by Ilya Gavrichenkov
04/05/2011 | 12:14 PM
The launch of Sandy Bridge processors disturbed the established hierarchy of Intel platforms. Really, try to answer the following question definitively: what platform would be the best fit for those who hunt for maximum performance? It used to be fairly simple when the top processor was the one and only Core i7 Extreme Edition. However, now its positions have been threatened by daring newcomers, such as Core i7-2600K. Of course, we cannot yet state that the top Sandy Bridge processors are indisputably faster than the six-core Gulftown ones. But at the same time, there are serious reasons to believe that in some cases Core i7 Sandy Bridge and LGA1155 platform compatible with it may turn out more appealing than the top LGA1366 CPUs.
The main advantage of the LGA1155 Core i7 processors is the second generation Core microarchitecture they are based on. Numerous tests show that the improvements introduced in Sandy Bridge make these processors more than 30% faster than their LGA1156 predecessors. And therefore, they can easily catch up with the top LGA1366 processors, which will never transition to the new microarchitecture. Moreover, Sandy bridge processors may also satisfy overclockers. The top processors in this family belong to the so-called K-series, which implies that the CPUs feature an unlocked clock frequency multiplier. The contemporary 32 nm process used for LGA1155 CPUs guarantees significant frequency potential. As a result, products like Core i7-2600K may be regarded as equal competitors to LGA1366 processors from the Extreme Edition series, although they cost significantly less.
However, despite their progressive nature, Sandy Bridge processors continue being pushed into the mainstream price segment, which is the primary target destination for the LGA1155 platform. As a result, even the top Core i7 CPUs based on the new microarchitecture have only four computational cores. Moreover, the integrated Intel HD Graphics core is an inalienable part of all Sandy Bridge products. Moreover, the LGA1155 platform also has a lot of limitations: it only supports dual-channel memory mode instead of triple-channel, and multi-GPU SLI and CrossFireX technologies can only work as PCIe 8x+8x.
So, despite its significant age in computer terms, LGA1366 platform is not yet ready to give in. it is free from all these enforced restrictions that spoil the impressions from the new LGA1155 platform. Moreover, the top LGA1366 processors are six-core ones. The only issue is that all of them are Gulftowns, i.e. are based on the previous generation Westmere microarchitecture. However, Intel believes that it is still current enough and even continues to announce new products based on it, like the recently released 1000-dollar flagship LGA1366 solution - Core i7-990X Extreme Edition.
As for us, we decided to disregard all prejudices about the positioning of the LGA1366 and LGA1155 platforms and compare Core i7-2600K against Core i7-990X Extreme Edition directly. Yes, these are very different processors, one of which is more than three times more expensive than the other, but we believe it will make our today’s test session even more interesting. Anyway, computer enthusiasts base their preferences not on the manufacturer’s recommendations, but on their own idea of the product’s consumer qualities. So, today we are going to compare top LGA1155 and LGA1366 processors and will try to determine, which one of them is better suited for today’s realities: fast quad-core newcomer or experienced six-core heavy-weight leader.
Well, Core i7-2600K is the top CPU of the existing LGA1155 solutions. There are three features that distinguish it from the other processors on Sandy Bridge microarchitecture. First, it has four computational cores and supports Hyper-Threading technology. As a result, the operating system sees it as an eight-core processor, which is exactly why it belongs to the top ranks in the Core i7 family. Secondly, it has the highest clock frequency of all LGA1155 processors – 3.4 GHz. I have to say that if we leave the Core i7-990X Extreme Edition aside for a second, it will be the maximum clock frequency of all Intel desktop CPUs period. And thirdly, its L3 cache is 8 MB big. Other quad-core Sandy Bridge CPUs from the Core i5 series has a 6 MB L3 cache.
I have to say that although Core i7-2600K processors boast significantly more advanced features than the rest of the LGA1155 CPUs, they do not run any hotter than the others. Just like the top Core i5, our today’s hero stays within the same 95 W thermal envelope as its younger brothers. In other words, even before we get to the tests we can conclude that Core i7-2600K will turn out better than any LGA1366 processor at least in power consumption.
Also it is important to remember that we are going to discuss a K-series processor in our today’s article. It means that it can be overclocked by simply changing its frequency multiplier, and in this respect it can be put in the same category as Core i7-990X Extreme Edition. Note that Intel also has a regular Core i7-2600 CPU. It costs a little less but it only allows increasing its clock frequency multiplier by four points above the nominal value. And keeping in mind that LGA1155 platform doesn’t allow overclocking by raising the clock generator frequency, K-series processor is not only the more convenient, but in fact the only acceptable choice for overclocking fans.
As for Core i7-990X Extreme Edition, Intel considers it to be the today’s best desktop processor. True: it not only works at the highest frequency of all Intel processors (3.46 GHz), but also has six cores and at the same time supports Hyper-Threading technology. Besides, this CPU boasts a large 12 MB L3 cache. Overall, if you forget about the use of the older Gulftown design for a minute, then Core i7-990X Extreme Edition is a top of the line high-performance processor, no doubt.
At the same time, we can’t call Core i7-990X Extreme Edition an innovative solution: it is simply the old Core i7-980X Extreme Edition CPU launched a year ago overclocked by additional 133 MHz.
Although Core i7-2600K and Core i7-990X Extreme Edition belong to different categories, let’s try and compare their features anyway:
Well, it may seem like a strange idea to compare Core i7-2600K and Core i7-990X Extreme Edition, since LGA1366 processor is superior to the competitor in almost every aspect, including the number of pins. But in reality, Core i7-2600K may get a hand from the following three things: improved microarchitecture providing higher relative performance, higher frequency potential during overclocking, and the fact that there are not so many applications out there that can really benefit from having more than four physical cores.
Core i7-990X Extreme Edition (left) and Core i7-2600K (right)
Trying to make our investigation more in-depth, we decided not to limit ourselves with only two main heroes of our today’s article - Core i7-2600K and Core i7-990X Extreme Edition. Besides these two, we also tested the today’s current products in the price range above $300, as well as less expensive Core i5-2500K and Phenom II X6 1100T, which top the corresponding families. As a result, this article may serve not only as a practical comparison of the current top overclocker CPUs in the Core i7 family, but also as a broader comparison of expensive contemporary processors.
As a result, we ended up using the following hardware and software components for our testbeds:
We performed two comparisons between the participating processors: in nominal mode and in overclocked mode, since they are targeted for computer enthusiasts.
As usual, we use SYSmark 2007 suite to estimate the processor performance in general-purpose tasks. It emulates the usage models in popular office and digital content creation and processing applications. The idea behind this test is fairly simple: it produces a single score characterizing the average computer performance.
Right away we get the results that fit perfectly into our theory about quad-core Sandy Bridge being better than six-core Gulftown. SYSmark 2007 test shows average performance in general-purpose applications, which are often not so well optimized for multi-threading. Therefore, in this case higher relative performance of individual cores offered by the new-generation microarchitecture turned out more useful than larger number of cores. Core i7-2600K is 8% faster than Core i7-990X Extreme Edition. Moreover, the performance of our “extreme” processor turned out even lower than that of another less expensive Sandy Bridge CPU - Core i5-2500K.
I would like to add the following table to the diagram above. It contains detailed SYSmark 2007 results sorted according to the types of applications:
As you know, it is the graphics subsystem that determines the performance of the entire platform equipped with pretty high-speed processors in the majority of contemporary games. Therefore, we do our best to make sure that the graphics card is not loaded too heavily during the test session: we select the most CPU-dependent tests and all tests are performed without antialiasing and in far not the highest screen resolutions. In other words, obtained results allow us to analyze not that much the fps rate that can be achieved in systems equipped with contemporary graphics accelerators, but rather how well contemporary processors can cope with gaming workload. Therefore, the results help us determine how the tested CPUs will behave in the nearest future, when new faster graphics card generations will be widely available.
Back during our first test session featuring Sandy Bridge processors, we stressed that they would be an excellent choice for gamers. Today we confirmed this statement one more time. Sandy Bridge processors are simply unrivaled under gaming workload: they leave far behind their quad-core as well as six-core counterparts based on the previous-generation microarchitecture. So, unlike Core i7-2600K, Core i7-990X Extreme Edition will hardly be the most reasonable choice for a gaming system.
As a result, even if you intend to use a multi-GPU graphics configuration, we can recommend Core i7-2600K, despite the fact that the LGA1155 platform doesn’t support PCIe 16x + 16x for two graphics cards in their default mode. It might be a good thing to use an LGA1155 mainboard with an additional Nvidia NF200 switch that turns one PCIe x16 bus into two almost fully-functional ones. For example, Asus Maximus IV Extreme or Gigabyte GA-P67A-UD7-B3.
To test the processors performance during data archiving we resort to WinRAR archiving utility. Using maximum compression rate we archive a folder with multiple files 699 MB in total size.
Core i7-2600K and Core i7-990X Extreme Edition processors show very similar results in this test, although the LGA1366 processor is a little bit ahead of the competitor.
The processor performance during encryption is measured with an integrated benchmark from a popular cryptographic utility called TrueCrypt. I have to say that it can not only effectively utilize any number of processor cores, but also supports special AES instructions.
Encryption is a process that can be well paralleled, so six cores of the Gulftown processor, which also support AES instructions, show brilliant results. So, Core i7-2600K gets defeated not only by Core i7-990X Extreme Edition, but also by a slower six-core CPU – Core i7-970.
We measured the performance in Adobe Photoshop using our own benchmark made from Retouch Artists Photoshop Speed Test that has been creatively modified. It includes typical editing of four 10-megapixel images from a digital photo camera.
Not all the filters in Adobe Photoshop are well optimized for multi-threading. As a result, during complex image processing large number of computational cores doesn’t matter that much. So, the victory here goes to Sandy Bridge processors with advanced microarchitecture and higher relative performance, namely, Core i5-2500K and Core i7-2600K. As for Core i7-990X Extreme Edition, the performance difference between it and its vis-à-vis reaches impressive 10%.
We have also performed some tests in Adobe Photoshop Lightroom 3 program. The test scenario includes post-processing and export into JPEG format of a hundred 12-megapixel images in RAW format.
However, when it comes to batch image processing in Adobe Photoshop Lightroom, it creates completely different type of processor workload. Here Sandy Bridge microarchitecture loses to the pure computational power of six Gulftown cores. As a result, Core i7-990X Extreme Edition turns out 15% faster than Core i7-2600K.
We use Apple iTunes utility to test audio transcoding speed. It transcodes the contents of a CD disk into AAC format. Note that the typical peculiarity of this utility is its ability to utilize only a pair of processor cores.
We knew the outcome of the audio transcoding test in iTunes even before we started it. The number of cores doesn’t matter here, it is the performance of a single core that does. And this is something Sandy Bridge processors are truly superior in. That is why quite naturally the top two lines on the chart are occupied by Core i5-2500K and Core i7-2600K.
The test in Cakewalk Sonar X1 measured the time it took to mix a short test track.
And again LGA1155 processors are the leaders. It is a pity that Sandy Bridge microarchitecture is not yet used in processors with more than four computational cores. But even without extra cores Core i7-2600K defeated all competitors, including Core i7-990X Extreme Edition, in most applications, including a few professional ones.
In order to measure how fast our testing participants can transcode a video into H.264 format we used x264 HD benchmark. It works with an original MPEG-2 video recorded in 720p resolution with 4 Mbps bitrate. I have to say that the results of this test are of great practical value, because the x264 codec is also part of numerous popular transcoding utilities, such as HandBrake, MeGUI, VirtualDub, etc.
Video transcoding using x264 codec is one of those few tasks where the number of processor cores matters a lot. That is why here Core i7-2600K turns out as much as 33% slower than Core i7-990X Extreme Edition. The top LGA1155 Intel processor also falls behind the junior six-core CPU from Intel – Core i7-970. However, it is important to point out that it is nevertheless faster than the fastest six-core processor from AMD – Phenom II X6 1100T, which looks very weak in this test anyway.
The performance in Adobe Premiere Pro is determined by the time it takes to render a Blu-ray project with a HDV 1080p25 video into H.264 format and apply different special effects to it.
We see the same picture in Adobe Premiere Pro. Six-core Gulftown processors manage to retain their leadership in video processing and transcoding applications.
We use special Cinebench test to measure the final rendering speed in Maxon Cinema 4D.
The more cores - the better the performance. This principle works great in Cinebench, that is why processors with old microarchitecture – Core i7-970 and Core i7-990X Extreme Edition – win the race here.
Rendering speed in Autodesk 3ds max 2011 with both, Scanline as well as Mental Ray, was measured using SPECapc test.
This is another rendering application, but the results are again almost the same: six-core Core i7-990 Extreme Edition outperforms quad-core Core i7-2600K by about 9%. So, even though Intel has already rolled out more progressive microarchitecture, six-core Gulftown processors remain the better choice for 3D modeling and CAD applications, just like for video processing.
High performance is only one of the advantages of the new processors with Sandy Bridge microarchitecture. Processors based on it are considerably more energy-efficient than their predecessors. The LGA1366 platform, which is competing against Sandy Bridge, is, on the contrary, far from energy-efficiency. And even without the tests we understand that a Core i7-990X Extreme Edition based platform with its 130 W TDP will be much more power-hungry than the LGA1155 system with a Core i7-2600K inside, which maximum TDP doesn’t exceed 95 W. Nevertheless, it will be interesting to see how big the difference is in real tests.
The graphs below show the full power draw of the computer (without the monitor) measured after the power supply. It is the total of the power consumption of all the system components. The PSU's efficiency is not taken into account. The CPUs are loaded by running the 64-bit LinX 0.6.4 utility. We enabled all the power-saving technologies for a correct measurement of the computer's power draw in idle mode: C1E, AMD Cool'n'Quiet and Enhanced Intel SpeedStep.
Even in idle mode Core i7-990X Extreme Edition based system consumes twice as much power. The Intel X58 Express chipset also contributes to this situation, as its TDP together with the South Bridge equals 29 W. As for the LGA1155 system, it is one of today’s most energy-efficient systems for high-performance desktops, and Core i7-2600K fits perfectly into this concept.
The difference in power consumption is just as dramatic under heavy load, too. While we can’t tell definitively, which of the two processors - Core i7-2600K or Core i7-990X Extreme Edition – is the fastest, the Sandy Bridge CPU consumes 67% less power. And it means that we have an indisputable winner in terms of performance-per-watt.
The manufacturer considers Core i7-2600K or Core i7-990X Extreme Edition processors to be specialized overclocker solutions. Both these CPUs have an unlocked clock frequency multiplier that makes overclocking quick and easy. Moreover, Core i7-2600K or Core i7-990X Extreme Edition are manufactured with the latest 32 nm process. All this means that they may boast significant overclocking potential that can be revealed during overclocking. From our experience with other Sandy Bridge and Gulftown processors we expect our today’s heroes to be able to work stably at up to 4-5 GHz frequencies. This overclocking may deliver substantial performance gain that is why we decided to dedicate the entire chapter of our article to this matter.
When we tested our processors in overclocked modes, we didn’t aim at reaching maximum overclocked frequencies. Instead, we chose a different approach: we tried to identify the frequencies at which our processors could remain stable 24/7 inside air-cooled systems. We had two tests planned for each of our processors. During the first test we determined the maximum frequency, at which the CPU could work stably without increasing its core voltage above the nominal. In the second test we overclocked our processors at moderately raised Vcore (+0.15 V).
During our overclocking experiments, we only changed processor clock frequency multipliers and left the base clock generator frequency (BCLK) at its nominal value for both CPUs. This is the only possible overclocking method for Core i7-2600K, while in case of Core i7-990X Extreme Edition this approach makes things a lot easier in the long run, because it doesn’t require adjusting the multipliers and voltages for the memory, Uncore and QPI bus. We disabled Turbo mode during our overclocking experiments. We used LinX 0.6.4 and Prime95 26.5 utilities to test systems’ stability during overclocking.
As you know, overclocking causes increase in power consumption and heat dissipation, especially, if you increase the voltages in the process. However, we did our best to keep these parameters within reasonable limits. We didn’t disable any power-saving technologies and monitored processor temperatures very carefully throughout the entire session.
Well, our first testing participant, Core i7-990X Extreme Edition, worked stably at 4.26 GHz (+23% above the nominal) without any Vcore increase.
I have to say that this overclocking didn’t have any negative impact on the temperature mode: during our stability tests the processor core temperature didn’t exceed 70°C with our Thermalright Ultra-120 eXtreme cooler.
When we increased the core voltage to 1.4 V, we could get our Core i7-990X Extreme Edition to work at 4.66 GHz, which is 35% higher than the nominal.
However, our stability tests revealed that the operational temperatures grew up substantially. The hottest core was warming up as much as 90°C.
However, this didn’t cause any stability issues whatsoever.
Our second testing participant, Core i7-2600K, works at similar clock speed as Core i7-990X Extreme Edition. So, it showed very similar results during our overclocking experiments. In particular, we managed to speed it up to 4.4 GHz at the nominal core voltage.
In other words, by changing only the processor frequency multiplier we gained 29% higher frequency. As for the core temperatures, they remained very far from critical thresholds throughout all of our stability tests.
Just like Core i7-990X Extreme Edition, Core i7-2600K didn’t warm up to more than 70°C during overclocking without Vcore increase.
As soon as we raised the Vcore by 0.15 V (to 1.425 V), we managed to increase the maximum stable frequency of our test CPU by another 300 MHz (to 4.7 GHz):
In other words, in this case we got 38% past the nominal frequency.
The temperature remained at an acceptable level during the stability tests and never hit 90°C with our Thermalright Ultra-120 eXtreme cooler, which is actually no longer among the most efficient ones.
Although we overclocked processors that are at the top of their lineups, we managed to uncover their significant overclocking potential. This can be explained by the fact that Intel assigned nominal clock frequencies to their processors not based on the maximum capacities of their semiconductor dies, but in accordance with their thermal envelopes. In other words, overclocking may not affect the operational stability, but may lead to serious increase in power consumption and heat dissipation beyond the declared maximums of 95 W and 130 W respectively. However, true enthusiasts do not get scared by problems like that. Contemporary mainboards and cooling systems are robust enough.
Keeping in mind that Core i7-2600K and Core i7-990X Extreme Edition allow increasing their clock frequency by 20-40%, we decided to also check out their performance and performance scalability during overclocking. In the second part of our today’s test session we will compare the performance of Core i7-2600K and Core i7-990X Extreme Edition processors overclocked with and without Vcore increase against the performance of other overclocker processors working in similar conditions.
So, we ended up testing Intel Core i5-2500K, Intel Core i7-2600K, Intel Core i7-875K, Intel Core i7-990X Extreme Edition and AMD Phenom II X6 1100T Black Edition processors in the following modes:
In all cases we overclocked our processors by raising their clock frequency multiplier. The memory frequency in these testbeds didn’t change and remained at 1600 MHz. The testbeds configuration also remained the same as in the first part of our today’s tests.
We didn’t provide detailed comments for each of the diagrams above for one simple reason: the relative performance of Core i7-2600K and Core i7-990X Extreme Edition does not really differ from the results obtained in nominal mode. And there is nothing surprising about it, actually. Although Core i7-2600K and Core i7-990X Extreme Edition are based on two different version of Core microarchitecture, they demonstrated almost the same hidden frequency potential. So, both these processors provide overclocking enthusiasts with similar performance gains.
At the same time, I can’t help mentioning that overclocked Core i7-2600K and Core i7-990X Extreme Edition processors are simply beyond any competition. No other overclocker processors can get close to them. 32 nm process combined with contemporary Intel’s microarchitecture is a real goldmine for those enthusiasts who are on a quest for highest performance and are not afraid to have their hardware overclocked.
The power consumption tests performed on overclocked processors promise t be considerably less predictable. Although according to the conclusions we drew after our special investigation, we can assume that old rules will still work for both our today’s processors - Core i7-2600K and Core i7-990X Extreme Edition. These rules are the following: if the processor clock frequency increases at the constant voltage, the power consumption doesn’t increase too much; however, if the Vcore increases, then this parameter goes up, too.
Nevertheless, our practical experiments show that almost any overclocking doesn’t affect the power consumption in idle mode. This can be explained easily. C1E and Enhanced Intel SpeedStep power-saving technologies continue working as usual in overclocked systems. It is especially true in our case, when all overclocking is done by simply adjusting the processor clock frequency multipliers, so that they go back to the same power-saving mode as they would without any overclocking, when the system is idling. And even if we overclocked by raising the processor Vcore, it doesn’t change anything: the core voltage also drops to its original level in idle mode.
The only sad exception to this rule is Phenom II X6, which Cool’n’Quiet technology gets partially disabled during overclocking by multiplier for some reason.
High processor load easily destroys the idyllic picture we have just seen on the graph above. Here overclocked and non-overclocked processors demonstrate drastically different results. However, the obtained results still comply with the rule described above: only if you raise the Vcore during overclocking, the power consumption increases significantly. Namely, Core i7-990X Extreme Edition working at 4.26 GHz frequency and at default voltage consumes only 22 W more power than at its standard clock of 3.46 GHz. As for Core i7-2600K overclocked to 4.4 GHz without increasing its Vcore, its power consumption is 28 W higher than in the nominal mode. Any further overclocking of both processors (by additional 300-400 MHz) that requires higher Vcore setting, causes about 50 W increase in power consumption.
Overall, outstanding power-efficiency of the Sandy Bridge microarchitecture doesn’t go anywhere during overclocking. Even when we push our Core i7-2600K to its maximum of 4.7 GHz, the system with this processor still consumes less power than the Core i7-990X Extreme Edition based platform working in nominal mode.
After we completed the first tests of Sandy Bridge processors, the verdict was extremely positive. Intel engineers did a great job on the new microarchitecture, so the actual products based on it ended up with a great combination of consumer features. However, the Core i7-2600K processor we discussed today is a unique thing. Although this CPU, like all other Sandy Bridge processors, is designed for LGA1155 platform and theoretically doesn’t belong to the top product category, things are totally different in reality. The new microarchitecture turns out so strong, that Core i7-2600K managed to outperform many pricier products for a higher-end LGA1366 platform.
In other words, in the face of the new Core i7-2600K Intel created some sort of diversion that from the inside totally messes up the company’s marketing strategy in the upper price segment. And now, as we can see, there is nothing they can do about it. Even the increase in the clock frequencies of their top-of-the-line six-core LGA1366 Gulftown processors doesn’t really help. As our tests showed, the recently launched Core i7-990X Extreme Edition processor with maximally increased clock speed can only compete against Core i7-2600K in a few individual cases – in those few applications that create heavy load that could be well-paralleled.
In fact, the most expensive processor in the market, the 1000-dollar six-core Core i7-990X Extreme Edition, can be titled the today’s fastest CPU only during video processing and transcoding, final rendering and a few specific tasks, such as encryption and batch image processing. As for the majority of general-purpose applications, including contemporary 3D games, it yields significantly to Core i7-2600K. Although the latter CPU has a smaller L3 cache, it can boast much higher relative performance per core.
Unfortunately, overclocking also doesn’t help to fix the situation for Core i7-990X Extreme Edition. Even though this CPU belongs to the elite Special Edition overclocker series, the potential of the regular Core i7-2600K is just as good. While its price is three times lower, it can be easily overclocked by simply raising its clock frequency multiplier, and the maximum frequencies for this processor are not any lower than those of the Core i7-990X Extreme Edition.
In the light of everything we have just said, the “Extreme Edition” part in the name of a 1000-dollar processor becomes a little ridiculous, because now it refers only to power consumption and price.
As for the specific recommendations, at this point it would make a lot of sense to walk away from the LGA1366 platform in general and the Core i7-990X Extreme Edition processor in particular. Much cheaper Core i7-2600K will be just as good in terms of performance in the majority of popular applications, and will leave Core i7-990X Extreme Edition behind in energy-efficiency and overclocking. Of course, there are specific applications where six-core Core i7-990X Extreme Edition is still very good, but a system based on it will cost an equivalent of two Sandy Bridge PCs.
So, in our opinion, those users who are looking to buy a high-performance system, have two possible options to consider. The first one is the following: they can decide on an LGA1155 platform and Core i7-2600K processor and spend the money they save in this case on a good mainboard, fast SSD and top-notch graphics accelerator (or even two). The second option is to wait until Intel releases something new to replace their high-end LGA1366 platform, which will include new generation multi-core processors, which will undoubtedly be superior. In fact, you won’t be waiting too long. By the end of this year they should roll out new enthusiast platform – LGA2011, which will boast all sorts of technological enhancements, including two PCI Express x16 slots and four memory channels. This platform will require multi-core (more than four cores) Sandy Bridge processors, which will undoubtedly faster than their LGA1155 relatives.