics of controlled systems — increasing or decreasing individual indicators (for example, the processor clock frequency).
- Technologies that participate in the operation of the system. The most striking of these are dual video card systems — Cross Fire or SLI. Optimization of RAM — DUAL RAM and caching of information on high-speed SSD storage — Smart Response Technology.
- Support in “communication” with specific devices . This also includes retired controller types (PCI, AGP).
Various buses are responsible for the implementation of the message system. However, it should be noted that the main chip controls, but does not directly intervene in the process.
How to find out the motherboard chipset
Drivers are often already installed on a personal computer, so this process is much easier. It is enough to go to the device manager and select item system devices . The corresponding line will indicate the type of the main chip .
The second way is to read the information on the box to the device.
The name is right after the brand of the chip.
If for some reason there is no box, you can use the appropriate utility, for example AIDA64. It is enough to download, install and run the application.
Which chipset should I choose?
When this question arises, the user, in fact, buys not only the chipset as a separate element, but the motherboard as a whole. The main condition for the purchase is compatibility with other components. At the moment, there are two leaders that offer a wide range of products — AMD and Intel. It is recommended to select both the CPU and the motherboard at the same time.
For a budget configuration, chips of set 9 will be enough0007 H110 or H310 . For medium power devices, B150 or B250 motherboards are recommended. For computers with powerful configuration, it is recommended to choose Z270 or Z170 .
In general, it is impossible to say which chipset is better — all depends on the requirements of for the computer being assembled and the budget for purchasing the motherboard.
Structure and principle of operation
The standard device of the chipset contains two or more chips (bridges). The standard version is divided into northbridge and southbridge. North interacts directly with processors and memory. South is divided into two parts: through one side it is connected to the north bridge using the built-in bus, on the other side — peripheral controllers.
Each bridge is based on the controller hub .
The northbridge houses a memory hub with a CPU connection. It powers the CPU and RAM. Some systems add graphics adapters.
The Southbridge houses the input and output hub. Interaction with the central processor is carried out through the north bridge. Controls hard drives and external drives.
Location
The chipset is located in the middle of the motherboard.
On some devices, it may be hidden by a cooling grille.
Temperature
The optimal temperature of the chipset depends on its type . On average, the temperature can fluctuate 40 to 120 degrees . It all depends on the type of device. Detailed information is indicated in the instructions, for example, as in the picture below.
Cooling
Normally a conventional radiator is used for cooling. This is quite enough. But, if the motherboard overheats, then there are two main options for enhancing cooling.
The most effective is replacement for thermal paste . When buying new thermal paste, you should pay attention to thermal conductivity. The higher this figure, the better the paste.
The second method is to replace the grille . It is difficult to find radiators of the right size in stores. Therefore, it is recommended to order them from the manufacturer’s website.
What is a chipset? Analysis using the example of Intel TigerLake h55
Smartphones, tablets and other mobile gadgets have taught us this scheme: there is a chip on which everything should fit — the central processing unit, graphics, neural, all kinds of signal processors, Wi-Fi and so on.
This arrangement is called SoC or system on a chip, or single-chip system.
And this seems to be the most correct method of IC layout. After all, this is how we get the smallest delays, everything takes up less space, consumes less energy, and so on. But what is good for a mobile phone is not always good for a computer. Therefore, today we will talk about the gray cardinals of printed circuit boards — about chipsets.
What is it? And how many do you need to be happy? Let’s figure out how processors for computers are arranged. Let’s find where the bridges disappeared from the motherboard? And let’s find out how fast SSDs affected processors.
One chip is not always good
So why is one chip not always good? Unlike mobile phones, PCs and laptops are much more complex and versatile devices.
We expect that the processor in the computer will support any video cards and other pieces of iron. We expect to be able to connect a bunch of different peripherals to the computer: monitors, keyboards, flash drives, hard drives — and all this will work through different ports.
With mobile phones, everything is simpler and much more predictable. Therefore, mobile platforms are simpler in and of themselves. They are not required to support everything and everyone, so everything is on one crystal.
A vivid example of this is the Apple M1 chip, which grew out of a mobile platform. This is an excellent, tightly integrated single-chip system, but with a number of limitations. The entire RAM is soldered, and there are only two USB4 connectors and they do not work at full speed. As you understand, such a solution does not pull on universality.
Therefore, in PCs and laptops, instead of stuffing everything on one chip, it is much more reasonable to distribute key functions between several chips that will work together with the central processor.
Actually, a set of these additional helper chips is called a chipset.
Intel TigerLake h55
First, a little introduction. The reason for the roll was the release of new Intel processors for Tiger Lake-H laptops.
These are Intel’s first really powerful laptop processors in their best 10nm SuperFIN process to date.
The fact is that last year’s processors were already very good in terms of single-core performance. But they lost to AMD solutions in multi-threading, because they supported a maximum of 4 cores and 8 threads. But now Tiger Lake-H are 8-core processors, which are very powerful, judging by the tests we looked at on the net. This is not surprising: the successful architecture of the Willow Cove core, 24 MB of L3 cache, and the ability to overclock to 5 GHz in Turbo Boost mode for the older processor in the line.
In general, brutal processors, but in addition to power they have one more feature. A very interesting two-chip layout, which we want to talk about in more detail.
Where did this come from?
North and South Bridge
Previously, the most common among computer platforms was a three-chip solution: a central processor and two bridges.
You probably remember which ones: northbridge and southbridge. And for sure, you remember the pain when one of these bridges burned down. After all, in this case it was necessary to change the entire motherboard. After all, two controllers were called the north and south bridges, which were responsible for the operation of all components of the motherboard. Why were they needed?
The northbridge is connected directly to the processor and usually contained a memory controller, had direct access to a graphics card, or even had an integrated video adapter on board.
A south bridge was also connected to the north bridge. The southbridge was slower than the northbridge and less demanding peripherals were connected to it: USB, PCI, SATA buses, all I / O devices, Ethernet, audio, and so on.
PCH
But progress did not stand still, and in the process of miniaturization of components, the circuit was reduced to two chips. By distributing all the functions between a chip with a central processor and a chipset, which in Intel processors was called the “Platform Controller Hub” or PCH.
This second piece of silicon that is often flaunted next to the central processor is the PCH. Although it can be soldered in another place somewhere on the motherboard.
TigerLake h55
But the most interesting thing in this whole story is not that there were three chips, but there were two. And what functions do these chips now perform. Indeed, in recent years, the requirements for modern processors have become more complicated and have changed a lot. And this is largely due to the console of the new generation.
Here’s a look at how the new Intel processors for powerful Tiger Lake H laptops are arranged.
There are as many as 20 PCI-e 4.
0 lanes directly connected to the CPU, which provide a throughput of just over 39 GB / s. And that’s a lot. In terms of speed, this is equal to 40 PCI-e 3.0 lanes. And it’s much larger than the new AMD Ryzen 5000 series, which only has 16 PCI-e 3.0 lanes. But why do we need such speed?
“To unleash the full potential of graphics cards. Top-end Nvidia RTXs require PCI-E 4.0 so that the bandwidth of the connection interface does not become a bottleneck”
First of all, to implement two very important technologies: Microsoft DirectStorage and Nvidia’s Resizable Bar.
DirectStorage technology can significantly speed up I/O operations from fast SSD drives. Roughly speaking, thanks to this technology, it becomes possible to directly load data from SSDs into the processor, bypassing RAM and using an SSD instead of RAM.
The technology was originally introduced as part of the Xbox Velocity architecture for new consoles. It allows you to maximize performance throughout the entire pipeline from NVMe disk to GPU and is primarily needed for lightning-fast game loading and on-the-fly loading of game assets.
But in the near future, the DirectStorage API will become part of Windows, which will allow you to download not only games on Windows, but also download Windows itself, and indeed any software that will support this API.
DirectStorage is closely related to another technology — Nvidia’s Resizable Bar, which allows the processor to access the entire amount of video memory, providing more efficient data exchange between the CPU and the video card.
In other words, these technologies will allow you to rotate data between SSD, CPU and GPU with maximum efficiency, turning this whole bundle into a single organism.
And the new Intel TigerLake H-series gives you a lot of freedom in this regard. Notebook manufacturers can come up with any configuration for different tasks. They can allocate 16 PCI-e 4.0 lanes for a discrete graphics card and leave 4 lanes for NVMe SSD. Or go the other way — leave 8 lines for the video card, and divide the remaining 8 lines between two NVMe SSDs in a RAID 0 array, which will provide simply prohibitive speeds of the SSD.
Let’s check how it works in practice right now. I have an ASUS ROG Zephyrus M16 in an Intel i7-11800H configuration for testing purposes. This is just an 8-core Tiger Lake-P. There is also 16GB of RAM, a 1TB SSD.
- Display 16″ WQXGA 165Hz
- Intel i7-11800H
- GeForce RTX™ 3060 6G
- RAM 16G
- 1T SSD
The speeds here are really very high. But if desired, this figure can be doubled in a laptop by adding a second SSD to a free slot and combining them into RAID 0. New consoles never even dreamed of such speeds.
Well, the most unexpected thing in the new Intel TigerLake: 36 more PCI-e 3.0 lines are routed to the second PCH chip, 12 of them are allocated for USB and 24 are free. This allows you to implement generally free connection of any peripherals at all. In other words, in laptops with Tiger Lake-H, you can at least stick all the free space on the case with different ports. You can simultaneously plug in four Thunderbolt 4 ports, 10 USB 2.