Modern chipsets for Intel processors. Intel products: chipsets and their features

Key Trends and a Brief Description of Six Semiconductor Variations on the Same Topic

We have already managed to get acquainted with some motherboards for the new Intel LGA1150 platform, and with new processors too. However, the chipsets have not yet been considered in detail. What is not entirely correct - you will have to "live" with them for a long time: at least two generations of processors. Moreover, in the new series, Intel approached the issue of platform rework in a rather radical way - if the seventh series was only a small revision of the sixth and existed in parallel with it (the budget H61 did not receive a successor at all) within the same LGA1155 platform, and the sixth inherited its features from the fifth, the eighth was designed almost from scratch. Not in the sense that it has nothing at all to do with previous products - in fact, it is still the same south bridge, in terms of basic functionality comparable to the "peripheral" hub of very old chipsets and interacting with the north bridge (which is already in the processor) via DMI 2.0 buses (the same as in 1155/2011) and FDI (the interface debuted in the fifth series of chipsets and is used to connect displays). But the logic of work has changed. And so are the peripheral interfaces. So it's time to talk about all this in more detail.

Quarter FDI ...

Let's start with the Flexible Display Interface, which, as we have already said, appeared in the LGA1156 framework. But not right away - the P55 chipset did not have this interface: it debuted in the H55 and H57, released simultaneously with processors with an integrated video core, since others are not needed. That within the framework of this, that within the framework of the subsequent platform, it was the only way to use the integrated GPU. Moreover, Intel also had a P67 chipset with a blocked FDI, which did not allow to install video outputs on boards on it. However, the company later abandoned this approach. The only difficulty left is the connection of a large number of high-resolution displays. More precisely, as long as we were talking about two digital sources of images and resolutions not higher than Full HD, everything was fine. As soon as attempts began to get out of this framework, problems immediately began. In particular, the fact that it is impossible to find a board with support for 4K on HDMI directly hints that it is not the manufacturers of the latter who have been smart;) Yes, Intel is promoting DisplayPort, which does not require royalties for use, but in consumer electronics, it’s not on fire during the day. you will find. And the appearance of the third video output in Ivy Bridge actually turned out to be a theoretical advantage of the new GPU line: it quickly became clear that it can only be used on boards with at least a couple of DPs. This was actually done only in the case of expensive models with Thunderbolt support.

What has changed in the eighth generation? The FDI has shrunk from eight to two lines, just like the title says. The explanation is simple - following the example of AMD APUs, all digital outputs (up to three) are transferred directly to the processor, and the chipset is now responsible only for analog VGA. Thus, if the latter is abandoned, the layout of the board is greatly simplified already at the stage of the "processor-chipset" connection. Of course, work around the socket becomes a little more complicated, but not much, if you do not require records from the board. For example, in ASUS Gryphon Z87 the manufacturer limited itself to two video outputs, which will be enough for many, since one of them is "standard" DVI, but the second is HDMI 1.4 with a maximum resolution of 4096 x 2160 @ 24 Hz or 2560 x 1600 @ 60 Hz. Or you can go for a record - as in the Gigabyte G1.Sniper 5, where there are two such outputs plus DisplayPort 1.2 (up to 3840x2160 @ 60 Hz) was added. Moreover, all three can be used simultaneously. Or it is possible not at the same time - for example, connect a pair of high-resolution monitors to HDMI. It is clear that suitable models are universally equipped with DP, and it is in them that HDMI may no longer be found, however ... see above about previous generations: most motherboards would not have supported two high-resolution monitors at all. It was possible to connect them to a computer only using a discrete video card, which is not always convenient, and sometimes impossible. Systems based on Haswell are forced to resort to discrete graphics only in cases of going beyond the needs of mass users: if you need the maximum performance of the graphics subsystem (in a gaming computer), or when you need strictly more than three monitors.

In general, purists who advocate that processors should be processors, and everything else from the evil one, may once again be indignant at the fact that more and more functions of the north bridge are being transferred under the CPU cover - let them. From a practical point of view, it is more important that the previously integrated video had, let's say, not always sufficient peripheral capabilities. The new is in many respects a groundwork for the future - it is clear that now no one will connect three 4K TVs (or, at least, high-resolution monitors) to a computer, and if it does, it is unlikely that they will use an integrated GPU. However, this at least became possible. And in the future, in terms of video support, the situation will not worsen, but it can already come in handy. In addition, this approach of the company, in fact, pushes manufacturers to completely abandon the analog interface. Which "healed" on the market to a large extent precisely because of Intel's early policy regarding video outputs: even in the fourth series of chipsets it was easier just to be limited to "analog", but "digital" required additional gestures. Now, on the contrary, which will obviously affect both motherboards and monitors: their manufacturers will no longer be able to nod that VGA is the most widespread.

By the way, one of the reasons why we started with FDI: this change makes new processors completely incompatible with old platforms, where video outputs were connected to the chipset. What is always worth remembering for those who decide to complain about a socket change. It is clear that for the sake of this alone, Intel would hardly go to the overdue, but radical overhaul of the platform, however, along with a change in the approach to power (integrated VRM and single circuits for both processor and graphics cores, as opposed to separate circuits previous generations) there were enough potential benefits. Actually, all of them lead to the fact that, despite the use of the same DMI 2.0, the platforms have become fundamentally incompatible with each other. But the possibility of using the PCH of the eighth series in the updated version of the LGA2011 platform (if it is deemed necessary) has been preserved: one interface is sufficient there, and FDI is not used.

... and PCI by-by

The PCI bus appeared more than 20 years ago, and all these years has served faithfully to computer users, first as a high-speed internal interface, and then simply as an interface. The historical aspect, we already, now just say that since the publication of this material PCI has become completely and irrevocably obsolete, but it is still often used. Another question is that its presence in chipsets has already become an anachronism - the layout of parallel buses is inconvenient, since the number of contacts of a relatively small chip increases sharply. Those. it is easier for motherboard manufacturers to use additional bridges even in motherboards based on PCI chipsets.

Why did PCIe-PCI bridges come to market at all? This is due to the fact that Intel has gradually begun to remove support for the second bus from its products already within the framework of the sixth series. More precisely, the PCI controller itself was physically in the chips, but its contacts were brought out only in half of the packaged microcircuits. The main line of the section was the positioning of the latter - in the business series (B65, Q65 and Q67, as well as their heirs of the seventh series) and the extreme X79, there was "innate" support for PCI, but in solutions aimed at the mass desktop segment and designed for mobile computers blocked. It seems to us that such a half-hearted decision was made because the company itself could not decide whether to “finish off” PCI or it was too early. It turned out that it was just right :) The dissatisfied, of course, were all the same, but mostly theoretically dissatisfied. In practice, many dispensed with PCI slots altogether, and some were quite satisfied with bridges. In general, the company did not have to do an urgent refresh of the chipset line, returning PCI to its place. Therefore, the eighth series of chipsets does not support this bus either de jure or de facto. Thus, the process of transition from PCI / AGP to PCIe, begun back in 2004, has come to its logical conclusion; ended, simply put. This is even noted in the names of microcircuits: for the first time since the notorious i915P and its relatives, there is no word “Express” there - just “Chipset”. What is logical - it makes no sense to emphasize support for the PCIe interface in an environment where there is only one. And very symbolic;)

Let us emphasize just in case (especially for the most timid) that there is no PCI support in chipsets, and not on motherboards - the latter can provide the user with a couple of PCIs in the usual way: using a PCIe-PCI bridge. And many manufacturers do it - including Intel itself. So if someone has a dear scarf lying around as a memory of his youth, it’s still easy to find where to stick it. Even when buying a computer on the latest platform.

SATA600 and USB 3.0 - the same and more

Six SATA ports appeared in the ICH9R south bridges as part of the third series chipsets (and, formally, the "fourth" X48), while the weaker ICH9 was limited to four. Within the framework of the fourth family, this injustice was eliminated - the ICH10 still did not support RAID, but it was also given six SATA. This scheme migrated to the fifth series without changes, while the sixth brought support for the faster SATA600 to Intel chipsets. But limited - the older models received two high-speed ports, the junior "business" B65 limited itself to one, and the budget H61 was cheated on all fronts: only four SATA300 ports and nothing else. In the seventh episode, nothing has changed. In general, the decision with a limited number of ports was logical: since some (and not always great) gain from SATA600 can only be gained by solid-state drives, but not hard drives, in budget systems it is still not needed at all. Yes, and in non-budgetary one or two ports is enough, especially since a larger number of high-speed devices will not be able to fully work at the same time, because DMI 2.0 has a limited bandwidth, however ...

However, AMD support for SATA600 was not only implemented almost a year earlier, but also in the number of all six ports. Of course, their simultaneous operation at full speed has never been discussed either - the bandwidth is that Alink Express III (the bus connecting the north and south bridge of AMD 800 and 900 series chipsets), that UMI (provides communication between FCH and APU on FM1 / FM2 platforms ) that DMI 2.0 is exactly the same, since the whole trio is a slightly redesigned electrically PCIe 2.0 x4. But such a solution was more convenient - if only because when assembling the system, you did not need to think about where to connect which drive. Moreover, it's easier to advertise - six ports sounds much better than two. And recently there were eight of them in the A85X.

In general, Intel decided not to put up with this state of affairs, and to increase the number of ports. True, they approached the question in their own way: there are two SATA controllers left, as in the previous families. But the one responsible for the SATA600 is now capable of connecting up to six devices out of six possible. Smaller than AMD's still, but comfortable too. And the total speed, as mentioned above, remains the same, so that quantity can be converted to quality not earlier than the inter-hub interface changes. And something tells us that this will not happen soon - until that moment, SATA Express will probably be able to try hard, which will make the bandwidth of SATA itself insignificant.

As for USB 3.0, initially Intel was generally lukewarm about the new interface. Later, the company realized it, and in the seventh series of chipsets, an xHCI controller with support for four Super Speed ​​ports appeared. And in the eighth, this part of the chipset was radically redesigned. First, the maximum number of ports has been increased to six, which is more than AMD's, so all mainboard manufacturers have already sent out victorious press releases on this topic. Many, however, did not calm down on this, and continue to "mold" discrete controllers or hubs into their products, bringing the number of ports to eight or even ten. To be honest, we see no more practical use in this than in six chipset ports, since no user will have a dozen USB 3.0 devices, and for a long time. Those. here are four ports - necessary and sufficient: a couple on the back panel, a couple more in the form of a comb to bring it to the "face" of the system unit, and where more? In laptops, there are often three ports in total. So it goes.

But, in general, there are more ports, which is only the upper part of the iceberg. The underwater one can turn out to be unpleasant - there is only one USB controller in the new chipsets. Why is that bad? Intel is nothing: the microcircuit has been simplified. PCB manufacturers also do nothing: the wiring is easier, because, in fact, it doesn't matter which legs to pull from. But for users ... Firstly, the older chipsets had not one, but two independent EHCI controllers, which theoretically could provide a higher speed of the “outdated” High Speed ​​peripherals with the simultaneous use of several devices. Secondly, this pair of controllers has not changed for many years, so it was perfectly "understood" by all more or less relevant operating systems without installing additional drivers. Under Windows XP, one, however, was needed, but under this OS all 14 ports worked (or fewer in low-end chipsets, but all physically present) - albeit only as USB 2.0. And for a new controller, you need to install a driver (in laptop SoCs, USB ports do not want to work without it at all), and it exists only for Windows 7/8 (maybe it can be “screwed” to Vista, but this is no longer very interesting) ... It is clear that support for Windows XP has long been anathema on the part of Microsoft, so Intel does not bother with it too much (it's not for nothing that they did not implement the full-fledged operation of USB 3.0 in the seventh series, although some discrete controllers fully work even under Windows 98) and not only This concerns USB, however, fans of the "old lady" will not be envied. It is easier for Linux fans and users of different LiveCDs based on these systems, although they will also need an update, but for the old scheme it was not required. In general, on the one hand - it is better, on the other - some habits will have to be changed.

Simpler and more compact

So, as we can see, the new chipsets have become more primitive in some respects than their predecessors. Support for video outputs has almost completely "moved" into the processor, there is no PCI controller, instead of three (in fact) USB controllers, there is only one, etc. However, if we compare consumer characteristics (the same number of ports of high-speed interfaces), we can see clear progress. And what about the physical parameters of the microcircuits themselves? All is well, since an active redesign was also needed to transfer the chips to new production standards. The fact is that, as the assortment of processors to 22 nm is more and more active, Intel began to release production lines designed for 32 nm, to which it was decided to transfer chipsets. Considering that previously the "standard" was the use of norms as much as 65 nm, the jump is impressive.

So, let's recall the top-end Z77 Express: a 27 x 27 mm chip with a TDP of up to 6.7 W. It seems to be a little, so it would be possible not to touch it. But the Z87 fits into 23 x 22 mm. It is more clear to compare the areas: 729 and 506 mm 2, i.e. from one plate you can get 40% more new chips than old ones. And the number of contacts has decreased, which also reduces the cost. And the maximum possible thermal package decreased even more significantly - up to 4.1 W. And if the first is relevant only for Intel itself (while maintaining the same prices for chipsets and without the need to modify the process of their production, you can earn much more) and a little for other manufacturers, then the second can be useful for end users as well. Not for buyers of motherboards based on Z87, of course, where no one will notice these 2.6 W (and manufacturers will gladly stick a fanciful cooler with a heat pipe on this too - don't go to a fortuneteller). But after all, similar changes apply to all chipsets, but in laptops and other compact systems, reducing heat dissipation will not hurt at least. And a decrease in the linear dimensions, coupled with a simplification of the wiring, will also not be superfluous: in this segment, they often struggle for every millimeter. Comparison of mobile HM77 Express and HM87 is no less significant: 25 x 25 mm and 4.1 W versus 20 x 20 mm and 2.7 W, i.e. the size was reduced even more than among the desktop modifications, and even with efficiency it was possible to squeeze out at least something (despite the fact that great importance was attached to it before). In general, in terms of increasing the consumer attractiveness of the platform as a whole, the chosen course can only be welcomed. And it is not known - it would have been possible without it to develop a SoC with "full" characteristics. For example, something like the Core i7-4500U, where everything that was left uncut during the development of standard component systems was cut, but the chip turned out to be less than 1000 mm 2 and with a full TDP of 15 W. In the very first implementation of the U-series, two chips were required (and, as I recall, we already focused on the fact that the processor is smaller than the chipset), and they needed more than 20 watts for a couple. Trifle? The tablet is not a trifle. And in the desktop, such improvements were not vital - for him they turned out to be a side effect.

Intel Z87

Well, now let's get acquainted in a little more detail with specific implementations of new ideas - both already supplied and predicted. Let's start, traditionally, with the top model, giving both a typical scheme and a list of basic functionalities:

• support for all processors on the Haswell core (LGA1150) when connected to these processors via the DMI 2.0 bus (with a bandwidth of 4 GB / s);
• FDI interface for receiving a fully rendered screen image from the processor and a block for outputting this image to a display device with an analog interface;
• support for simultaneous and / or switchable operation of the integrated video core and discrete GPU (s);
• increasing the frequency of processor cores, memory and integrated GPU;
• up to 8 PCIe 2.0 x1 ports;
• 6 SATA600 ports with support for AHCI mode and functions like NCQ, with the possibility of individual disconnection, with support for eSATA and port splitters;
• the ability to organize a RAID array of levels 0, 1, 0 + 1 (10) and 5 with the Matrix RAID function (one set of disks can be used in several RAID modes at once - for example, on two disks you can organize RAID 0 and RAID 1, for each array its own part of the disk will be allocated);
• support for Smart Response, Rapid Start, etc .;
• 14 USB ports (of which - up to 6 USB 3.0) with the possibility of individual disconnection;
• Gigabit Ethernet MAC controller and special interface (LCI / GLCI) for connecting a PHY controller (i82579 for Gigabit Ethernet, i82562 for Fast Ethernet);
• High Definition Audio (7.1);
• strapping for low-speed and outdated peripherals, etc.

In general, everything is very similar to the Z77 Express, with the exception of some points, most of which were described above. There are only two things left behind the scenes. Firstly, as we can see, the possibility of splitting the "processor" PCIe 3.0 interface into three devices has not disappeared, but any mention of Thunderbolt has disappeared - on the contrary: the diagram clearly says "Graphics". Thus, we will not be surprised when faced with motherboards that implement three "long" slots without any bridges. The second change concerns the approach to overclocking. More precisely, there are two changes. On the LGA1155 platform, it was possible to have fun with the multiplier of non-K-series quad-core processors - now Limited Unlocked is dead. But overclocking on the bus returned in a form similar to LGA2011: before feeding it to the processor, the reference frequency can be increased by 1.25 or 1.66 times. Unfortunately, our initial optimism about this information has not yet stood up to practical tests - this mechanism does not work with processors other than the K-series. In any case, for the three motherboards we have already tested on the Z87, this is true, so you can, of course, continue to hope and believe that these are all flaws in earlier versions of firmware, but ...

Intel H87

Unlike the sixth and seventh families, there are no intermediate chipsets between the top-end and mainstream solutions. And the differences between them have become less - in fact, only splitting of 16 "processor" lines is absent, so there is nowhere to "cram" an analogue of some Z75 (especially, this chipset has remained largely a virtual product, unclaimed by manufacturers boards). Even in terms of their attitude to overclocking, the chipsets are close: there are no bus modifiers, but they are generally useless on the Z87, and the multiplier on some Core i7-4770K is not forbidden to “twist” on motherboards based on H87. Moreover, the latest chipset also has some advantage over its more eminent relative, namely, support for Small Business Advantage technology inherited from the business line of the seventh series. However, it does not work out to consider it an unambiguous advantage for a "single enthusiast" (if only because these very "enthusiasts" of the SBA do not discuss too much), and where it is needed, it is often the business line of chipsets that have been and are being used ... But the fact of expanding its scope of application is indicative. You look, in time we will inherit something else.

Intel H81

This chipset has not yet been announced, but with a high degree of probability it will appear no later than inexpensive processors for LGA1150. Moreover, after being published, it can become quite popular among expensive buyers, since a new budget solution is able to cover 80% of user requests. At the same time, it is still budget, which allows you to hope for motherboards of $ 50 at retail. Why so cheap? The H61 inherited a bunch of limitations that can bring a real enthusiast to a nervous fit: one memory module per channel (that is, only two full slots), six (not eight) PCIe x1, four SATA ports without any RAID "s and others. bourgeois excesses, 10 USB ports. the H61 was not enough, although, we repeat, the chipset has not been officially announced yet, so most of the information about it is rumors and leaks, but they are quite plausible.

Business line: B85, Q85 and Q87

Let's go over these models briefly, since most buyers are not interested in them. The B75 was an extremely attractive chipset for LGA1155, but mainly because the H61s were too disfigured to make them cheaper and were not updated within the seventh series. As we can see, the H81 will support new interfaces (albeit in a limited number due to positioning), so the B85 has only quantitative advantages over it: +2 USB 3.0, +2 SATA600 and +2 PCIe x1. True, there is not so much benefit from increasing the number as from the very presence of these interfaces, and the price is higher, so you can already swing on the board on the H87, since there is even more there, and SBA support is also there. Again, built-in PCI support was an exclusive feature of the "old" business series, often turning into a significant advantage, but now there is nothing left of it.

Here is Q87 - the chipset is traditionally unique, since it is the only one from the entire line that supports VT-d and vPro. The rest is almost identical to the H87. And Q85 is a strange thing, which occupies an almost intermediate position between H87 and B85: the main difference is the optional support for AMT in Q85. Why is he so needed - do not ask. There is a suspicion that Intel is developing the Qx5 line more "just in case", since there are not too many motherboards on such models, and not only on the open market. At least not comparable to the Qx7. And in our area, “business solutions” most often mean not even the B-series, but something on the youngest chipset of the line (earlier G41, later H61, then, apparently, H81 will take this place), which is logical - the same SBA , in principle, it could be useful in a small office, but its implementation still requires at least Core i3, and not the popular Celeron in such offices. In general, for greater beauty and in order to increase the general education, we present diagrams of systems based on these three chipsets.

But, again, the probability of meeting the majority of our readers with them is close to zero. Except, perhaps, for the Q87, since VT-d is of interest not only in the corporate market, and no other chipset can boast full support for this technology. In any case, officially - unofficially, some motherboards based on the Z77 supported it, so this is certainly possible with the Z87. True, earlier sometimes attempts to use such genetic engineering products did not always end successfully, so in order to avoid problems and save time, it is easier to immediately focus on Qx7 (especially now, when processors with VT-d support can’t be overclocked anyway, but amenable K-series tuning did not support I / O virtualization either).

Total

	Z87	H87	H81	B85	Q85	Q87
Tires
PCIe 3.0 Configurations (CPU)	x16 / x8 + x8 / x8 + x4 + x4	x16	x16	x16	x16	x16
PCIe 2.0 quantity	8	8	6	8	8	8
PCI	Not	Not	Not	Not	Not	Not
Overclocking
Cpu	Multiplier / Bus	Factor	Not	Not	Not	Not
Memory	Yes	Not	Not	Not	Not	Not
GPU	Yes	Yes	Yes	Yes	Yes	Yes
SATA
# Of ports	6	6	4	6	6	6
Of which SATA600	6	6	2	4	4	6
AHCI	Yes	Yes	Yes	Yes	Yes	Yes
RAID	Yes	Yes	Not	Not	Not	Yes
Smart response	Yes	Yes	Not	Not	Not	Yes
Other
# Of USB ports	14	14	10	12	14	14
Of which USB 3.0	6	6	2	4	6	6
TXT / vPro	Not	Not	Not	Not	Not	Yes
Intel Standard Manageability	Not	Not	Not	Not	Yes	Yes

If we consider processors for LGA1150 as an isolated product, then they do not have any significant advantages over their predecessors in terms of consumer characteristics, which we have already written about. As you can see, this also applies to chipsets to the same extent: some things have gotten better, some have simply become larger, but the implementation of some things was more interesting earlier. On the other hand, there is practically no separate market for processors and chipsets in the form in which it existed 15-20 years ago: manufacturers are actively and aggressively selling "platforms" in the form of complete (laptops and other portable) and semi-finished solutions (desktop computers). Accordingly, when developing both processors and chipsets, you can not think about some kind of global compatibility, simply "adjusting" one to the other and transferring an increasing part of the functionality directly to the processor (all the same, they have to be produced according to fine norms, so this is economically justified , and the rejection of "long" lines of high-speed buses also simplifies the creation of a finished product). As a result, we have what we have: FDI and DMI 2.0 are still used to connect the processor and the chipset, but neither new processors can be combined with old motherboards, nor vice versa. Theoretically, it is possible to "screw" the same Z87, abandoning video outputs, to LGA1155, but it will still be a new board. Well, the reverse procedure does not make sense at all.

In general, if anyone is going to buy a fourth-generation Core, they will definitely have to buy a board based on one of the eighth series chipsets. All freedom of choice is limited only by a specific model. Which one? It seems to us that out of the entire six chipsets, only half of the models are interesting: Z87 (top solution for entertainment), Q87 (no less top-end chipset for work needs) and the forthcoming H81 (cheap, but enough for many). Intermediate models, as practice shows, are in much more limited demand from individual buyers, simply because the contribution of the cost of the chipset to the price of the motherboard is noticeable only in the budget segment (but every dollar is saved there), but quickly disappears in the models , with a retail price in the region of a hundred. So, perhaps, a more correct approach on the part of Intel would be to stop pretending to be the illusion of choice altogether, and to release only a couple of models: expensive (where everything is available) and cheap (where there is only an absolute minimum). On the other hand, it will not be possible to develop a hundred motherboards in a line on just two chipsets (which manufacturers who focus on the retail market of components simply adore), so we will have less work on describing all these twists of engineering and marketing thought, and users of various near-computer forums will become there is nothing to discuss, so let everything remain as it was for now.

This article will examine and describe in detail the chipsets produced by Intel for the latest generations of processors from this manufacturer. Also, recommendations will be given regarding the choice of logic of the motherboard when assembling a new computer system.

What is a "chipset"?

Behind the word "chipset" is the chipset that is installed on the motherboard. It connects together the various components of a computer system. Its second name is system logic. As a rule, it is bound to a specific socket, that is, a processor socket. This article will consider the most relevant solutions from Intel, which can still be found on sale.

Sandy Bridge and Series 6 Chipsets

The most "ancient" of those produced, which can still be found on sale now, belong to the 6th series. Their announcement took place at the beginning of 2011, and any CPU of the Sandy Bridge and Ivy Bridge families can be installed in them. If you are installing a second family of CPUs, you may need to. All of these chips were installed in and were often equipped with an integrated graphics solution. Another important feature of this platform was that it consisted of only one microcircuit - the "south bridge". But the "north bridge" was integrated into the processor. The most affordable of these was the chipset. It made it possible to create inexpensive office systems. Also, on its basis, it was possible to make a good PC for study. But the ligaments "Kor Ay5" or "Kor Ay7" and "N61" look absolutely ridiculous. It is foolish to install a high-performance processor in a MiniATX motherboard with minimal functionality. This chipset allowed installing only 2 RAM modules, equipped with one PCI-Express 16x v2.0 slot for installing an external graphics accelerator and had 10 YUSB 3.0 ports and 4 SATA ports for connecting hard drives or an optical drive.

The middle segment was occupied by Q65, B65, Q67 (these chipsets did not support Ivy Bridge chips). The difference between them and the H61 was the number of slots for RAM (in this case there were 4 instead of 2) and ports for drives (5 versus 4). Initially, the H67 and P67 were used for the most productive. The first of them supported integrated video, but was equipped with only one slot for installing an external graphics accelerator. And the second was aimed only at use (it had 2 slots for these purposes), but the built-in graphics accelerator on such motherboards did not work. In turn, the solutions based on the Z68 combined the best aspects of the H67 and P67. This particular chipset can be considered the best for this platform.

"Ivy Bridge" and motherboards for them

A new generation of Ivy Bridge CPUs came in 2012 to replace Sandy Bridge. There were no fundamental differences between these generations of chips. The only thing that has essentially changed is the technological process. The previous generation of processors was manufactured using 32nm technology, and the new one - using 22nm process technology. The socket for these chips was the same - 1155. Entry-level systems in this case were also built on the basis of the Intel H61 chipset, which perfectly supported both generations of semiconductor crystals. But the middle and premium segments in this case have changed significantly. Although the characteristics of the Intel7 series chipsets indicate that they practically did not differ from their predecessors. The mid-range solutions in this case included B75, Q75, Q77 and H77. All of them were equipped with 1 slot for a video card and had 4 slots for installing RAM. The B75 has the most modest parameters: 5 SATA 2.0 ports and 1 SATA 3.0 port for organizing a disk subsystem and 8 YUSB 2.0 ports and 4 YUSB 3.0 ports. By the way, all chipsets of the 7th series could boast of exactly this amount of YUSB 3.0. The Q75 differed from the B75 only in the number of YUSB 2.0 ports, of which in this case there were already 10 instead of 8. The H77 and Q77, unlike the Q75 and B75, could boast of having two SATA 3.0 ports. The premium segment in this case was represented by the Z75 and Z77. If the previous four chipsets only allowed overclocking the CPU and the graphics accelerator, then these two semiconductor crystals could also increase the RAM frequency. Also in this case, the number of slots for video cards increased. There were 2 of them in solutions based on the Z75, and 3 in the Z77.

Haswell, Haswell Refresh and its system logics

In 2013, 1150 replaced it. Its processors did not make any revolutionary changes. The only exception in this regard was the power consumption of chips, which in this particular family of CPUs was significantly redesigned and this made it possible, without changing the technological process, to significantly reduce the thermal package of semiconductor crystals. New sets of system logic were released for the new socket. Their parameters have a lot in common with the previous generation of the 7th series. There were 6 chipsets in total: H81, B85, Q85, Q87, P87 and Z87. The most modest in terms of parameters was the H81. It only has 2 RAM slots, 2 SATA 3.0 ports, 2 SATA 2.0 ports and 1 graphics card slot. Also, the number of YUSB 2.0 and 3.0 ports, respectively, was 8 and 2. Motherboards based on this set of system logic, as a rule, were equipped with Seleron and Pentium chips. The Intel B85 chipset differed from the H81 in the increased number of RAM slots (there were already 4 of them), USB 3.0 and SATA 3.0 ports (4 in both cases versus 2). The Q85 could, in comparison with the B85, boast only 10 USB ports of version 2.0. These two chipsets are most often used in conjunction with Kor Ay3 chips. The characteristics of the Q87, P87 and Z87 are identical. They have 4 slots for RAM, 8 YUSB 2.0 ports, 6 YUSB 3.0 ports and 6 SATA 3.0 ports. The Q87 and P87 chipsets were perfect for Kor Ay5 and Kor Ay7 with locked multipliers. But the Z87 was focused on chips with the "K" index, that is, on its basis computer systems were built to overclock the CPU.

Broadwell and chipsets for it

In 2014, the Haswell generation was replaced by new chips codenamed "Broadwell". They are manufactured using a new 14 nm manufacturing process and are not fully compatible with 8 series logic kits. The number of processors themselves was released a little and, as a result, there was no specific update of the chipsets. Only 2 of them were released - H97 and Z97. The first of them was intended for a CPU with a locked multiplier and completely repeated the parameters of P87. Well, the Intel Z97 chipset was an exact copy of the Z87, but supported the 5th generation Cor processors. By the way, chips of the 4th generation, that is, Haswell, can be installed in the same motherboards.

System logic for Skylike

A total of 5 sets of system logic were presented for the latest generation of CPUs, codenamed "Skylike": H110, B150, H170, Q170 Z170. Comparison of Intel chipsets of the eighth and hundredth series clearly indicates the positioning of the latter. Moreover, their technical parameters are almost identical. The first of them - Н110 - is intended for use in budget and office computer systems together with Celerons and Pentiums. В170 and Н170 are oriented towards "Kor Ay3", "Kor Ay5" and "Kor Ay7" with locked multipliers. Well, with unlocked multipliers "Kor Ay5" and "Kor Ay7" (that is, the CPU with the index "K"), it is most correct to install it in motherboards based on Z170. There is one important difference in this family of chipsets, which is support for a new type of RAM - DDR4. But all earlier versions of the system logics of this manufacturer supported only DDR3.

What's next?

The life cycle of the hundredth series of Intel chipsets is just beginning. These decisions will be valid for exactly 2 more years. And the replacement process itself in the future will not be so fast. But, in any case, its receivers will have a similar division into niches. Even their designations will be similar.

Enthusiast solutions

Separately, it is necessary to consider the chipsets for enthusiasts from Intel. Chipsets of the 2011 platform were different from all previously described. The first of these was the X79. It allowed installing the most efficient Sandy Bridge and Ivy Bridge chips. It was replaced in 2014 by the X99, which was intended to install Haswell solutions. Among other differences, it is necessary to highlight in the latter support for DDR 4 standard RAM, while the X79 could only work with DDR 3. Also, these processors, in comparison with the previously described chips, could boast of an improved memory controller (4 channels) and an increased number of computational modules (the most productive solutions included 8 such units).

Intel motherboard chipsets are clearly divided into niches. The least productive solutions are recommended to be built on the basis of H81 and H110. The most productive PCs for computer enthusiasts are best built on the basis of the Z87, Z97 and Z170. The rest of the chipsets are focused on mid-range computer systems. Their performance will definitely be enough for the next 2-3 years, but at the same time the possibility of overclocking is minimized. Well, the latest BIOS updates generally indicate that such an opportunity will not be available soon. The chipset manufacturer himself is blocking it. From the standpoint of novelty, it is better to choose solutions of the hundredth series, which are now only beginning to actively appear on store shelves. But in case of budget savings, you will have to purchase more affordable 80 series motherboards.

Outcomes

This article took a closer look at the chipsets released since 2011 by Intel. This semiconductor giant updates its chipsets almost every year. As a result, each new generation of CPUs requires the purchase of an updated motherboard. On the one hand, this increases the cost of the PC, and on the other hand, it allows you to constantly improve its characteristics.

Intel with the release of the fourth generation of processors (Haswell) and the transition to a new socket (LGA 1150) launched a new line of motherboards (Lynx Point). Now there are five different chipsets Z87, H87, Q87, Q85, B85 (the Z75 did not have a receiver), divided, as always, into two segments: business and consumer. Consumer segment (Z87, H87) with a number of features designed to improve overall performance. The business segment (Q87, Q85, B85), on the one hand, has fewer options, but it contains a lot of useful information for IT departments of large and small companies.

The latest Intel processors (including Haswell) are designed to move more and more functionality from the motherboard to the processor itself. For example, integrated graphics (where it was), a RAM controller (RAM), PCI-E and DMI bus controllers, as well as processor power management are no longer located on the motherboard. This means that things like onboard video and RAM compatibility now depend more on the CPU than on the particular motherboard's chipset. Based on this, the differences between the chipsets will now be small, mainly in options, the number of peripherals supported.

The most important changes in fact in the new chipset are support for up to six SATA 6Gb / s and up to six USB3.0. Thunderbolt is not yet integrated into Haswell chipsets, but can be added with a separate controller on the motherboard.

Consumer segment (Z87, H87)

Z87

The z87 set is the most feature-rich and the only one that offers overclocking capabilities (K-series processors). The chipset also has support for SLI / Crossfire connectivity with three configurations.

As for the rest of the features, the Z87 supports Rapid Storage Technology, Smart Response Technology (SSD Caching), six SATA 6Gb / s and six USB 3.0 ports. Also, when using Smart Response Technology (SSD Caching), it optimizes the performance and power consumption of the SSD.

H87

The H87 chipset is very similar to the Z87, but it lacks some very important features: CPU overclocking and support for triple SLI / Crossfire configuration.

The H87, like the Z87, supports Rapid Storage Technology, Smart Response Technology (SSD Caching), six SATA 6Gb / s and all the same six USB 3.0. However, unlike the Z87, this chipset has Small Business Advantage support.

In general, the H87 provides almost all the same functions as the Z87, yes, without overclocking, but most likely you will choose the z87 since the motherboard manufacturers are simply pushing for this by cutting the number of ports and removed USB.

Business segment (Q87, Q85, B85)

Q87

The Q87 chipset is the most functionally rich among the business line, it supports technologies such as vPro, Active Management, Intel TXT. Plus six SATA 6gb / s and six USB 3.0 in addition to 14 and USB 2.0. This chip will definitely suit you if you are using vPro or AMT or TXT or you just want a board that supports them.

Q85

The Q85 chipset is very similar to the Q87, only it does not support these wonderful business technologies. Also, chipsets of the 85th series do not support, unlike the others, Rapid Storage technology, which allows to reduce power consumption and increase the speed of operation when using multiple disks. If you do not need these technologies, you are looking for a cheaper platform but do not want to buy the weakest one, then this is the choice for you.

B85

B85 is a budget business solution that not only does not support business technologies, but it also has four USB 3.0 and Serial ATA 600 ports, in contrast to six ports in other chipsets. The B85 is a great option for budget processors (Core i3, Pentium, Celeron).

Conclusion

Note: to be able to work with the specified technologies, they must be supported by the processor.

We can conclude that fragmentation has decreased, there are fewer chipsets, but this does not make the choice easier, the differences are minimal, often additional logic on motherboards removes them altogether.

Published: 26.01.2017

Hello friends.

This time we will look at such an important part of the motherboard and computer as a whole, like the chipset. Let's talk about the main manufacturers, the differences between the chipsets. Let's go through the price categories of chipsets of different series.

What is a chipset

Chipset (eng. Chipset) - a set of microcircuits located on and acting as an intermediary between various elements of the computer. It provides understanding of processor commands by RAM, video card, hard disk and other equipment connected to the motherboard.

Chipsets differ by manufacturer, number of internal chips, speed, supported connectors and their number, as well as many others. Let's consider the differences.

Name history

A group of motherboard control chips was originally called a chipset. These were the North Bridge and the South Bridge. Also, sometimes the chipset included a Super I / O chip connected to the south bridge and controlling low-speed connectors (PS / 2, floppy, COM, LPT).

North bridge

Northbridge or Memory Controller Hub - coordinates the processor with memory and graphics adapter. It uses high-speed buses, allowing information to be exchanged at tens of gigabits per second. Physically, it is located above the south bridge, hence its name.

South bridge

South bridge or controller-hub I / O - through the north bridge connects the processor and steel equipment connected via SATA, USB, IDE and other connectors.

Manufacturers

Such companies as Intel, AMD are engaged in the production of chipsets. Among the companies that stopped producing chipsets are NVidia, VIA and SiS, the markings of which can still be found on motherboard chipsets. The chipsets of modern manufacturers differ mainly in the supported socket. Intel manufactures chipsets for their sockets, AMD for theirs.

Differences between chipsets

The main difference between modern Intel chipsets is the absence of a Northbridge. Not so long ago, they removed it as part of the processor.

Chipsets come in different classes and categories. Of the modern chipsets from Intel, it is worth highlighting the 100 series chipsets:

H110- for budget home or office computers;
B150 and H170- for medium computers;
Q170 and Z170- for serious gaming or work computers. Only the Z170 has overclocking capability.

All of them have USB 3.0, SATA 3, PCI-E x16 connectors. The main difference between these chipsets is in the number of supported connectors and slots. All of them are able to work with modern i series processors (i3, i7, i5).

Modern AMD chipsets are divided into 2 categories: A series and 9 series. The main difference between the 9th series is that it can work with 8-core AMD processors. The 9-series supports AMD OverDrive fine tuning and socket FX support for 8-core processors. Chipsets A series today are presented:

A58- for very budget and low-speed systems, without support for SATA 3 or USB 3.0;
A68H- for budget computers;
A78- for medium and multimedia machines;
A88X- for high-performance work or gaming PCs, with the ability to overclock.

AMD chipsets are cheaper than Intel chipsets, but at the same time they have fewer supported slots.

2016-2017 will not present new platforms to the personal computer market: fans of Intel products are in full swing mastering the recently presented Skylake architecture, and AMD fans have patience until the end of this year - early next year, when the first products with support for the new AM4 socket are expected to go on sale. However, those consumers who want to radically improve an existing or buy a new computer are in a difficult situation. Now the question of how to choose the best motherboard (system) board does not have a single answer.

What should you pay attention to?

The motherboard is the heart of the computer. It is she who determines which processor, memory, hard disk and other components can be installed in the system.

Some characteristics of motherboards have become the de facto industry standard, therefore, they are true for all modern models. Among them are the presence of USB 3.0 ports (a universal means of communication with almost all external peripherals and gadgets), Ethernet (local area network adapter), and one or more PCI-e x16 slots (video cards are connected to them). Thus, when choosing a suitable motherboard, you should pay attention only to:

• form factor - the physical dimensions of the board. They determine the type of computer case and the possible number of expansion slots (it is impossible to place a large number of large parts on a small piece of PCB). Nowadays mini-ITX, micro-ATX, ATX, extended-ATX are relevant (arranged in order of increasing sizes). The first ones are designed for very compact computers, contain only one expansion slot and in some cases the central one is already wired to them. Extended-ATX cards are designed for the highest power systems possible;

The motherboard is the foundation of the computer

• type of processor socket;
• a set of system logic (chipset), on which support for individual proprietary technologies depends, the maximum amount of RAM, a list of expansion slots and ports for peripherals.

New or tried and tested old?

The latest addition to the personal computer market is Intel's Skylake architecture. It brought the LGA1151 processor socket, support for DDR4 memory and a number of technologies that are not so important for the average consumer. However, at present, the practical benefits of these innovations are not obvious - the increase in performance compared to the previous generation is not noticeable to the eye.

In most special test applications or computer games, the increase in computing power does not exceed a few percent. DDR4 also has yet to reveal its potential, but it will require more advanced chipsets, memory modules and processors. As a consequence, Haswell platform with LGA1150 socket and DDR3 is still relevant.

Attention! Skylake processors support DDR4 and DDR3L memory. The latter operates at a lower voltage than DDR3 (1.35 V versus 1.5). DDR3 and DDR3L modules are not interchangeable. Installing memory that is not supported by the processor and motherboard may result in component failure.

The only choice for users looking for maximum performance is motherboards with an LGA2011-3 socket. This platform supports quad-channel DDR4 memory and up to 40 PCI-e 3.0 lanes (up to 4-5 graphics card slots).
Relatively modern platforms from AMD corporation are AM3 + and FM2 +. Motherboards with these connectors support the mainstream of technology. However, AMD processors are inferior to competing solutions from Intel in terms of performance, heat dissipation and power consumption. The feasibility of building a system based on the AM3 + and FM2 + platforms is now questionable.

Finally, there are boards with pre-installed processors and AMD's AM1 platform. They are cheap, but their performance is only sufficient for working with text, browsing the Internet and playing games from 10 years ago.

What chipset should the motherboard be on?

For each of the platforms, manufacturers have presented several chipset models:

1. Intel LGA1150:
   • H81 - component overclocking is not supported (a special setting that increases operating frequencies and performance), no more than 2 memory modules can be installed;
   • B85 - overclocking is not supported, installation of up to 4 memory modules, a set of proprietary technologies for building a business infrastructure is supported;
   • Q87 over B85 supports more USB ports and business software technologies;
   • H87 is aimed at home users, therefore, unlike Q87, it does not support business technologies;
   • Z87 fundamental differences from other models are reduced to overclocking support.
2. Intel LGA1151:
   • H110 - no overclocking support, the number of memory slots is limited to 2;
   • H170 - the number of memory slots has been increased to 4;
   • B150 supports fewer USB ports compared to H170, the chipset is designed for business users;
   • Q170 - support for more technologies for business;
   • Z170 - support for overclocking, more USB ports, increased bandwidth PCI-e (useful when installing multiple video cards).
3. Intel 2011-3:
   • X99 - overclocking is supported, a large number of USB ports, technologies for business, the maximum possible bandwidth of the PCI-e bus is provided.
4. AMD FM2 +:
   • A88X, A78, A68H, A58 - support up to 4 memory slots and overclocking. Significant differences boil down to the availability of CrossFire technology (needed to install two video cards on AMD graphics processors, present in the A88X), the number of USB and SATA ports (for connecting optical drives and). Overclocking options vary depending on the individual characteristics of specific motherboard models.
5. AMD AM3 +:
   • 990FX - up to 4 PCI-e x16 slots, maximum overclocking stability, 4 memory slots;
   • 990X - up to 2 PCI-e x16 slots, overclocking support, 4 memory slots;
   • 970 - 1 PCI-e x16 slot (motherboard manufacturers increase their number to 2 by third-party means), overclocking support, 4 memory slots.

Attention! For effective overclocking, the corresponding technologies must be supported not only by the motherboard, but also by the processor. Chips with an unlocked multiplier are marked with a K index, for example, A10-7870K or Core i7 6700K. At the same time, all processors for the AM3 + platform of the FX series have a free multiplier.

Intel Corporation produces under the Core i5 trademark quad-core processors without support for multi-threading technology - Hyper Threading. It allows you to simultaneously process 2 computational threads on one core, while a four-core processor approaches an eight-core in terms of computing power. The performance of the Core i5 chips is sufficient to meet all the needs of the home user.

Motherboards for Intel Core i5

Modern chipset models support the entire line of processors of the corresponding generation. So, for Haswell Core i5 chips, motherboards on any set of system logic - H81, B85, Q87, H87 or Z87 are suitable. The situation is similar with the Skylake architecture.

Advice. Overclocking support increases the cost of the processor and motherboard. If you do not plan to increase the factory frequency, there is no point in overpaying for components. The combination of a locked multiplier processor and a Z-series chipset will not provide any practical benefit. The influence of the sets of system logic on the overall performance of the system (all other things being equal) is currently reduced to statistical error.

Gaming motherboards

Throughout the history of personal computers, games have been one of their main purposes. This type of entertainment has come a long way from being a hobby for geeks, children and teenagers to being officially recognized as a sports discipline. At its core, a computer game is not much different from other software, for example, a text editor or three-dimensional models.

The latest innovation in the digital entertainment industry will run on any system capable of providing sufficient processing power - with a certain amount of RAM and graphics memory, free hard disk space, a suitable graphics and central processing unit. However, component manufacturers are trying to break this axiom.

Gaming motherboard

In the past 5-10 years, marketers have been actively promoting the concept of a "gaming computer", referring to the maximum processing power and striking eye-catching design. This term is also used by motherboard manufacturers. In the assortment of each of them there is a specialized line of products for gamers.

Gaming motherboards have unusual PCB colors, LED backlighting and large decorative panels or radiators on the chipset and key nodes of the power circuits. Such components are more expensive than their counterparts, but in fact only demonstrate the external attributes of the gamer subculture. The key characteristics of a regular motherboard are no different from a gaming PC product based on a similar chipset.

The modern motherboard market allows you to choose a product that best suits the individual preferences of the end user. At the same time, bright design, maximum practicality or system performance may be presented as the main requirement. A careful analysis of the basic characteristics of motherboards will save you from rash purchases and help save your money.