It is sad, but endless battles fans AMD and Intel processors in the past. Now to answer the question, the processor which company is more profitable to buy for productive desktop system, it has become very easy. AMD almost refused to compete with Intel in the upper price segment, and therefore, dating from about 150-dollar threshold, no choice really is not. Any suitable processor for the price of the latest generation Core family, and will be the best option, and not because of some its undeniable benefits, as corny because of the lack of decent alternatives. However, amidst all this simplicity reason enough to discuss find still possible.
The fact that Intel offers at the same time not one, but several platforms for desktop computers, which may partly be similar in its scope. And we’re not talking about that on the market at the same time are preceding and following her platform, which differ in their characteristics and performance are not so noticeable. There is something just very straightforward, because if we compare, for example, LGA1155 and LGA1150, the choice is in any case worth doing in favor of the latter, newer version – more promising and though a bit, but still more productive. It pushes us and herself Intel, forming their pricing policy so that the new platform, the output of which is synchronized with the full cycle of the strategy of “tick-tock”, are not more expensive than their predecessors. In other words, pay attention to aging products when they have a more modern replacement, there is no rational sense.
Same logical reason for the choice of torment arises because for the most productive desktop Intel offers not familiar commonly used LGA1150-generation processors Haswell, a specialized elite CPU design Haswell-E in kvaziservernom form factor LGA2011-v3. Formally platform LGA1150 and LGA2011-v3 in terms of positioning do not intersect, because Intel has tried to separate them into different price segments. But in fact, the difference in cost for older CPU LGA1150 (Core i7-4790K) and primary processor LGA2011-v3 (Core i7-5820K) is not so fundamentally – and it does not exceed 15 percent. Of course, this should be taken into account and the fact that she related ecosystem for LGA2011-v3-processor will cost a bit more expensive, however, and the difference in price to many may not be crucial. Therefore, here and there you can hear is a reasonable question as to which version of the system is better to choose for a particular purpose – many users are willing to seriously consider the Core i7-5820K in performance LGA2011-v3 as an alternative for the Core i7-4790K. And if so, we decided to dedicate this issue a separate study.
Way Intel to release it such as Core i7-5820K, Jr. processor for high-performance platform was not easy. His predecessor, the first of which appeared in early 2012, the family of Sandy Bridge-E, Core i7-3820, was a four core, does not even have an unlocked multiplier. In the next generation, Ivy Bridge-E, Jr. Core i7-4820K processor was overclocking, but I still like the older Core i7 processors in LGA1155-performance, satisfied with only four cores. However, the most significant step forward was made only with the commissioning of the platform LGA2011-v3 and processor design Haswell-E, whereby the Core i7-5820K became a full six-core CPU, has a complete set of overclocking features. Thus, today’s low-end processors to Intel’s high-performance desktop platform can finally boast a fundamentally better performance than the older vernacular processor platform – if only because that in it one and a half times more processing cores.
Meanwhile Core i7-5820K and interesting. Respect for elders for LGA1150 processors it is representative of the other weight class, but its recommended price is set at the level of only $ 389, while the Core i7-4790K is estimated by the manufacturer only $ 50 cheaper – at $ 339. Very small margin for two additional cores, is not it? But in addition, Core i7-5820K boasts more capacious memory cache in the third level, the volume of which is 15 MB.
However, in the presence of additional cores there is a downside – lower clock speeds. Passport frequency Jr. Haswell-E is set at a modest level – 3.3 GHz, 700 MHz clock frequency is less than Core i7-4790K. Of course, with a Core i7-5820K implemented technology Turbo Boost, but it can only increase the frequency to 3.6 GHz, while senior Devil’s Canyon activation Turbo accelerates to 4.4 GHz.
As a member of the Core i7-5820K platform LGA2011-v3, and is easy to guess about his other fundamental differences from the LGA1150-processors. Firstly, the youngest member of the family Haswell-E quad is entirely appropriate memory controller supporting DDR4 SDRAM. But to call it a clear advantage after all it would be wrong. At this stage, DDR4-memory distributed fairly weak, and tangible productivity gains, it does not. Same margin, which will pay for the novelty of such a memory will be, at best, 50-60 percent.
Secondly, Core i7-5820K compared with conventional Haswell has developed a controller PCI Express 3.0, not supporting 16 and 28 lines. This can be useful for gamers that use multi-GPU-configuration, or for enthusiasts who want to create a performance of the disk subsystem is based on RAID-controller or server-class solid state drives with interface PCI Express. Moreover, only the LGA2011-v3-processors (and Core i7-5820K among them) can ensure the functioning of the three PCI Express 3.0 slots simultaneously, which allows you to collect in the system based on these three-component SLI- or CrossfireX-configuration. However, it should be borne in mind that more expensive than the Core i7-5820K, processors for LGA2011-v3 provide the user with an even greater number of lines PCI Express – 40. But for the vast majority of cases available in the low Haswell-E 28 lines will be enough . For example, the difference in the two-speed multi-GPU-system with a full scheme of slots PCI Express 16x + 16x and provides the Core i7-5820K version of PCI Express 16x + 8x virtually invisible.
Speaking of that, what other differences there are in Haswell-E and Haswell, not to mention the lack of a high-performance processors for the LGA2011-v3 integrated graphics core. Of course, users of the upper price range it does not disappoint, but there is one subtlety. No built-in GPU also means the lack of function Quick Sync, which could be useful for the Core i7-5820K, as it allows to carry out very fast and inexpensive video transcoding.
However, if you sum up all the above, the Core i7-5820K looks very attractive proposition. This processor is only slightly inferior in their performance 600-dollar Core i7-5930K, which means that from it we can expect roughly the same performance as that of the recent flagship, Core i7-4960X Extreme Edition generation Ivy Brige-E. But now that the 1 000 dollars enthusiasts are offered eight cores, the price is so high computing power has declined substantially – shestiyadernik become much closer to the people.
As follows from the given table, Core i7-5820K operating frequencies are in the range from 3.3 to 3.6 GHz. But according to the diagnostic utility CPU-Z, real frequency of the Core i7-5820K at high load almost always is 3.4 GHz.
If the load falls on one or two cores, this may increase the frequency up to 3.6 GHz.
Operating voltages from Core i7-5820K, as well as the older processors Haswell-E, are low: about 1,03-1,08 V. Uncore-processor portion, which includes including L3-cache and memory controller in the nominal mode clocked at 3.0 GHz. This frequency is the same for all members of the series Haswell-E.
Curiously, the Core i7-5820K, as well as two other models Haswell-E, based on an eight semiconductor chip with an area of 356 mm2. However, the pair of cores together with their corresponding part of the cache is disabled at the production stage, and in fact we are talking about the fact that in shestiyadernik used culling from the production of eight-processor. This fact may explain very similar to the behavior of all processors Haswell-E, which, for example, is shown at their dispersal.
Despite the fact that the Core i7-5820K – is the youngest representative in the lineup Haswell-E, it has all the same overclocking capability as its more expensive counterparts. That is, first, the frequency multiplier is unlocked, and secondly, not fixed and factors responsible for the formation of memory and frequency Uncore-block. In addition, the processor allows you to choose between three options for the base frequency – 100, 125 or 166 MHz, which are optimized for frequency dividers tires DMI and PCI Express. As the supply voltage and all the cores of adjacent nodes with them, it is formed integral to the processor stabilizer.
Similarly, a senior fellow in the Core i7-5820K and arranged his body: diffuser cover is attached to the semiconductor chip by soldering, which is considered as the best option in terms of the efficiency of the heat sink. And in this case it can be argued with confidence: Japanese enthusiasts trying to scalp a copy of the processor and not only personally convinced of the presence of solder under the hood, but also captured the seen for a commemorative photo.
However, despite all this, we can not say that the Core i7-5820K could please us the results of their dispersal. The fact that it is based on exactly the same semiconductor chip as a senior Haswell-E, caused its similar behavior with increasing frequency and voltage beyond their nominal values. Unfortunately, excessive heat, which we noted when testing the Core i7-5960X and Core i7-5930K, again made it impossible to achieve however some outstanding results. And we use for heat removal is one of the best air coolers – Duplex Tower Noctua NH-D15 – before the Six-Core microarchitecture Haswell was powerless even at only 25 percent overclocking.
As a result, stable operation Core i7-5820K was possible only at a frequency of 4.1 GHz and with an increase in the supply voltage up to 1,225 V. The further acceleration c such voltage leads to a loss of stability, and increase the voltage has not been possible due to overheating occurring at the same time processor. During the same passage of stability tests with a frequency of 4.1 GHz, our sample temperature Core i7-5820K reached 95 degrees, which can be considered for Haswell-E is quite permissible temperature regime, as these CPU throttling is activated by heating to 105 degrees.
Notice to test overclocking tool we use LinX 0.6.5 with support AVX2-instructions – this is the program for checking overclocking LGA2011-v3-systems well. The fact that AVX2-instruction while not widespread in commonly used programs, but they cause rampant heating cores microarchitecture Haswell. This means that once in LinX 0.6.5 Core i7-5820K processor maintains its stability, the problems in other cases, there will almost certainly not.
Unfortunately, the overclocking potential of the Core i7-5820K was not better than those tested in our laboratory before the Core i7-5960H and Core i7-5930K. All this clearly indicates that all Haswell-E – birds of a feather. And the fact that the representatives of the family of six-core crystals are used, for whatever reasons, have not been chosen for the full eight-modifications are well explained by the low acceleration junior Core i7-5820K. In other words, we obtained during overclocking experiments, a 25 percent increase in clock speed – not an unfortunate exception to the rule, but it is a natural result that reflects the general trend. In terms of maximum achievable under normal conditions of clock frequencies Jr. shestiyadernik platform LGA2011-v3 seriously inferior to the older Devil’s Canyon, even though they are based on the same microarchitecture Haswell.