Connect with us

PC / Laptop

Details about Intel H370, B360 and H310: what will have to tolerate those who will save on the motherboard

Advertisement
Samsung J7 V just $5 mo. New device payment purchase req'd. Plus, free shipping.

Intel Coffee Lake are present on the market since October of last year, and their release was a great event in the life of a computer community. For the first time in many years, Intel made a step in the direction of increasing the number of cores, significantly improving the performance of their processors in each price segment. It is considered that such an act is microprocessor giant has pushed action of AMD, which has a very good multi-core Ryzen, and Coffee Lake is a kind of response to Intel, allowing it to save the changed market landscape the competitiveness of their proposals. Anyway, the announcement of the first batch of Coffee Lake was really accelerated, and, until recently, Intel as a temporary solution offered six models of desktop processors of this family and one chipset that is compatible with them – Z370. The full announcement of the Coffee Lake and associated desktop platform took place only last month when pioneer modifications added an additional 18 desktop CPU and four additional chipset.

Held extension of the model range is a very important stage of the implementation of the new platform. The fact that until recently any of the processors Coffee Lake could only be used with a relatively expensive overclocking motherboards, regardless of whether he has the acceleration or not. And if the Intel Core i7-8700K and Core i5-8600K motherboard on base of chipset Z370 looked quite organically, for buyers more affordable locked processors like the Core i3-or Core i5 8100-8400, they were not just meaningless, but expensive for the price. This ultimately increased the cost of ownership platform in a cheap Coffee Lake and did not allow them to become a viable alternative for Ryzen Ryzen 5 and 3, which AMD from the beginning offered a wide selection of compatible platforms in different price categories. In other words, without having to buy the motherboard fully disclosed to the market potential Coffee Lake had no chance.

But now everything fell into place. The release of fresh sets of logic H370, B360 and H310 solves all the problems with high cost platforms for the Coffee Lake. The motherboard is built on a new associated with Coffee Lake “hub controller platform” (PCH – platform controller hub) must have a much more affordable and allow to build a relatively inexpensive build on those processors with the increased number of cores. But interesting they are not only this.

With the announcement of new chipsets for Coffee Lake is associated a curious fact, which largely was the reason for the appearance of this material. Despite the fact that the new chipsets are deprived of any possibilities to overclock the CPU and DDR4 SDRAM, truncated versions Z370 they really are not. The fact is that while the previously proposed overclocking chipset for Coffee Lake was only adapted modification based on Kaby Lake processors chipset Z270, new chipsets for Coffee Lake is a separate series of circuits related to the next generation. What this means from a practical point of view, we try to understand this material.

#Chipsets three hundredth series: what’s new

Intel has accustomed us to the fact that the sets of system logic, marked with the letter Z – it’s not just overclocking platform, but the chipset, which is the maximum possible among the entire family. However, in the case Z370 it is not so. Despite the fact that Intel has included Z370 old and new H370, B360 and H310 in single series c three hundred rooms, this is a fundamentally different product, with different origins, and therefore markedly different in the details. Internal generic name of three hundred new chipset taken Intel – “Cannon Lake PCH”. It indicates that H370, B360 and H310 were intended for the seriously delayed 10-nm processors (according to new data, wait for them before 2019, alas, not worth it). The same set of system logic Z370 terms of design refers to a series of Kaby Lake PCH, that is, it is a close relative of the two hundredth series chipsets and Z270 in particular. There is therefore nothing surprising in the fact that many things H370, B360 and H310 were better and osnashenie “senior” Z370.

These chipsets differ even from the industrial point of view. While the Z370 is manufactured using 22 nm process, the new chipsets three hundred series have been translated into more modern process technology with the norms of 14 nm. Thus, in the updated version of the platform LGA 1151v2 processor and the chipset was manufactured by the same production standards, that still never happened. Therefore, the chip H370, B360 and H310 compared to the Z370 needs at least less warm at work.

Cannon Lake PCH

Cannon Lake PCH

But only one that benefits from the new chipsets are not limited. The first key innovation is to consider appeared native support for up to six ports USB 3.1 Gen 2 with a bandwidth of 10 Gbit/s. Previously, such ports in those chipsets were not supported at all, and the motherboard manufacturers had to implement them, resorting to external controllers, most of the production Cost. Now it is in the past. However, the innate support ports Type-C in the chipset however has not appeared. This means that adding on fees of symmetric connectors still require the motherboard manufacturers additional spending on chip redriver, so cheap computers new-fangled double-sided USB connector, apparently EN masse until you get.

Appeared in the new chipsets and support for promising Thunderbolt interface, although many were expecting. Implementation of high-speed Thunderbolt ports, as before, will require integration of on-Board external chip connected to two PCI Express. This means that the widespread introduction of Thunderbolt in the near future to wait still not worth it.

The second key feature of the three hundredth series chipsets should be considered the integration part of the components required for implementing Wi-Fi wireless controllers (2T2R 802.11 ac) with a capacity of up to 1733 Mbps. Scheme chipsetov Wi-Fi was taken from the platform Gemini Lake. The bottom line is that the chipset is integrated the most complex and expensive functional units such as logic, memory, and MAC, while the controller of the physical layer and the antenna must be placed on the external radio frequency module that is installed in the dedicated M. 2 slot or raspravama on the Board. As for the connection between chipsety logic and the external physical controller you intend to use proprietary CNVi-interface. Compatible with CNVi external M. 2 2230-modules 9560 Wireless AC (2T2R), Wireless AC 9462 (1T1R) Wireless AC and 9461 (1T1R) are present in the range of Intel, beginning in the summer of 2017.

However, as in the case with the ports of the USB Type-C, the widespread emergence of integrated Wi-Fi on the new LGA 1151v2 cards is not worth waiting for. The fact that the implementation of the necessary circuits and equipment motherboard RF module-companion increases the cost of the final product by about $15. As a result, to receive support for Wi-Fi has a chance only in a relatively expensive platforms where the producers not to chase the lowest price.

There are sets of logic H370, B360 and H310 and the third big contrast to the Z370. It concerns energy saving: along with the new chipsets in desktop computers comes support energy-saving States of the CPU C8-C10, which were originally presented in the solutions for ultrabook-class Haswell ULT. Furthermore, additional energy saving mode S0ix got themselves chipsets.

In the list of the platform Coffee Lake with access H370, B360 and H310 were added and Intel Modern Standby. Thanks to her desktop had the opportunity like intelligent column “listen live” and solve some background tasks like checking e-mail even in the sleep state. This functionality already has been implemented in laptops, and now may find a place in desktop computers.

#Perspective: Intel Z390

Speaking of new sets of logic for Coffee Lake, belonging to the family Cannon Lake PCH, mention should be on Z390, which has not yet been announced by Intel. Now it so happened that the younger chipsets H370, B360 and H310 has gained tangible benefits in the possibilities of overclocking on Z370. But this should not be. Aimed at use in the most tricked-out systems, the chipset needs to be a leader in everything, therefore, to replace the suddenly outdated Z370 soon will come an improved version of Z390.

Z390 announcement scheduled for the second half of the year, and, apparently, already at Computex in early June, we will see samples of motherboards based on it. However, the official announcement and start of sales will likely take place later, for example, in anticipation of the start of the season back-to-school in late summer.

However, do not think that the Z390 can lead the platform LGA1151v2 on some new level. This chipset, in fact, be another variation of Cannon Lake PCH with a view to overclocking the system. That is, in comparison with the Z370 he will offer the same improvement in the capabilities that are already in H370, B360 and H310: native support of ports USB 3.1 Gen2 and embedded controller 802.11 ac channel level.

Roughly speaking, about Z390 can be thought of as analogous H370 with support for overclocking, and also slightly increased the maximum number of lines PCI Express, and the maximum number of ports USB 3.1 Gen2.

#Specifications Intel H370, B360 and H310

Generally speaking, the number of new sets of logic includes four products: H370, Q370, B360 and H310. But in this article we are only talking about three products, but says nothing about Q370. This is due to the fact that the chipsets group Q aimed at the corporate market, and is unlikely to meet ordinary users. In General, however, Q370 can be described as advanced version H370-enabled advanced security and administration for Intel VPro technology. As for consumer of three chipsets, then their main characteristics are in the table below, where they are simultaneously mapped to the familiar set of system logic Intel Z370.

  Z370 H370 B360 H310
Processor support Intel LGA 1151v2 (Coffee Lake)
Process technology 22 nm 14 nm
PCI Express 3.0 (via CPU) 1×16
2×8
1×8+2×4
1×16
Memory channels/DIMM per channel 2/2 2/1
HSIO ports only 30 24 14
PCI Express 3.0 (via chipset) 24 20 12 6 (PCIe 2.0)
Support overclocking There No
USB ports, just 14 12 10
Ports USB 3.1 Gen2/Gen1 0/10 4/8 4/6 0/4
SATA ports 6 4
Rapid Storage Technology (RST) There No
The number of PCIe devices that support the RST 3 2 1 0
RAID support There No
Support to the RST of the devices connected to CPU There No
Intel Optane Memory There No
Intel Gigabit LAN There
Integrated WLAN-ac (CNVi) No There
The price (roughly) $$$$ $$$ $$ $

With the advent of new chipsets no successor Z370 have not yet formed. So until then, until it is announced Z390, it Z370 will continue to be the “gold standard” for enthusiasts. Because this is the only option that allows you to overclock the processors and memory on the platform LGA1151v2. In addition, this set of logic laid the greatest number of lines PCI Express 3.0 and the highest number of USB 3.1 ports (though only with a bandwidth of 5 Gbps). Added to this are the most extensive means of Rapid Storage Technology with the capability to collect RAID arrays from SATA and NVMe storage devices.

But the old Z370 new chipsets and 300-series belong to different generations PCH, and this leads to the fact that H370 on a number of characteristics superior fellow with a higher positioning. Earlier chipsets group H was repeated characteristics Z with the exception of the support for the dispersal and dividing lines the processor interface PCI Express. Now the situation has changed dramatically. Overclocking in the H370 and no, as there is no support for bifurcation 16 processor lines of PCI Express, but this chipset has to offer absent in Z370 USB 3.1 Gen2 with a bandwidth of 10 GB/s and integrated 802.11 ac Wi-Fi. In total, this means that in cases where acceleration is not required, the Board on the basis H370 can be interesting not only from the point of view of prices, but the characteristics.

However, there are nuances. Despite the fact that Z370, H370 have the same set of 30 HSIO ports, configure them in lines PCI Express 3.0. Board on the basis of the older chipset can use up to 24 chipset lines PCI Express 3.0, while in the case H370 is available in 4 lines less. In addition, Z370 – the only chipsets for desktop processors Coffee Lake, which allows you to assemble RAID arrays from NVMe storage connected to the processor lines of PCI Express.

Despite the fact that H370 seems the best-equipped variant of the second wave three hundredth series chipsets, we are not inclined to think that he will be able to gain widespread popularity due to relatively high prices and excessive opportunities for low-cost systems. A much more appropriate choice in this case seems B360 or perhaps a very stripped-down and cheap H310. It is interesting that Intel put away in Z370 Wi-Fi controller, even in its most simple set of logic. However, since the implementation of Wi-Fi on the boards requires adding an additional controller to the physical level, it is unlikely that wireless network is often to meet in real platforms H310.

Anyway, in the background, what are the limitations inherent in the H310, the presence integrirovannoi Wi-Fi like some kind of mockery. In particular, the boards on this chipset can be equipped with only two DIMM-slots and lack of technology Optane Memory. In addition, H310 no support for USB 3.1 Gen2 and PCI Express 3.0. The number of high-speed ports in HSIO H310 is limited to 14 pieces, which should be divided between the ports SATA, USB 2.0 and USB 3.0 and PCI Express 2.0 standard. In other words, H310 is very similar to the “old man” H110 and difficult to characterize as anything other than a bunch of compromises. It is obvious that due to the lack of resources it is quite ill-suited to create a full-sized motherboards of the current level. Most likely, a typical habitat H310 will be simple and extremely cheap motherboard Micro-ATX with a price range from $50 to $80.

The most balanced option to the mass LGA1151v2-motherboards seem to be a set of logic B360. There is ample support for the modern system the number of 24 HSIO ports, which somehow will be divided into 6 SATA ports, 12 USB ports (6 of which can work in the USB 3.0 standard and 4 – in the USB 3.1 Gen2) and 12 lines PCI Express 3.0. Cut in B360 only legkoizvlekaemye functions, for example, support for RAID arrays, or division of the processor lines of PCI Express. At the same time, in the B360 is Intel Optane Memory, has high speed USB ports with a bandwidth of 10 Gbps, and there is an integrated Wi-Fi. In sum, it gives the opportunity to build on the basis of this set of logic is quite functional and modern system processors Coffee Lake, but the average price of plat on the basis of the B360 is projected no higher symbolic boundary of $100.

#Overclocking no, but what about Multi-Core Enhancements?

Traditionally, the boost function of Intel processors is available only on motherboards based on the chipsets of the group Z. there is no change has not occurred: if you want to overclock a Core i5-or Core i7 8600K-8700K to frequencies in the neighborhood of 5 GHz, the choice of appropriate platforms is limited only by the motherboard based on set logic Z370 (and in the future – Z390).

But overclocking in relation to Coffee Lake exists in various forms and can be not only to increase the CPU multiplier, which is available at a special unlocked CPU. There is another approach – Multi-Core Enhancements, which are quite legitimately be considered the Lite version of acceleration, suitable for all models of processors.

Recall, the inclusion of Multi-Core Enhancements allows output of the processor to the maximum allowed by the specification frequency, regardless of what level of energy consumption at this point he demonstrates. Roughly speaking, Multi-Core Enhancements “pumping” Turbo Boost technology 2.0 processor and displays the limits of inherent TDP. The advantages of Multi-Core Enhancements in the case of Coffee Lake is obvious: the processors of this type have an increased number of computing cores, and high stress quickly leads to excess heat generation and power consumption limits of the thermal packet 65 or 95 watts. In normal conditions, this would have entailed resetting the clock frequency of the CPU to normalize the electrical and thermal performance, but the Multi-Core Enhancements allows to neglect this and allows the processor to work on a frequency, which is the maximum allowed to load on all cores.

This trick is akin to gentle acceleration and allows significantly raise the effective frequency of six-core Coffee Lake in the modes with high load, and it works well for neversmokers versions of the CPU. As an illustration, look at how the frequency of six-core Core i5-8400 with passport design dissipation 65 W when testing in LinX 0.9.2 in its normal mode:

And when you enable Multi-Core Enhancements.

The differences can say, dramatically. If you do not resort to the inclusion of the Multi-Core Enhancements, then under high load, the Core i5-8400 operates at a frequency of 3.1-3.3 GHz, although maybe 3.8 GHz is the frequency is maximal for the mode with the load on all cores. But to get to the maximum gives the energy consumption. At a frequency of 3.8 GHz in LinX CPU requires 100 watts of electricity. In order to allow the processor to go beyond the TDP and need a Multi-Core Enhancements, which essentially disables the internal control over consumption and has built in mechanisms to throttle frequency.

Second example: here’s the picture with the frequency observed in 95-watt Core i5-8600K in the nominal mode.

And this is when you enable Multi-Core Enhancements.

This processor is initially more liberal TDP – 95 W, so the restrictions on the power consumption here is not as critical. However, the maximum permitted load on all six cores frequency of 4.1 GHz LinX consumption can reach up to 110 watts, so without the function of Multi-Core Enhancements, the frequency is forcibly reduced from 4.1 to 3.8-3.9 GHz.

All this, naturally, affects productivity, for example, in the benchmark Linpack 2018.2.010, which is based LinX application.

 

Core i5-8400 (65 W)

Core i5-8600K (95 W)
MCE off 262 GFlops 312 GFlops
MCE included 304 GFlops 321 GFlops
Performance increase 16 % 3 %

Of course you need to keep in mind that Linpack draws some border picture, since this test actively uses energy-intensive instructions set AVX2. But still the results are striking: enable Multi-Core Enhancements 65-watt processor can add to their performance of 16 percent. And this, incidentally, demonstrates once again that the new six-core processors Intel is not so economical as one would think, looking at the specs, their performance is artificially slaughtered boundaries of the thermal package.

Thus, the Multi-Core Enhancements – an important tool to increase the performance of a new generation of Coffee Lake. The greatest efficiency it provides in conjunction with the blocked powerbrokers processors with the reduced to 65 watts within a heat pack.

In light of this, it naturally arises the question about the function support Multi-Core Enhancements the new motherboards built on the chipsets H370, B360 and H310. And the short answer is Yes: despite the fact that the acceleration in the classical meaning of these chipsets is not supported, a new generation of cards, however, allows you to cancel for Coffee Lake the boundaries of the thermal packet and bring them to the maximum allowed frequency. However, in the General case, it is not as straightforward as in the case of circuit boards on the basis of the Z370, which in the UEFI BIOS prepared the corresponding option, which is also frequently activated by default and does not require any user intervention.

Of course, much depends on how the BIOS is programmed a specific motherboard, but none of the boards on the basis H370, B360 and H310 that have passed through our hands, and one simple switch to Multi-Core Enhancements in the UEFI shell had. Moreover, the boards is the new generation most often, this functionality is deactivated by default. That is, the vast majority of people who have decided to opt for LGA1151v2 – overclockers motherboards on chipsets, we recommend you to devote some time fine-tuning the processor and to remove the limiting performance in demanding applications limits the heat dissipation and power consumption.

#How to enable Multi-Core Enhancements?

To enable Multi-Core Enhancements on the motherboard does not support overclocking and has no corresponding item in the UEFI, will have to dig around in the settings for turbo mode. The fact that the implementation of the Multi-Core Enhancements is based not on the overclocking capabilities of the chipset and Turbo Boost technology 2.0, which is made for any LGA1151v2-configurable processors. Opportunities for power management and, as a consequence, the frequency put into it in order for the system builders can configure the settings for Turbo Boost 2.0 under specific operating conditions, given the chosen means of cooling, power supply and, in the end, formats system.

Overall, the Turbo Boost 2.0 enables you to boost the CPU frequency in those moments when the burden on computing resources is small, and such a “legalized acceleration” is not in danger of exceeding limits on temperature and power consumption.

Within the technology Turbo Boost 2.0 is defined by five modifiable variables:

  • Power Limit 1 (PL1) – the border for which should not exceed the average power consumption of the processor. By default, it is considered to be equal to the TDP of the processor;
  • Power Limit 2 (PL2) – the boundary to which a short-term excess of energy consumption. By default, this value is taken to do equal to 1.25∙TDP;
  • τ – the time during which it is allowed to exceed power level PL1 PL2 in the framework. Is set equal to from 1 to 127 seconds depending on the inertia of the cooling system.;
  • Power Limit 3 (PL3) and Power Limit 4 (PL4) – peak limiting short-term emissions of energy consumption not longer than 10 MS. In the desktop CPUs is disabled.

Thus, it becomes clear the essence of the Multi-Core Enhancements: this feature, in fact, raises the limit value PL1 above the TDP. This is what allows the processor within the technology Turbo Boost 2.0 to reach the maximum permitted frequency and not to drop them when the real long-term power consumption exceeds the rated TDP.

It remains only to clarify the question how the user can change the value of PL1 in practice. Here again, all highly dependent on manufacturers of motherboards, but most of the motherboards based on chipsets H370, B360 and H310 of the relevant option still not deprived, just need to know where it is in the structure of the UEFI BIOS and how it is called:

  • At ASUS it is called the Long Duration Package Power Limit and located under Ai Tweaker/CPU Power Management Internal;
  • ASRock have this option called Long Duration Power Limit and is available under OC Tweaker/CPU Configuration;
  • Gigabyte it can be found in the subsection M. I. T./Advanced CPU Core Setting called Package Power Limit1 – TDP;
  • At MSI it is located in the Overclocking/CPU Features called Long Duration Power Limit.

After assigning the appropriate parameter values large enough, which obviously exceeds achievable in practice, the power (e.g. 200 watts), you allow the CPU to always run at its maximum frequency, which is determined only by the number of utilized cores and not depend on what energy appetite shows CPU.

  TDP, watts Nominal frequency, GHz The maximum frequency depending on the load, GHz
6 cores 5 cores 4 cores 3 core 2 cores 1 core
Core i7-8700K 95 3.7 V 4,3 4,4 4,4 4,5 4,6 4,7
Core i7-8700 65 3,2 4,3 4,3 4,3 4,4 4,5 4,6
Core i5-8600K 95 3,6 4,1 4,1 4,2 4,2 4,2 4,3
Core i5-8600 65 3,1 4,1 4,1 4,2 4,2 4,2 4,3
Core i5-8500 65 3,0 3,9 3,9 4,0 4,0 4,0 4,1
Core i5-8400 65 2,8 3,8 3,8 3,9 3,9 3,9 4,0

In the end, even cheap motherboards have everything you need to ensure that the processors Coffee Lake worked for them in modes that go beyond the spec in the TDP. And it makes the platform with the chipset H370, B360 and H310 are much more attractive. Even if they do not support overclocking, but the frequency of about 4 GHz at full load with them quite unable to give any six-core processor family Core i5.

The second conclusion concerns the fact that the use of a card on chips H370, B360 and H310 overclocking Intel Core i7-8700K and Core i5-8600K makes no sense: almost the same performance if properly set will ensure a regular Core i7-8700 and Core i5-8600. The opposite is true: to buy CPUs other than Core i7-8700K and Core i5-8600K, the Board on the basis of the Z370 is clearly not worth the higher price they will not give any significant advantages. The Multi-Core Enhancements in one form or another is feasible in low-cost platforms, and this is important.

#Summary: the performance

Despite the fact that motherboards based on chipsets H370, B360 and H310 function in a Multi-Core Enhancements can be activated detours, in the General case, the performance of the systems based on them will still be slightly lower than that of systems with boards based on the Z370.

Yes, after correct configuration of the parameter Power Limit of 1 real operating frequency of the processor in both cases would be identical. However, the difference in performance may occur for other reasons, at the expense of memory. The fact that the sets of logic H370, B360 and H310 severely restrict the choice of modes DDR4 SDRAM, and do with it anything.

While the Board based on the Intel Z370 with any processor family Coffee Lake can run the memory at frequencies DDR4-4000 or even higher, the new platform without the support of the crackdown will give the memory as much as possible only the frequency which is designated in the specifications for the specific CPU. This means that the platforms H370, B360 and H310 memory frequency with CPU Core i7 and Core i5 will be limited to the mode DDR4-2666, and with processors Core i3 and Pentium – mode DDR4-2400.

Of course, the impact of memory speed on the performance of systems with Coffee Lake is not as significant as in the case of Ryzen. But to ignore this factor is still not worth it. In other words, the construction of the computer with the highest level of performance possible only with the use of boards based on Intel Z370. But how serious is this difference that should show the tests.

SOURCE

Advertisement
Samsung J7 V just $5 mo. New device payment purchase req'd. Plus, free shipping.
Code: VZWDEAL. Enter this coupon code at checkout to get $100 discount on Samsung Galaxy Note 8. Includes free shipping. Restrictions may apply. Device payment purchase required.
Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

PC / Laptop

The 2018: PC processors

Advertisement
Save $120 on iPhone SE 16 GB with Sprint Flex Lease when you visit Sprint.com. $7.09/mo. for 18 mo. after $120 instant savings for well-qualified customer with new-line activation.

If we talk about processors, then 2018 will be remembered first and foremost not because of some new and impressive products, but for the opposite reason. Of course, we cannot deny that AMD and Intel doing a good job of that last year left at least boring, but the thing that dominated the CPU market for the entire year is a never – ending problem. Various difficulties arose here and there, they raised both the manufacturer of the CPU, and left a significant imprint on everything that happened. Moreover, many of these problems was not resolved last year and in varying degrees, be relevant in 2019.

#Security issues

The main trouble with which the CPU market had to face in 2018, of course, became the vulnerability of families Spectre and Meltdown. And if it’s not a disaster, it is very deep and difficult problem, because attacks with their use exploit basic principles of modern micro-processors, which are widely used to increase performance: branch predictions, and speculative (anticipatory) execution of commands. Very indicative of the fact that since about the Spectre, and Meltdown was notified manufacturers of processors, until the beginning of 2018, when the information about these vulnerabilities was posted in public access, passed for six months. However, there is no clear response during this time was not followed and, moreover, certain types of attacks through third-party channel using the principles of Spectre and Meltdown remains possible to this day!

In varying degrees, were prone to Meltdown and the Spectre of the topical products not only Intel and AMD, and ARM, and even Power. However, more than any other from among the manufacturers of x86 processors have suffered all the same Intel. Microarchitecture Core was open for a solid number of varieties of attacks, which completely neutralize one only microcode fixes and patches the operating system has failed to this day. Intel have to make changes in the design of their chips at the hardware level, but it takes time, and the first CPU microprocessor giant, where the vulnerability will be eliminated in the best case will be a 10-nm Ice Lake, which is expected to be available by the end of 2019.

In the meantime users have to be satisfied with the OS patches and service packs installed firmware, installing of which, however, is not without a trace. The speed of those processors, in some scenarios, when you enable protection against Spectre and Meltdown drops by up to 30% and significantly reduces the performance of operations of input-output, in particular for calls to the disk subsystem. And this has to be tolerated, because even the processors of the Whiskey Lake and the Coffee Lake Refresh, where the developer already made some changes that make it difficult to attack, installing required patches, still leads to performance degradation.

Against this background, AMD was in a more advantageous situation: attack Meltdown for them not scary at all, and the practical operation of the Spectre requires much more effort. Therefore, we can say that AMD got off lightly: although close the vulnerability patches and service packs installed firmware is also needed, in case Ryzen this does not lead to obvious deterioration of consumer qualities of the processors.

Later, however, security systems with AMD processors, there were several other specific claims that have been announced under the name MasterKey, Chimera, Ryzenfall and Fallout. However, these vulnerabilities belonged to the security coprocessor and the chipset, that is does not directly affect the architecture of the CPU. And in addition, their operation required elevated privileges. In other words, vulnerability was secondary, and AMD also promised to eliminate them quickly via a BIOS update, so in the end, and then for the company all came together extremely well.

#Problems process

Intel, meanwhile, managed to fall into another loud scandal – the company suffered a fiasco with the introduction of 10-nm process technology. Initially, a 10-nm process technology was announced by microprocessor giant in 2017, with mass production of semiconductor devices with its use was to begin in the second half of 2017. However, in reality this did not happen. By the beginning of last year, Intel launched only a pilot delivery of limited volumes of 10-nm chips Cannon Lake only one client, and these processors were obviously a test product with limited capabilities: they had only a few cores and have been deprived of whatever graphics core.

Further development of this story was even more sad. In April, Intel reported that supplies of 10-nm chips continue to wear the trial the nature and the mass production of 10-nm products will be postponed until 2019. And still later, in the summer, it was announced that further delay the introduction of advanced technology, this time to the end of 2019. At this point, was beginning to seem like Intel is ready to completely cancel its 10-nm process, which the company has encountered numerous difficulties. And the biggest of them was the fact that Intel has traditionally been tied to the development of process technology new microarchitecture, so the delay in the implementation of the 10-nm technology automatically led to impossibility of implementation of any innovation in processor design. For this reason, all processors, which Intel brought to market in 2018, not only made for 14-nm process, but relied on the old microarchitecture Skylake, developed in 2015.

However, at the end of last year, Intel still has confirmed that recalcitrant 10-nm technology will still be implemented for mass production of chips and processors first mass produced with the application, will become chips of Ice Lake, built on a new microarchitecture Sunny Cove. Intel has promised that by the end of this year the market will be ready-made systems based on them, and explained why it could not cope with the “thin” standards.

It turns out that the problem was that the microprocessor giant has set himself too ambitious in terms of scaling the size of transistors. The transition from 14 – to 10-nm technology, as originally envisaged in the project was to increase the density of semiconductor crystals was 2.7 times and less aggressive goals, the company informed in front of him never put. For example, 14-nm process technology has increased the density of the crystal is only 2.5 times, and before that the typical ratio was a value of 2.1 and 2.3. However, the priorities originally set Intel still leaves in place. In other words, the two-year delay in the commissioning of the 10-nm technology still gave the company the necessary time to debug the equipment, and semiconductor crystals for future processors Ice Lake will be exactly the targets that I initially thought.

With the problems concerning the technological processes, in 2018 he faced, and AMD, although they were quite different in nature. The fact that long-standing manufacturing partner AMD, GlobalFoundries, suddenly changed the strategy and changed their plans. In August, the semiconductor forge announced a complete waiver of the development and implementation of 7-nm process and the desire to focus on the production of chips solely on the technical processes with the standards 12 and 14 nm and the improvement of its technology FDX (FD-SOI). Along with that GlobalFoundries has suspended all of its work on the implementation of EUV-lithography and even began to seek buyers for already acquired the lithographic equipment. This solution is one of the leading contract manufacturers of semiconductors was due to purely economic reasons: GlobalFoundries felt that received the required long-term pool of customers on old technical processes and the introduction of new technology sucking out her finances and does not promise profit in either the short or in the medium term.

Therefore, although AMD has traditionally used the power of this contract manufacturer to place orders for the release of all its CPUs, and now was forced to build relationships with new contractors. AMD plans included the transfer of all advanced products for 7-nm process technology for 2019, so looking for a new partner had very quickly. And it was the company TSMC, which will now be responsible not only for the GPU release, but also for the production of future processors Ryzen and EPYC, built on microarchitecture Zen 2. This change raises some concerns whether the new contractor to provide the proper amount of supplies. And the unequivocal answer to this question yet. But by the end of 2018 download 7-nm TSMC production lines for the first half amounted to only 80-90 %. Therefore, it is hoped that a short supply of promising chips with the architecture Zen 2 still will not occur.

#Of problems with backorders

Story short supply in 2018 is very painful for Intel. Despite the fact that ahead of the microprocessor giant has finally led to successful resolution of the situation with the introduction of another semiconductor technology, the whole story still “came out sideways”, so much so that Intel now it is time to worry about loss of market share. The fact that, starting to convert production lines for the production of advanced 10-nanometer chips, Intel was forced to restrict the output of 14-nm products, and as a result, by the middle of 2018 faced with the fact that it fails to fully meet the demand on the processors current model range.

Provoked such situation events in the server market, which demonstrated an unexpected rapid growth and was in need of a larger number of chips for data centers, but it is reflected primarily in consumer decisions. Still, the interests of major customers Intel puts above all. So when we are talking about the fact that the company is unable to produce the required number of processors, the decision was made primarily to limit the supply of low-cost solutions for laptops and desktops. As a result, in the middle of the year began a significant underdelivery of massive models the CPU, which resulted in shortages and rising prices, eventually affecting the entire range of Intel’s consumer platforms.

At the peak of the deficit, which peaked in September-October, prices on popular desktop processors like the Core i5-and Core i3 8400-8100 has risen by 30-40 percent, and never returned to normal even today. However, nothing surprising in this. Although Intel and directs a huge effort to expand production of processors for 14-nm process and even invest in the expansion of aging the production of an additional billion dollars, install and configure equipment, as well as the start of the production process – it is not one month. Therefore, even the most optimistic projections the shortage of Intel processors we will have to live until at least the end of the first quarter of this year, and many analysts suggest that the repercussions of the production problems will be felt throughout the first half.

Изменение цены Core i5-8400 на Amazon.com

Change the Core i5-8400 in Amazon.com

All this gave a great chance for AMD to increase market share in consumer processors, which is what she may fail to take advantage. While Intel offers took new price levels, AMD has consistently met the demand, we do not supply and kept prices at a constant level. This led to the fact that Ryzen become much more favourable purchase terms of the combination of price and consumer qualities, and it was appreciated by the buyers. As a result, in the last months of last year, retail sales of AMD processors in some regions (e.g. Germany and Russia) managed in unit sales to exceed the sales of Intel processors, which eventually led to some reduction in the market share of Intel in the segment of desktop systems with 88% in the second quarter to 85-87 % in the third-fourth quarter (exact estimates Mercury Research will appear a little later).

#Problems with the heads

Amid such serious security issues, new technological process and backorders processors a real trifle may seem to be another significant incident with Intel: in 2018, the company lost its Executive Director. Brian Krzanich (Brian Krzanich), who has worked at Intel for 36 years, started his career with the post of mechanical engineer and has come down in 2013 to directorship was in the middle of 2018 sacked and expelled from the membership of the Board of Directors due to the violation of internal rules regarding fraternization between employees.

There is a theory that under this pretext, Intel got rid of the Manager who committed a series of strategic miscalculations, for example, with the development of 10-nm process technology, but whatever it was, from June and to this day, the chief Executive officer of Intel remains vacant. Temporary management of the company assigned to financial officer Robert Swan (Robert Swann), and how many will continue the search for a permanent person to the vacant position, is completely unclear.

By the way, personnel problems are not bypassed in 2018 and AMD. Last year she lost many specialists from the field of marketing, and engineers. The majority of employees who left the company after Coduri Raja (Raja Koduri) has weakened the graphic direction of the company. But among them was valuable processor engineer – Jim Anderson (Jim Anderson), who headed the Department of computing and graphics solutions and led the improvement of the microarchitecture Zen after leaving AMD, Jim Keller (Jim Keller).

How this loss will affect future activities of AMD, time will tell.

#An overview of the main announcements

Problems, and how they had to fight the producers for a very interesting story. However, to circumvent the final article is a story about the new products that appeared on the market, it would be unfair. While it is true breakthrough technology, we showed neither Intel nor AMD. All appeared in last year new items were, by and large, secondary. Companies are preparing for a fundamental breakthrough in the next year and in 2018, they produced chips that are built on older technologies and architectures.

For this reason, the most innovative product in 2018, was submitted 12 months ago processor the most productive on the market of integrated graphics, which is the result of cooperation between Intel and AMD – Core graphics Radeon RX Vega. This Quad-core mobile processor crystal Kaby Lake merged with a graphics core Polaris and HBM2-memory, which were assembled on the CPU Board with the use of implanted in a semiconductor substrate of a bridge EMIB (Embedded Multi-Die Interconnect Bridge). As a result, the output will get processors with a TDP of 65 to 100 watts, which could offer very good graphics performance without having to install an additional discrete graphics card. These processors were adopted by HP and Dell, which came to put them in some of their gaming notebooks, and Intel, who offered the compact NUC system. But now, after a year, it becomes clear that it was rather a bold experiment, and not a mass product with a great future. Further development of the project, unfortunately, has not received, and computers, which can be detected with Core graphics Radeon RX Vega are gradually moving away from sale.

But a month later, in February, AMD already without the help of Intel has announced its own desktop processors with integrated graphics, which have taken a worthy place in the lineup of the company. Combining in a single semiconductor crystal core computing Zen and Vega graphics core, AMD has released a couple of chips for desktop – Ryzen Ryzen 5 2400G, and 3 2200G, immediately became bestsellers among the audience of buyers focused on the Assembly budget systems. The success of such processors was due to the fact that they were able to offer acceptable gaming performance in 720p resolution and four processing cores at a price in the range of $100-$170. However, to consider Ryzen Ryzen 5 2400G, and 3 2200G novelty of 2018 is still not quite correct. A similar Raven Ridge chips for mobile computers AMD announced back in the fall of 2017, so in this case it is logical to talk about the expansion of their habitat, and not about presenting a fundamentally new product.

But do something interesting, AMD could be releasing in April: this month saw the light of older processors Ryzen bimillenary series: Ryzen 2700X 7, 7 2700 Ryzen, Ryzen Ryzen 2600X 5 and 5 2600. And it is really legitimate to classify the second generation Ryzen, since they switched to the newer microarchitecture Zen+ and were produced by a more perfect process technology with the norms of 12 nm. However, for many of these chips was a disappointment. AMD did not perform any “mistakes”, did not improve the memory controller and did not reduce the latency of inter-core connections. All that could offer new Ryzen is only a 3% improvement in IPC (the number of executable per clock instructions), reached by the reduction of delays in the subsystem cache memory, and a slight increase in clock frequencies. In the end, the benefits Ryzen second-generation over predecessors lie within 10% performance increase, which at first glance was not enough to change the landscape of the processor market. But in fairness it should be noted that, despite the rather restrained progress in the technical specifications, the new processors Ryzen were still able to obtain considerable popularity. Fueled it as favorable price that AMD has set for their new and emerging shortage of competing offers Intel successfully for Ryzen second generation from the end of summer began to rise in price.

Were held in April and another announcement: I added a lineup of desktop processors and Intel. To the already existing overclocking six-core processors, Coffee Lake, the company added 35 – and 65-watt new the acceleration, which turned out to be six-core Core i7 and Core i5, Quad-core Core i3 and dual core Pentium. And I must say that some of these processors managed to attract considerable attention, at least until until all their appeal did not spoil the shortage and increased prices. Users willingly chose for their systems Junior Junior six-core and Quad-Core Core i5-8400 and Core i3-8100, which at some time could even become the best choice for not too expensive gaming systems. In addition, Intel has updated and sets of system logic, suggesting that for these processors cheap chipsets with innate support for USB 3.1 Gen2 and CNVi interface, providing a simple implementation of a WiFi controller on the motherboard.

At the same time with desktop processors, the microprocessor giant has introduced a large group of chips Coffee Lake for mobile systems, made shestiyaderny available including and laptops. Other models were also options vPro, productive integrated graphics Iris Plus, for the first time a mobile processor, class Core i9. However, in this case it is about eight computing cores is not. The first bearer of a new brand in the mobile segment has become the Core i9-8950HK – 45-watt six-core mobile processor with high clock speeds and an unlocked multiplier.

First, in 2018, the strengthening of the lineup of desktop processors Intel occurred in June, when the company introduced the Core i7-8086K – anniversary chip, the output of which is formally dedicated to the 40th anniversary of the Intel 8086, the first incarnations of the x86 architecture in silicon. However, despite all expectations, Core i7-8086K was not so interesting on the background already available on the market flagship six-core Core i7-8700K. The anniversary processor was only able to boast the achievement of a frequency of 5.0 GHz in turbo mode, but have not received any additional cores or improved internal thermal interface.

But this does not mean that the summer has not been any really noteworthy announcements. Just assumed they are not from Intel, and from AMD, which in August introduced the second generation Ryzen Threadripper. After the usual Ryzen they moved on microarchitecture Zen+, but more importantly, AMD decided to increase in your family HEDT offers the maximum number of cores from 16 to 32. Thanks to this AMD was able to seize the leadership of Intel in the maximum number of cores processors HEDT systems. And this time this advantage was completely overwhelming, and wait for the Intel processors of this class with a comparable number of cores now I don’t have to.

However, seniors Ryzen Threadripper second generation with 24 and 32 cores were very unique processors. Due to the fact that they are built on four crystals Zeppelin, the access to the RAM of which have only two crystals, these processors turned out to be strong only in the render tasks that do not require large amounts of information. In addition, such heterogeneous structure of the processor was not ready and the Windows operating system, the dispatcher which distributes threads across the cores Ryzen Threadripper not the most optimal way. As a result, being very interesting and attractive product on paper, senior Ryzen Threadripper turned out to be niche products with a very narrow sphere of applicability. What can be said about the 16-core Ryzen Threadripper 2950X – this product is really liked many professionals who appreciated offer them the best combination of cores and cost.

Similarly, offering a great combination of price and performance, AMD has made in the segment of budget solutions, budget releasing Socket AM4-APU processor Athlon 200GE. Presenting a somewhat abridged version of Raven Ridge, Athlon 200GE could boast of two cores Zen with the multithreading support in the graphics subsystem Radeon Vega 3, attractive price of $55. Opened later the same overclocking Athlon 200GE made a very interesting choice for budget builds.

At the end of summer came the announcement and new mobile processors from Intel, Whiskey Lake and Amber Lake U-Y. However, despite the use for naming their new code names, in this case we are talking only about the new Quad-core and dual-core versions of the mobile Kaby Lake with the target heat dissipation of 15 watts and 5 to 7 watts.

Real big announcements Intel we waited until October, when the market came processors Coffee Lake Refresh. Although Intel once again is not suggested for micro-improvements, continuing to exploit the design Skylake, the new processors got up to 8 computing cores and improved thermal interface between the chip and the heat-spreading lid, based on Bashlykova solder. Most surprising in this announcement, of course, was the fact that Intel doubled the number of computational cores in their senior mass offerings for literally two years.

Together with the OCTA core and shestnadcatiletnim processor Core i9-9900K was also presented and the Intel Core i7-9700K and Core i5-9600K, who redefined the basic characteristics of the representatives of the processor Intel average. As a result, Core i7 –now eight-core processors without Hyper-Threading, while Core i5 – shestiyaderny without Hyper-Threading. And this means that the technology of Hyper-threading from Intel are now gone from the main mass consumer proposals, remaining only in the flagship product for the ecosystem LGA 1151v2 and class HEDT processors.

Incidentally, simultaneously with the submission of the Coffee Lake Refresh Intel has updated and processors for high-performance workstations. But the new Skylake-X, in contrast to Coffee Lake Refresh of the increasing number of compute cores has been proposed, and senior LGA 2066-Intel Core i9-9980XE remained 18-nuclear. But according to the solder instead of thermal paste under the heat-spreading lid, the younger members of the family grew the capacity of the cache memory of the third level, and additionally, Intel no longer limit the number of lines available PCI Express processor value below $1000. In addition, by the amount of approximately from 5 to 15 % increased and the clock frequency.

At the same time held another preview: microprocessor giant has prepared aimed at workstations Xeon processor status W-3175X with 28 cores. Nominal characteristics of such a monster promise clock speed at 3.1 GHz, the peak frequency in turbo mode to 4.3 GHz and beyond reasonable dissipation typical 255 watts. Of course, in the framework of existing platforms, the performance of this processor is provided, it could not be, for he was offered a special socket LGA 3647, the appearance of which means the need for new motherboards. Currently, it is known that supported the initiative of Intel ASUS and Gigabyte, but as of today no circuit boards or processors Xeon W-3175X on sale. So we can’t even guess how many will have to pay for the possession of such a miracle of engineering.

#Conclusion: what to expect now

Apparently, no breakthroughs and high-profile announcements in the first half of this year on the processor market is expected. And AMD and Intel are already described in sufficient detail its plans and according to available data, processors with new designs start coming out closer to fall, when will start a new round of competition.

In the first half of the year can only be expected the advent of the AMD Picasso desktop – improved variants Ryzen with integrated graphics, transferred to 12-nm process technology. However, judging by what the data processor design has been presented for the mobile market, much innovation should not wait. It will be exactly the same Ryzen Quad-core graphics Vega, which is available now, just with increased clock frequencies.

Real innovation from AMD will have to wait until the third quarter, when the company plans to introduce its processors Ryzen third generation, built on microarchitecture Zen 2 and produced by a totally new CPU for 7-nm process. They can wait for a really large-scale improvements. Future microarchitecture involves the increase in IPC due to the optimizations of design and, most importantly, the extension to 256-bit block floating-point operations. The new technological process will allow to increase clock frequency. And besides, Ryzen third-generation AMD is going to resort to a modular design, in which the processor will be composed of several semiconductor crystals – chipsetov, which gives the manufacturer the possibility of a relatively simple to increase the number of cores. Therefore, it is possible that the efforts of AMD in 2019 we will be able to witness another breakthrough in the further development of multithreading when the mass processors for desktop systems will be able to offer consumers more than eight cores.

Answer whether it is a symmetrical step, Intel is not yet clear. But it is known that microprocessor giant is working on a seriously advanced microarchitecture Sunny Cove, which would become the basis of a mass processor company by the end of this year. Processors Ice Lake, which will be used for Sunny Cove, are expected to be produced at 10-nanometer process technology, which will allow you to place in the processor chip increased the number of cores. But while Intel focuses on micro-improvements, which should give an increase in IPC of up to 20 %. The promised extension of the cache memory and increasing the capacity of all the Executive pipeline, which should provide an impetus to accelerate the work of individual cores. If Intel adds desktop Ice Lake more cores, the result can be very interesting.

In other words, to be bored in 2019, we clearly don’t have. The competition between AMD and Intel will only be exacerbated, and both companies will probably continue to take a somewhat different approaches. While AMD is betting on the number of cores, Intel is struggling to increase the specific performance of individual cores. Some of the ways to increase productivity will be a more advantageous strategy for the desktop market, we will see in the relatively near future.

⇣ Content

If You notice a mistake — select it with mouse and press CTRL+ENTER.
Materials

SOURCE

Advertisement
Samsung J7 V just $5 mo. New device payment purchase req'd. Plus, free shipping.
Click this link to get the Moto Z2 Force for just $31.50/month. Unlimited and device payment activation required. Includes free shipping. Restrictions may apply.
Continue Reading

PC / Laptop

What cooling to choose when using 8-core Intel processors. Find out on the example of the ASRock Z390 Phantom Gaming 6

Advertisement
Code: FROSTY. Save 20% discount on ink orders $60 or more when you use this code at checkout plus free shipping. Exclusions apply.

It would seem that everything is obvious: if you’re buying 8-core Coffee Lake Refresh, then the motherboard will select the appropriate level. However, the Intel Core i7-9700K and Core i9-9900K introduced Taiwanese manufacturers into a stupor. Test laboratory 3DNews over the past couple of months and have looked at several motherboards based on the chipset Z390 Express, and only to MSI MEG Z390 GODLIKE we have had no claims, if it is the Converter power and heat during heavy workloads. Other devices cost from 14 to 25 thousand rubles ( ASUS ROG STRIX Z390-E GAMING, MSI MEG Z390 ACE, Z390 GIGABYTE AORUS PRO) if you use a liquid cooling system sufficiently strongly heated, especially in overclocking. And now Core i9-9900K not overclock, do not use “dropsy” and to take only the most expensive “vundervafli”? Let’s answer these questions by testing another cost of Middle-end class- ASRock Z390 Phantom Gaming 6.

ASRock Z390 Phantom Gaming 6

ASRock Z390 Phantom Gaming 6

#Specifications and packaging

Phantom Gaming, as you know, is a new series of ASRock motherboards. Model Z390 Phantom Gaming 6 belong to the middle class. In selling you will encounter a version with the numbers “9” and “4” and mini-ITX-the variation called Z390 Phantom Gaming-ITX/ac and the successor of the budget series Killer SLI — Z390 Phantom Gaming SLI. In my opinion, with the device names from ASRock came all too confusing. Basic specifications of the ASRock Z390 Phantom Gaming 6 is given in the table below.

ASRock Z390 Phantom Gaming 6
Supported processors Intel 9-th and 8-th generations (Core, Pentium and Celeron Gold) for LGA1151 platform-v2
Chipset Intel Z390 Express
Memory subsystem 4 × DIMM slots, up to 64 GB of DDR4-2133-4300 (OC)
Expansion slots 3 × PCI Express x16
3 × PCI Express x1
The storage interface 1 × M. 2 (Socket 3, 2242/2260/2280), support a SATA 6GB/s and PCI Express x4
1 × M. 2 (Socket 3, 2242/2260/2280/22110) with support for PCI Express x4
6 × SATA 6GB/s
RAID 0, 1, 10
LAN Intel I219V, 10/100/1000 Mbps
Realtek Dragon RTL8125AG, 2500 Mbit/s
Audio ALC1220 Realtek 7.1 HD
Interfaces on the rear panel 1 × HDMI
1 × DisplayPort
1 × D-Sub
2 × RJ-45
1 × optical S/PDIF
1 × USB 3.1 Gen2 Type A
1 × USB 3.1 Gen2 Type C
4 × USB 3.1 Gen1 Type A
5 × 3.5 mm audio
Form factor ATX
Price 15 500 rubles

The device is Packed in a small but colorful cardboard box. In addition to the cost, it was not a lot of accessories:

  • user guide, card Phantom Gaming, and optical media with software and drivers;
  • the cover on the rear panel of the housing;
  • four SATA cable;
  • three screw to secure the SSD;
  • HB SLI-bridge to connect two graphics cards.
 

In principle, almost everything you need to build a gaming PC in a box has been found. I would only like to see additional connection cables for RGB devices. Yet now, wherever you look — will definitely come across the “piece of tin” with light.

#Design and capabilities

Agree, it would be strange if the device-level Z390 Phantom Gaming 6 collected on the basis of a stripped-down PCB. No, in product design, ASRock uses full-size (305 x 244 mm), printed circuit Board, and this fact can not but rejoice.

As always, let’s talk about the pros and cons of wiring the main components of the device. ASRock Z390 Phantom Gaming 6 has six expansion slots. The ones closest to the CPU socket turned out to be PCI Express x1, and then comes the main PEG port which will be installed graphics card. On the one hand, everything is done right — no matter what CPU cooler you have installed, it will not collide with a 3D accelerator. At the same time, PCI Express x1 is close enough to the socket. If you use too large supercooler (for example, its Archon), then this port will not work to install a discrete device is longer than 100 mm. by the Way, all PCI Express x1 slots on the Board do not have latches.

Three PCI Express x16 is reinforced. According to the manufacturer, this metal sheath not only increases the connectors are, but also protects the contacts from electromagnetic interference. In this work the ports according to the scheme x8+x8+x4 mode for PCI Express 3.0. Of course, supports AMD CrossFire and NVIDIA SLI. All three PCI Express x16 slots are well removed from each other — you can safely install a graphics card with translatewiki coolers.

Moving on. On-Board soldered five 4-pin connectors for fans connection. In General, their location I would characterize as successful, but I am confused by their number. I would like to see the fees for 15 000 at least seven such connectors. Still buying ASRock Z390 Phantom Gaming 6 involves building a very powerful gaming system unit. In this PC can be used two – and three-piece “dropsy”, and the case with three or four (or more) fans. So be careful about buying all the necessary adapters or controllers required.

Note that the connector for CPU cooler supports CPU power up to 12 watts (1 amp). Connectors CPU/Water Pump and Chassis Fan/Water Pump Fan support 24-watt impeller and pump. Unfortunately, this model is not able to control the fan speed without PWM.

 

Sure, when you look at “six” immediately struck by the presence of three M. 2 ports. One of them is equipped with aluminum radiator. The main port is soldered between the CPU socket and PCI Express x16 slot, additional cooling is not. Apparently, the manufacturer decided that it is installed in the SSD so bad to be blown by a fan and CPU cooler. But the second SSD in the array using CrossFire/SLI with a video card he obviously not be good. This storage device will not prevent further cooling.

The upper socket (M2_1) allows you to install SSD 2242/2260/2280 formats, supports SATA 6 Gbps and PCI Express 3.0 x4. The lower M. 2 slot (M2_2), supports the installation of SSD formats 2242/2260/2280/22110, it also summed up the four lines PCI Express 3.0, and it can also operate in SATA mode. If you use M2_1 entries ports SATA3_0 and SATA3_1. If you use M2_2, turns off SATA3_4 and SATA3_5. Middle M. 2 connector with the key “E” is for installing Wi-Fi module, which in itself looks quite punning. We know that the Z390 Express chipset has native support of the wireless controller communication Intel Wireless-AC 9560.

Pads eight SATA 6 GB/s soldered in the usual ATX boards. A pair of such ports is implemented by the controller ASMedia ASM1061.

To be honest, I don’t really like it when internal ports USB 3.1 Gen1 soldered on the PCB of the motherboard, as is done in ASRock Z390 Phantom Gaming 6. You know very well that hull with blue wire connectors — very hard. During Assembly they form a large and conspicuous loop, and this affects the appearance of the system unit, especially if you use a case with a window instead of the side wall. But this is all nitpicking, nothing more. Why internal connector USB 3.1 was the first version? Some saving “on matches” turned out, honestly.

In the bottom of the Board there are pads for connecting the front panel audio connector, 12 – and 5-volt RGB strip, TPM-module, COM and two USB 2.0. One of the connectors for the backlight can be placed on top of — corps are different, for example, here such.

Continuing the theme of illumination, I note that the Board itself is equipped with RGB elements in the area of the chipset, the sound path and the plastic casing called I/O Armor.

As for the interfaces are soldered on the input panel, then I note the presence of two ports USB 3.1 Gen2, which are now implemented through the chipset. One of them is C-type. Among the video outputs are HDMI 1.4, DisplayPort 1.2, and (for some reason) D-Sub.

Wired network connectivity is provided by Intel I219-V and Realtek Dragon RTL8125AG. With the first chip we are familiar with, because it is being used in a large number of motherboards middle and high price levels. Controller Realtek Dragon RTL8125AG meets the requirements of standard IEEE 802.3 bz and is able to provide bandwidth at 2.5 GB/s. unfortunately, to test this I have not been able to.

For sound ASRock in Z390 Phantom Gaming 6 answers Realtek ALC1220. Traditionally, expensive boards in the composition of the audio included high-quality Nichicon capacitors and operational amplifier NE5532. The audio chip is shielded, and all elements of the sound system is separated from the other components of the PCB strip, which does not conduct current. Plus the left and right channels are separated by different layers of the PCB.

Not to mention the presence on the Board of the post indicator lamps and keys on and reboot the system. I last spent 10 years next to the outdoor test stand, such “trimmings” very useful.

 

The power of the CPU is provided by two connectors — 8 – and 4-pin. Don’t forget that power supply with this cable set do not occur very often, and basically we are talking about powerful devices — watts on the 700 and more. However, extra food processor is necessary in extreme cases.

On the official website of ASRock Z390 Phantom Gaming 6 is attributed to 12 phases for the CPU, because marketers Taiwanese company something should get his salary. The basis of VRM subsystem is 8-channel PWM controller uPI Semiconductor uP9521P. It turns out, the real diagram of the power Converter looks like a 5+2 because the structure of each channel allocated to the stabilization of the CPU consists of two inductors and two field-effect transistor Sinopower SM7341EH and driver uPI Semiconductor uP1965P (on the reverse side of the PCB is soldered in five pieces). The remaining two phases formed by four transistor assemblies Sinopower SM3308N, two ferrite coils and a pair of drivers from uPI Semiconductor uP1962S.

On paper and on photo the Converter power supply ASRock Z390 Phantom Gaming 6 looks impressive. Besides “ranzuki” are cooled by a fairly large aluminum heat sinks that are copper deplorable. But how efficiently the power subsystem will work together with Core i9-9900K? I will tell you that the second part of the article.

SOURCE

Advertisement
Click this link to get $200 discount on iPad. Includes free shipping. Restrictions may apply.
Click this link to get the Moto Z2 Force for just $31.50/month. Unlimited and device payment activation required. Includes free shipping. Restrictions may apply.
Continue Reading

PC / Laptop

AMD Ryzen Threadripper 2920X whether to take 12-core AMD instead of a Core i9-9900К?

Advertisement
Click this link to get $200 discount on iPad. Includes free shipping. Restrictions may apply.

When the summer of 2017 AMD has released its first processors Threadripper, many took it as a revolution in the market HEDT (high-end desktop systems). Indeed, offers Intel for enthusiast high performance for a long time did not undergo significant progress. An increase in the number of compute cores was very slow, processors kept the high price and also often had limitations on the number available for use by PCI Express. Ryzen Threadripper steel in these circumstances a breath of fresh air: they all previously adopted rules were swept aside. While Intel was getting ready to release a 10-core chip with 44 lines PCI Express, senior AMD Ryzen Threadripper for the new HEDT platform immediately offered to enthusiasts 16 cores Zen and 60 PCI Express at a price of only one thousand dollars.

And that decision worked out perfectly, despite all the shortcomings of micro-Zen. No one will argue with the fact that the specific performance per core of AMD is obviously worse than the Intel Core. But professional users still perceived multi-core AMD processors with great enthusiasm as the challenges they face to their activities, usually suggesting the possibility of efficient parallelization. The first 16-core processor family Ryzen Threadripper, 1950X, was able to gain much popularity among the digital content creators. Now, a year later, AMD decided to radically increase its offer and launched the second generation of Threadripper, which included even more impressive monster – 32-nuclear Ryzen Threadripper 2990WX. However, fresh Ryzen Threadripper the second generation had at its disposal not only the increased number of cores, but a newer design Zen+, involving the use of modern 12-nm production technology, improved performance and higher clock frequency.

However, not Ryzen Threadripper 2990WX is the main character of this material. Family Ryzen Threadripper is not from one model, and the interest in this can represent not only the older version with the maximum number of compute cores. For example, the first generation Threadripper in which the processor was built on two eight crystals Zeppelin, consisted of three processors from 8, 12 and 16 cores. In the second generation the number of crystals in the structure of the processors has doubled, now there are four, and this gave the opportunity to release a 24 – and 32-nuclear Threadripper. At the same time updated the family survived and the processors of the two crystals, which provide 12 or 16 cores. In the end, the full family Ryzen Threadripper the second generation consists of four models, the lineup grew to include seven different HEDT processors.

  Cores/ Threads Base frequency, GHz Max. frequency, GHz L3 cache MB Memory support Lines PCIe TDP, watts Price
Threadripper 2990WX 32/64 3,0 4,2 64 4 × DDR4-2933 60 250 $1799
Threadripper 2970WX 24/48 3,0 4,2 64 4 × DDR4-2933 60 250 $1299
Threadripper 2950X 16/32 3,5 4,4 32 4 × DDR4-2933 60 180 $899
Threadripper 2920X 12/24 3,5 4,3 32 4 × DDR4-2933 60 180 $649
Threadripper 1950X 16/32 3,4 4,0 32 4 × DDR4-2667 60 180 $779
Threadripper 1920X 12/24 3,5 4,0 32 4 × DDR4-2667 60 180 $485
Threadripper 1900X 8/16 3,8 4,0 16 4 × DDR4-2667 60 180 $319

Processors Ryzen Threadripper the second generation with 16 and 32 cores available for sale from August and on our website you can find a separate dedicated review. But to date, the list of available to buy Ryzen Threadripper second generation finally, was supplemented by a 12-core model 2920X and 24-core processor 2970WX. Among these processors in the first place, our attention was attracted by a younger model.

The fact that the 12-core Ryzen Threadripper 2920X has a relatively low official price of $649. And this means that the CPU do not necessarily need to be regarded as relatively affordable, but still expensive HEDT model. From the position of prices it is quite possible to oppose and mass solutions — for example, enterostomy LGA1151v2 the flagship Core i9-9900K. The recommended price of the senior mass osmeteria Intel installed for $488, but to buy at this price. Due to the current market situation, when the availability of Intel CPU are very limited, the real price of the Core i9-9900K starting from $570, and it allows us to talk about Ryzen Threadripper 2920X as his opponent without much exaggeration.

In other words, in this article we will try to assess how Ryzen Threadripper 2920X may be of interest as the basis for normal desktop top price category. It may be that the youngest Ryzen Threadripper second generation, not the Core i9-9900K is worth recommending for those users who want to swing for the maximum or extreme build.

#Family Threadripper 2: the basics

Although we plan to oppose the 12-core Ryzen Threadripper massive improvement osmeteria, we should start with a reminder that ideologically 2920X as any other processor of this family, cannot be considered a simple improved Ryzen. The AMD approach to the creation of Threadripper was to adapt the EPYC server processors for desktop systems, so the processors have received some operational features which in certain circumstances can become a problem.

Processors EPYC is monstrous not only in performance but also in size to 4096 CPU pin socket, constructed using four hidden under the CPU cover eight crystals Zeppelin. Each of these crystals has its own dual channel memory controller and PCI Express controller on the line 32, and the sum of all that agriculture provides up to 32 cores up to 64 threads up to 128 PCI Express and up to eight memory channels.

With the release of the first generation Threadripper AMD decided not to give into the hands of enthusiasts all the capabilities available in server configurations. The initial opportunity was halved: users desktop got only two working crystal Zeppelin, that is up to 16 cores, Quad channel memory and 64 PCI Express, four of which were reserved for communication with the chipset. However, in the new generation Threadripper 2 of EPYC back: models 2970WX and 2990WX were built using four crystals Zeppelin. Do not add memory channels or lines PCI Express, which were originally cut at the platform level, but has allowed AMD to increase the number of compute cores, bringing the number up to 32 pieces in the older model.

At the same time in processors Threadripper 2920X and 2950X all have remained the same. They continue to build on a couple of crystals Zeppelin — the only difference is that now it is the improved silicon, which is produced by the 12-nm technology and has a microarchitecture Zen+. Due to this, even those Threadripper 2 that do not have extra cores, got a bit increased increased IPC (number of instructions cycles) and increased clock frequency. The increase in specific productivity is achieved through optimization of the cache memory. Zen+ the latency of L1-cache is reduced by approximately 8 %, the latency L2-cache – 9% and the latency L3 cache – 15 %. Using the same advanced 12-nm process technology GlobalFoundries (12LP) allows ceteris paribus to reduce the supply voltage of 80-120 mV, which in the end provides about 200-megahertz frequency extension capability.

Despite the fact that the processors Threadripper 2920X and 2950X against their older counterparts 2970WX and 2990WX look not so impressive, they have obvious advantage. The fact that, due to the design of the platform Threadripper, the older processors with 24 cores and 32 receive crystals Zeppelin two types: those that have their own memory controller, and those that don’t have it. Therefore, the memory access is uneven and half of the cores to work with the data forced to go through additional intermediaries in the face of neighboring crystals. This leads to a significant growth delay in loading this diversity of cases is fatal, leading to performance degradation. With processors Threadripper 2, constructed of two crystals, such a problem does not exist.

When we tested Threadripper 2990WX and 2950X, all of this was fully apparent. While the 16-core 2950X always worked better than its predecessor, 1950X, due to higher frequency and some improvements in microarchitecture, 2990WX was very capricious thing. In applications that are well rasparallelivanija and do not require accesses to large amounts of data, a 32-nuclear Threadripper 2 looked like “king of beasts”, inaccessible to any other chip. However, where the algorithms are tied to the memory, Threadripper 2990WX could significantly lose performance 16-core Threadripper 2950X. In other words, the older 32-core model can only be considered as a niche solution for rendering and nothing more. With younger modifications 2920X and 2950X no such problems there, they are versatile and perfectly suited for a wide range of tasks.

Speaking specifically about the main character of this review, Ryzen Threadripper 2920X, he, unlike 2950X, even more simple processor with 12, not 16 cores. It is based, as in the 16-core Threadripper 2950X used two 12-nm crystal Zeppelin. The only difference is that 2920X in each of these crystals is blocked by two cores, one in each block CCX (CPU Complex). And this organization has its pros and cons. On the one hand, the load generating higher number of threads will require more active treatment to relations between crystals which strong place in design Threadripper be considered as definitely not. But on the other cores have at their disposal more than Threadripper 2950X, volume cache of the third level and are better able to accelerate in frequency due to the technology of avtoradgona.

#Threadripper 2920X in detail

Overall Threadripper 2920X can be considered a simple update of last year’s Threadripper 1920X with better performance. The increase is mainly the increased frequencies. Although the nameplate base frequency for a of 3.5-GHz level, the maximum frequency in turbo mode is now declared at 4.3 GHz instead of 4.0 GHz.

It should be borne in mind that the performance increase will be supported with improved technology, Precision Boost 2 (PB2), which finely adjusts the operating frequency to the parameters of the loading and current power and Extended Frequency Range 2 (XFR2), further increasing the frequency of the processor in a favorable temperature mode. The result provided a high quality cooling Threadripper 2920X managed to keep the real frequency above 3.8 GHz at boot up to 12 cores.

This is clearly illustrated in the following graph, where the documented behavior Threadripper 2920X in the nominal mode when rendering in Blender using different numbers of cores.

As shown, the Threadripper 2920X works on the average on 300 MHz frequency frequency in comparison with the predecessor of the last generation, Threadripper 1920X.

The processor power is limited to specified in the specification, the TDP of 180 watts. Frequency Threadripper 2920X adjusted so that does not exceed the established specification limits.

The difference of the frequencies of the 12-core Threadripper first and second generation allows a high degree of reliability to estimate the level of performance that can provide a 12-core new product, because it is directly related to frequency. Micro-improvement in Zen+ minor and no changes in the design and implementation of mikrotechnik interactions in Threadripper 2920X no. This processor is based on two crystals Zeppelin with six active cores, dual channel memory controller and PCI Express controller 32 lines each. The connection between the crystals is realized through Infinity Fabric with a bandwidth of 50 GB/s (when operating mode memory DDR4-3200). In other words, the internal structure Threadripper 2920X identical to the structure of 1920X.

As before, the internal topology of double-crystal Threadripper 2920X allows to use system memory UMA and NUMA modes. Mode, the default, performs UMA when the memory that is physically connected to two different crystals, collected in a single domain, which is a four-channel access. It is this balanced configuration provides the best performance in most cases.

Режим UMA (Distributed)

Mode UMA (Distributed)

However, the utility Ryzen Master users can switch to the NUMA mode, when each of the crystals runs in dual channel mode memory. In the case when the load is low-flow in nature and all active processes that belong to the same application, you can assemble within the framework of the six-core NUMA node, it reflects the latency.

Режим NUMA (Local)

Mode NUMA (Local)

#Acceleration and Precision Boost Override

Since we decided to oppose Threadripper 2920X typical desktop processor, the Core i9-9900K, subject acceleration it is impossible to ignore. Typically, the processors of class Ryzen Threadripper not differ noticeable undisclosed frequency potential. For example, 16-core Threadripper 2950X in our laboratory was able to overcome a mark of 4.1 GHz, but no more. With Threadripper 2920X the situation was similar, except that 12-core CPU managed to overclock a little better – to 4.15 GHz.

This dispersal required an increase in voltage to 1.35 V and enable Load-Line Calibration. When testing stability with Prime95 29.4 the CPU temperature did not exceed 82 degrees, and the maximum CPU power consumption was 240 watts. However, no problems with stable operation of the computer did not arise, good for heat dissipation we used a liquid cooling system MSI motherboard MSI 240 TR4 with a special waterblock that completely covers the surface Threadripper.

Speaking on the topic of overclocking, not to mention about another opportunity – function Precision Boost Override, which appeared in all processors Ryzen second generation, including new Threadripper. It can realize intelligent overclocking CPU to a fixed frequency, and dynamically so that the frequency is automatically adjusted to the current load in the process.

The point is that AMD implemented turbo mode, which is controlled by the Precision Boost technology 2, does not work by a simple formula that matches the frequency and the number of loaded cores, and relies on quite different parameters. Operating frequency is determined based on the current consumption and calculated heat dissipation, which allows you to manage performance is much more subtle.

The Boost Override function of Precision in turn allows you to change the pre-programmed standard limits for electric and thermal characteristics of the CPU. Raising them, the user can increase the aggressiveness of the turbo, allowing the processor adjusting its own frequency to take the high levels as with a weak and under multithreaded load. Such acceleration in many cases more interesting than just setting a fixed frequency, because when you use Precision Boost Override processor continues to interactively adapt the frequency under load, but makes it much bolder than in the nominal condition.

AMD said that with the installation of the CPU powerful enough cooling system performance through Precision Boost Override can be raised by about 13 percent. In other words, this acceleration is a (or even the best) analogue antalovsky the function of Multi-Core Enhancements, which, sacrificing the efficiency, also displays the CPU beyond the rated regime. And there all done so that not to create problems with the stability of the system. The user only need to accept the fact that in the name of higher performance will have to sacrifice heat and power consumption.

Strictly speaking, to configure a Precision Boost Override the motherboard BIOS or the utility Ryzen Master you need to change three parameters: PPT Limit – the maximum allowable level of heat dissipation of the processor (CPU Package); TDC Limit the maximum current which does not lead to overheating the VRM circuit Board; and EDC Limit – the maximum current that does not cause electrical overload of the VRM Board. However, the majority of motherboard manufacturers offer a simple method of modifying these parameters in one click.

For example, in the BIOS we use to test the motherboard MSI MEG X399 Creation option Precision Boost Override allows you to raise the bar allowed power consumption of the processor 300, 400 or 500 watts in one fell swoop.

For the experiment, we chose the maximum option — 500 watt, and it really increased the working frequencies of the CPU by 200-300 MHz without any damage to the stability of the system. The profile of the frequency is illustrated in the following graph, which displays the behavior Threadripper 2920X when rendering in Blender using different numbers of cores.

As you can see, activate, Boost Precision Override is allowed to transfer the operating frequencies of the test Threadripper 2920X in the range of 4.0-4.3 GHz. The most significant increase in frequency occurred in the modes with high CPU load, but at low-flow loads the processor has not lost its ability to avtoradgona. Therefore, Precision Boost Override can indeed be a better option than a simple overclock of 4.15 GHz to fixed. Especially in this case, the processor not only receives the higher rate during low-flow load, but retains the ability to fold it together with the voltage in idle moments.

Incidentally, this mode proved to be more economical overall. As can be seen from the measurement results, the power consumption of the processor when testing in Blender increased in comparison with the performance in nominal mode is not stronger than 10-15 %.

Although AMD considers Precision Boost Override a variety of acceleration and does not provide in respect of this mode are no guarantees, we are pretty sure that most owners Threadripper the second generation will want to provide them a free additional boost to performance. And deter, we will not: Precision Boost Override does not violate the stability of the system and works in such a way that the CPU temperature stays within acceptable limits. Therefore, there is no reason not to take advantage of this opportunity, especially in the case of a 12-core Threadripper 2920X, which are unlikely to overload the power system on the motherboard.

SOURCE

Advertisement
Code: VZWDEAL. Enter this coupon code at checkout to get $100 discount on Samsung Galaxy Note 8. Includes free shipping. Restrictions may apply. Device payment purchase required.
Samsung J7 V just $5 mo. New device payment purchase req'd. Plus, free shipping.
Continue Reading

Deals

Advertisement
Click this link to get $200 discount on iPad. Includes free shipping. Restrictions may apply.
Code: VZWDEAL. Enter this coupon code at checkout to get $100 discount on Samsung Galaxy Note 8. Includes free shipping. Restrictions may apply. Device payment purchase required.
Promo code not required. The deal applies to a specific group of goods. The deal is available for all customers of the e-store. The deal is available without amount restrictions

Trending