Fully-Fledged LGA1155 Platform: Intel Z68 Express Chipset Review

Intel has finally launched the last chipset in the Cougar Point family, which should meet all the needs of enthusiasts who prefer LGA1155 processors. This core logic set has full overclocking functionality, supports Intel Quick Sync, Intel Smart Response and Lucid Virtu technologies. Looks like Sandy Bridge becomes even more appealing...

by Ilya Gavrichenkov
05/26/2011 | 12:54 PM

No doubt that Sandy Bridge processors launched in the beginning of this year is a great success on Intel’s part. The company engineers managed to significantly optimize progressive Core microarchitecture and substantially increase the performance lowering the power consumption at the same time. As a result, we here at X-bit labs as well as other reviewers declared LGA1155 platform to be the today’s best choice for mainstream desktop computers.

 

However, the brilliant debut of the enhanced microarchitecture was darkened by an entire myriad of unfortunate circumstances. First of all, LGA1155 platform turned out not as overclocking-friendly, as the previous solutions. The so-called “unlocked” CPUs were necessary in order to get the system working at frequencies other than the nominal ones. Secondly, Intel’s marketing team wasn’t very successful in differentiating between platforms targeted for enthusiasts and regular mainstream users. As a result, there was no universal chipset with all-embracing functionality among the first wave of core logic sets for Sandy Bridge platform. You had to sacrifice overclocking if you wanted to use the integrated graphics core, and the other way around. And thirdly, shortly after the LGA1155 launch they discovered an issue with the entire Cougar Point chipset family, which required recalling first chipset shipments and replacing them with the new chipset revision.

In other words, until recently the CPU components was theoretically the major appeal of the LGA1155 platform, but the existing Sandy Bridge chipsets scared away the potential customers. Luckily, not only the users understood where it was coming from, but Intel guys also started getting concerned about the chosen strategy. That is why back in early January when the processors with new microarchitecture were launched we already knew that the manufacturer would later offer not only P67 and H67 chipsets (including their modifications), but also some advanced universal Z68 chipset, which should undoubtedly make the new platform a true hit.

And finally this day has come: first mainboards based on the long-anticipated Z68 tart appearing in retail. Do they prove up to everyone’s expectations? Today we are going to try answering this question in detail, because the situation seems to be very complicated. On the one hand, Z68 combined the features and functionality of P67 and H67, at the same time allowing to overclock the processor and use the integrated graphics core. Moreover, it also boasts a few additional functions, such as using the integrated graphics core and an add-on graphics card sat the same time as well as Intel Smart Response Technology. But on the other hand, Z68 still lacks many demanded features. For example, it doesn’t support USB 3.0 ports, can deliver only two 6 Gbps SATA ports, and besides, it doesn’t allow overclocking processors by raising their BCLK frequency, just like its predecessors. However, there is really no other choice available at this time. Intel Z68 is the best core logic set for LGA1155 platform and there won’t be any other chipsets for Sandy Bridge processors.

Closer Look at Intel Z68

Intel’s marketing department tried to introduce Z68 as a completely new solution targeted for enthusiast segment. In the meanwhile, a closer look at its features reveals that it is very similar to Intel H67 that allows using integrated Intel HD Graphics 2000/3000 core.

As you can see from the flow-chart, Z68 doesn’t have any new peripheral interfaces compared with H67. Just like its predecessor, the newcomer supports eight additional PCI Express 2.0 lanes, 14 USB 2.0 ports and 6 SATA ports, two of which can work at 6 Gbps speed. In other words, the new chipset doesn’t have any unique peculiarities from the architectural standpoint. Moreover, Z68 chips are fully pin-compatible with all other LGA1155 chipsets.

So, the uniqueness of Z68 is achieved in the following two ways: by allowing access to everything available in Cougar Point chipset family right from the start and by providing enhanced software support. Here I would like to remind you that P67 chipset could be used for CPU overclocking, but didn’t support integrated graphics. The alternative H67 chipset, on the contrary, allowed using the graphics core integrated into the processor, but didn’t have the option to adjust the processor clock frequency multiplier. Z68 has both: it supports integrated graphics and can overclock processors.

As for the software component, Intel paired the launch of their new Z68 with the announcement of two new technologies. First, they enabled dynamic switching between the discrete graphics accelerator and the integrated graphics core inside the processor. This served as green light for Virtu technology from Lucid. The main idea here is to alternate the use of the add-on graphics card or Intel HD Graphics 2000/3000 core depending on the graphics sub-system utilization at a given moment of time. Second, Z68 supports Smart Response Technology that allows using a small SSD drive to cache the operations of the storage sub-system built with regular conventional hard drives.

As a result, the specifications of major LGA1155 chipsets look as follows:

Although Z68 boasts richer functionality than any other LGA1155 chipset, we can’t call it all-embracing. We were particularly upset that there was no native USB 3.0 support and very few 6 Gbps SATA ports. It is the mainboard makers again who will have to fix these issues by installing additional controllers onto their products.

I have to say that it looks like Intel will soon try to fully replace P67 with Z68. The price of these chipsets differs by only a few bucks, but the new core logic set boasts broader functionality, which should encourage mainboard makers to prefer Z68. At the same time, integrated graphics support is not a must for the new chipset, so the mainboards based on it do not have to have monitor outs at all. As for the compatibility with discrete graphics accelerators, Z68 is equivalent to P67: it supports one graphics card with PCI Express 2.0 x16 interface, or two graphics cards connected as PCI Express 2.0 x8 + x8.

From user standpoint, clean record is a great advantage the new Z68 has to offer. Since its launch was delayed until May, it doesn’t suffer from degradation of the integrated SATA controller. All existing Z68 chipsets are based on the issue-free B3 revision.

Asus P8Z68-V PRO Mainboard

We are going to introduce the features and functionality of the new Intel Z68 chipset to you with the help of a product from one of the leading mainboard makers - Asus P8Z68-V PRO. For pretty obvious reasons this mainboard is a direct successor to a popular Asus P8P67 PRO, which we already discussed in one of our reviews. However, Asus engineers decided not to copy the existing design of the Intel P67 based product. Overall, P8Z68-V PRO is similar to P8P67 PRO, but if we take a closer look at both of them, we will immediately notice dramatic differences.

The similarity between Asus mainboards on P67 and Z68 is determined by the close characteristics of the two chipsets. Therefore, Asus P8Z68-V PRO is compatible with the same LGA1155 processors and the same DDR3 SDRAM as its predecessor. The number of expansion slots and their configuration also remained the same. P8Z68-V PRO offers three PCI Express x16 slots, which operation mod depends on the number of graphics cards in the system. The blue slot works in x16 mode when there is one graphics card installed, and switches to x8 mode together with the next white slot in case of two cards. The black slot connected to the chipset and not to the CPU works as x1, but it can also be switched to a faster x4 mode if necessary, although in this case the mainboard will disable other PCI Express x1 slots, two USB 3.0 and eSATA ports.

As for the differences, you can immediately notice them in the processor voltage regulator circuitry. Although all power components are hidden beneath the shaped aluminum heatsinks, the digital processor voltage regulator circuitry on P8Z68-V PRO has more phases – the total of 16. However, the additional four phases appeared here for quite explicable reasons: they power the processor graphics core, which is simply not powered at all on Intel P67 based boards.

It is highly positive that the implementation of additional phases didn’t eat up any of the free space around the LGA1155 socket. There should be no problems installing large CPU coolers on Asus P8Z68-V PRO. Moreover, Asus engineers decided to increase fan connectivity in their new mainboard and located 6 power connectors with two four-pin connectors assigned to the CPU cooler alone.

The chipset is cooled passively with a short aluminum heatsink on top of the chip that takes up a lot of space. Some users complained that Intel P67 heated up a lot and the new chipset didn’t get any better in this respect. The chipset heatsink warms up substantially during work.

Speaking of the innovations introduced in P8Z68-V PRO, we can’t help mentioning that Asus decided to give up the popular NEC USB 3.0 controllers used all over the place these days. Now the four USB 3.0 ports available on this mainboard are provided by Asmedia chips with slightly higher performance. However, changing the chip maker didn’t help solve the native support issue: all USB 3 devices won’t be available until the drivers have successfully been loaded.

Asus P8Z68-V PRO has eight SATA ports, six of which are provided by the chipset, and the remaining two are connected to Marvell SATA III controller. So, half of the SATA ports on this mainboard support up to 6 Gbps data transfer rate.

There are six USB 2.0 and twp USB 3.0 ports on the back panel alongside with an eSATA port, an RJ-45 Gigabit network connector, six analogue audio-jacks, an optical SPDIF and a Bluetooth module. There are also monitor outs – an analogue D-Sub and a digital DVI and HDMI 1.4. I would also like to remind you that Intel HD Graphics 2000/3000 core can work with two monitors at the same time, so any two out of available three monitor outs may work in tandem.

In addition to the back panel connectors and ports there are pin-connectors for two more USB 3.0, six more USB 2.0 and two IEEE1394 ports. By the way, even though there are no FireWire ports on the back panel, Asus didn’t include the corresponding bracket among the mainboard accessories. So, if you decide to use IEEE1394 devices with Asus P8Z68-V PRO mainboard, you will need to purchase the corresponding bracket separately.

Although Asus P8Z68-V PRO is not targeted for computer enthusiasts, the developers provided it with convenient Power On and Reset buttons anyway. Moreover, in its regular spot we find a great MemOK! button that allows to boot the mainboard even if the memory sub-system settings are incorrectly configured.

Here is our traditional specifications table for Asus P8Z68-V PRO:

The BIOS of Asus P8Z68-V PRO is like a twin brother of the BIOS of the predecessor – Asus P8P67 PRO. The new board uses EFI BIOS with a traditional graphics interface. As we have expected, all overclocking-friendly features have been simply moved into the BIOS of the new Intel Z68 based board without any changes.

However, this is hardly a surprise. It is much more interesting that among the options for setting the processor frequencies and voltages are similar settings for frequency and voltage of the graphics core built into the processor.

 

 

There is also a new section dealing with the initialization order for the graphics adapters, which allows enabling multi-GPU mode involving the integrated graphics core in the processor.

The hardware monitoring section also looks richer due to the fact that the board supports more fans now.

 

Testbed Configuration

The new Intel Z68 chipset is a very interesting review subject in comparison to the predecessors as well as individually. For our today’s test session we bundled our Asus P8Z68-V PRO mainboard with a Core i5-2500K CPU, 4 GB of memory and an AMD Radeon HD 6970 graphics card. For comparison purposes we also used an Intel P67 based mainboard – Asus P8P67 PRO.

As a result, our testbeds included the following hardware and software components:

Overclocking

We decided to start our test session with overclocking, because the new Intel Z68 chipset is particularly desired on the enthusiast community, who may be pinning too many hopes upon it.

I would like to remind you that LGA1155 platform architecture implies that we use the same clock frequency generator to form all system frequencies, starting with the CPU and memory clock and finishing with SATA and USB controller frequencies. It means that LGA1155 systems allow changing the base clock generator frequency (BCLK) in a very minor range, so that the peripheral bus controllers could handle it without issues. Therefore, the only way you can more or less substantially overclock processor and memory in Sandy Bridge systems is by adjusting the frequency multipliers for the CPU, memory and graphics core.

Here I have to point out right away that the release of Intel Z68 chipset doesn’t change anything in this respect. There is only one clock generator and the maximum BCLK speed when the system remains stable is only 5-7% higher than the nominal 100 MHz. For example, our Asus P8Z68-V PRO mainboard could work stably at 106.5 MHz BCLK frequency, but as soon as we hit 107 MHz, the board wouldn’t even boot.

It means that Intel Z68 doesn’t offer anything principally new to overclockers. And the only improvement of its overclocking functionality is the fact that this chipset provides simultaneous access to all available multipliers: for the CPU clock, memory clock and for the Intel HD Graphics 2000/3000 core integrated into the processor.

As for the way it works, everything is exactly the same as it used to be. Overclocker K-series processors can be overclocked without limitations, non-K series Core i7 and Core i5 CPUs allow increasing their multiplier only by 4 points above the nominal with Turbo mode intact, Core i3 and junior Sandy Bridge modifications cannot be overclocked at all. At the same time, the memory frequency may be increased to 1600/1866/2133/2400 MHz for all processors, and the graphics core may be overclocked indefinitely in 50 MHz increments.

In reality, we didn’t find any significant changes during CPU overclocking compared with our experience using Intel P67 based boards. We overclocked our Core i5-2500K processor with the voltage increased by 0.125 V to the same 4.7 GHz as before.

In other words, you shouldn’t expect the new Intel Z68 to work any overclocking wonders. In terms of overclocking, mainboards based on this chipset aren’t any different from the older models on Intel P67, and the actual advantages of the new core logic set will shine in a completely different field.

Performance

Chipsets in contemporary systems have very little effect on performance. It is so because the contemporary chipset is nothing more but a South Bridge with numerous controllers for external devices inside. The components that play integral part in performance, such as computational cores, graphics bus controller and memory controller, are currently inside the CPU. Therefore, no one expected that the launch of the new Intel Z68 would bring any performance improvement to the LGA1155 platform.

Nevertheless, we decided to compare the performance of similar systems built on the same type of mainboards with Z68 and P67. They are Asus P8Z68-V PRO and Asus P8P67 PRO. The tests were run twice: with a CPU working at its nominal frequency and with the CPU overclocked to 4.7 GHz by raising the multiplier. In nominal mode we kept Turbo Boost and Intel Enhanced SpeedStep technologies up and running 9these are the technologies allowing interactive adjustment of the processor frequency). During overclocking Turbo Boost was off, but Intel Enhanced SpeedStep kept working as usual.

To estimate the average platform performance PCMark 7 measures the speed of typical real-life algorithms that are very popular in every-day tasks.

The additional Computation index shows how fast systems can cope with resource-hungry video and image processing tasks.

3DMark 11 test focuses on the graphics sub-system performance in the first place.

However, besides the total score, 3DMark 11 also generates another interesting score – Physics rating. This score is obtained in a special physics test that emulates the work of complex gaming mechanics with a large number of objects.

To test the performance during data archiving we took WinRAR archiving utility. Using maximum compression rate we archived a folder with multiple files 1.1 GB in total size.

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.

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.

We use special Cinebench test to measure the final rendering speed in Maxon Cinema 4D.

In addition to that, we tested mainboards on different chipsets in popular 3D games.

All obtained results indicate unanimously that there is no performance difference between mainboards for Sandy Bridge processors built on different chipsets. All differences lie within the acceptable measuring error margin.

Power Consumption

Theoretically, the typical heat dissipation of all Cougar Point chipsets, including the new Z68, is the same and equals 6.1 W. However, practical power consumption measurements taken for P67 and Z68 based mainboards revealed pretty interesting peculiarities.

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 C1E and Enhanced Intel SpeedStep power-saving technologies for correct measurement of the computer's power draw in idle mode.

Even in idle mode Asus P8Z68-V PRO and Asus P8P67 PRO mainboards with identical functionality and performance consume different amounts of power. The Intel Z68 based solution is a few watts more resource-hungry than its predecessor.

The newcomer consumes noticeably more power under single-threaded load as well as during full CPU utilization.

So what’s the matter? It is important to remember what we said when we studied the voltage regulator circuitry layout on the new Asus P8Z68-V PRO mainboard. It has four additional phases powering Intel HD Graphics 2000/3000 core. P67 based mainboards do not have these phases at all.

It means that mainboards based on the new Intel Z68 chipset supporting integrated graphics power the graphics core anyway, even if the add-on graphics card is used. In fact, this is exactly what Lucid Virtu technology is based upon (we are going to talk more about it later in this review). As for Intel P67 based mainboards, they do not imply that the graphics core integrated into the processor may be involved at all, so they do not send any power to it. As a result, P67 based systems turn out a few watts more energy-efficient in any case.

Lucid Virtu Technology

So, our power consumption tests showed that the new Intel Z68 chipset doesn’t disable the graphics core integrated into Sandy Bridge processors even when the system used a discrete graphics card. Of course, if you have a fully-functional graphics accelerator in your system, it results into a frustrating loss of power, but at the same time it has a few indisputable advantages. In order to get to get to these advantages Lucid Company, which specializes on introduction of hybrid video systems, developed the so-called Virtu Technology. The idea behind it is to open access to Sandy Bridge graphics core capabilities even if there is a discrete graphics accelerator in the system.

You may believe that Intel HD Graphics 2000/3000 graphics core should hardly be of interest to the owners of high-performance graphics accelerators, because contemporary AMD and Nvidia GPUs provide way higher 3D performance anyway. However, there is one very good reason for the opposite: the graphics core inside Sandy Bridge contains not only traditional graphics units, but also Intel Quick Sync technology, which currently has no analogues from any of the discrete GPU makers.

This technology provides Intel HD Graphics 2000/3000 with a hardware encoder and decoder that should encode and decode high-definition video in popular MPEG2, VC-1 and H.264 formats. And the performance of these specialized units is so high that when we use them for video transcoding the performance gets several times higher compared with what we would get using software transcoding with AMD Stream or Nvidia CUDA.

Lucid Virtu technology contains special software that allows interactively switching the load between the processor graphics core and the discrete graphics card depending on the application you are running. In fact, Lucid Virtu is a purely software tool, but at the same time it is closely connected with the Z68 chipset. In Virtu the Intel Z68 chipset supports on the hardware level those multi-GPU configurations that allow using discrete graphics cards together with the graphics core integrated into the processor.

Lucid Virtu technology supports two operational modes:

Both these modes use the same software where you should manually select the applications, in which the system should switch to the secondary graphics accelerator. There is a list of applications that require access to Intel Quick Sync in d-Mode, and a list of games that should be rendered on a discrete graphics card in i-Mode.

 

Of course, d-Mode seems to be the most suitable for everyday use. In fact, it simply provides the owners of high-performance graphics accelerators with access to Intel Quick Sync, but at the same time doesn’t impose any limitations on the discrete graphics accelerator.

At the same time Lucid can’t stress enough that i-Mode is not totally useless, but makes a lot of sense in terms of lowering the power consumption, because in this case a power-hungry discrete graphics card is in passive mode most of the time. But unfortunately, the graphics card doesn’t get totally turned off and keeps working in idle mode. Therefore, we can argue about energy-efficiency and power savings in this mode.

Of course, we had to check out Lucid Virtu technology in real life. We were overall pleased by the approach offered by Lucid software developers, however, we were still able to reveal a few frustrating drawbacks. They were particularly obvious in the i-Mode. Some games have compatibility issues in this mode: they would still run on the integrated graphics core despite the Virtu settings, or would simply crash during work. Things are much better in d-Mode in this respect: the system works stably almost all the time in this mode. Besides, there are almost no limitations concerning the graphics cards in this mode: you can even use SLI and Crossfire configurations.

We decided to study the system performance in 3D games with Lucid Virtu technology and without it, i.e. in i-Mode, d-Mode and with Virtu completely off. We used a discrete Radeon HD 6970 graphics card. We set 1920x1080 resolution in all tested games and used only maximum image quality settings.

The numbers speak for themselves. Enabling Virtu causes the gaming performance to drop in any case. However, in d-Mode, where discrete graphics accelerator plays the dominant part, the fps rate drops only by a few percents and is, actually, barely noticeable. Things get much worse in i-Mode. The speed drops substantially in some games because of the need to additionally transfer the rendered images along the PCI Express bus. And, frankly speaking, i-Mode seems to be totally unacceptable for gamers because it doesn’t allow you to get the best of your graphics sub-system.

However, could our opinion about the i-Mode and d-Mode change after the transcoding speed test? To answer this question we checked how fast the system could transcode a 3 GB 1080p movie in H.264 format (a 40-minute TV show episode) into lower resolution for viewing on an iPhone 4. We performed transcoding using popular commercial utilities supporting Intel Quick Sync technology: Cyberlink MediaEspresso 6.5.1704_37777 ? ArcSoft MediaConverter 7.1.15.55.

Here Lucid Virtu technology works impeccably. Transcoding speed is hardly affected by enabled Virtu neither in d-Mode, nor in i-Mode. Systems with Virtu take about the same time to transcode as systems with integrated Intel HD Graphics 2000/3000 core. So, due to the use of Intel Quick Sync Virtu really can significantly speed up HD video transcoding in Sandy Bridge systems with an external graphics accelerator.

However, there is also the third aspect – power consumption. Let’s see what will happen to this parameter when we enable Virtu, especially, since the developer – Lucid – promotes the i-Mode as an optimal solution for users who care about energy-efficiency.

It is obvious that Virtu technology cannot lower the power consumption of a system equipped with a discrete graphics accelerator to the level of an integrated LGA1155 platform. As for the energy-efficiency of the advertised i-Mode, it only shows during video transcoding or playback or at the expense of a catastrophic performance hit. In all other cases we didn’t see any noticeable improvement. The reason for that is because the discrete graphics accelerator is never fully off even in i-Mode: even though it may be idle, it is still powered on. As for 3D modes with enabled Virtu technology, the discrete graphics card is used in exactly the same way as during its normal gaming operation, and the alleged energy savings are only achieved at the expense of performance.

So, i-Mode is in fact of pretty limited interest, because it lowers the 3D performance considerably, but doesn’t really offer any advantages in return, with the exception of some energy savings in several specific cases, like during HD video playback or transcoding. d-Mode, however, looks much more appealing. The 3D performance losses here are minor, but video transcoding is performed via Intel Quick Sync allowing to take real advantage of this technology even with a discrete graphics card in the system.

Intel Smart Response Technology

We decided to save the most interesting feature of the new Intel Z68 for last. The new chipset didn’t impress us with its overclocking potential, and Virtu technology seems to make sense only for those gamers who often transcode video. Therefore, Intel added another feature to their new Z68, which may become the major argument in favor of the new core logic set. This technology is called Intel Smart Response or simply SSD-caching, which is the name used previously.

Flash-based storage devices continue to expand their market presence with great success. SSD-drives have one indisputable trump – high speed. Therefore, SSDs are very welcome guests in any computer system: any user will immediately feel that with SSD applications load faster and files open quicker. However, SSDs are still pretty expensive, which is preventing a lot of users from replacing their conventional HDDs with the solid-state drives.

No wonder that in this situation many users choose the “transitional solution”, when there is a small and fast SSD functioning as a system drive and a slow large-capacity HDD used to store all the data. Having analyzed the situation, Intel decided to come up with a solution for these rationalists. Intel Smart Response technology implemented in Intel Z68 should create even more optimal conditions for small and fast system SSDs. Intel suggested turning them into a high-speed cache for the slow disk sub-system of conventional HDDs. In other words, Intel Smart Response allows combining SSDs with traditional hard drives within the same storage sub-system and enjoying maximum advantages of this symbiosis by automatically duplicating the most frequently requested data in a fast SSD. As a result, you will be able to access the most popular data stored in the SSD cache very quickly, which should dramatically improve the overall system responsiveness.

In fact, just like Lucid Virtu, Intel Smart Response is a software technology. It is implemented in the new Intel RST driver version 10.5. However, Intel has strictly tied up their Smart Response technology to the chipset: it won’t work on those LGA1155 mainboards that aren’t based on the Z68 core logic set. But at the same time Intel Smart Response will be supported in some mobile platforms on Intel HM67 or QM67 chipsets.

Intel Smart Response works in a very simple manner. The SSD is set as a caching storage device in the Intel RST driver interface. Right after that the operating system starts using it as a cache and sees the SSD + HDD combination as a single storage volume.

You can use SSDs with up to 64 GB storage capacity to serve as caching devices. Moreover, when Intel Smart Response is activated for larger capacity SSDs, it is possible to assign only part of the SSD drive for the needs of this technology.

Intel Smart Response technology has another limitation. You can only assign one caching SSD drive for one HDD or one disk array. In other words, if there are two hard disk drives in the system that aren’t within the same RAID array, you will need to get two SSDs to cache them both, or to split the capacity of your SSD in two halves to create two independent caches.

The algorithm behind Intel Smart Response technology is fairly simple. During the initial request for data on the hard drive (read or write), this data is simultaneously duplicated onto the caching SSD. In case of another request for the same data, it will be delivered not from the slow HDD, but from the fast cache, which speeds up the disk sub-system. When the SSD is filled with data completely, but a new position needs to be cached, then the system will remove the least frequently accessed data from it to free up some space. And due to the fact that SSDs use power-independent flash-memory, the cache contents are saved after all reboots and power-downs. This very easy to implement principle guarantees superb efficiency of Intel Smart Response technology.

However, Intel developers made Smart Response extra intellectual, which makes the use of SSD cache even more effective. Firstly, it is important that this technology is abstracted from the file system: it caches not the actual files but clusters. This makes the use of this cache more efficient. Secondly, Intel Smart Response can identify specific data access scenarios, which do not require transferring the data into the cache. For example, system virus scanning, HD video playback or simply copying large chunks of data from one drive to another.

Intel Smart Response technology offers two cache usage strategies. You can use Intel RST to switch between them at any time:

Knowing how Intel Smart Response technology works, we should understand the importance of choosing the right SSD for the cache. A lot of inexpensive SSDs offer relatively low write speed and may even lose to contemporary HDDs in this aspect. In a Smart Response array the write speed is limited by the SSD writes even in Maximized mode, so in some cases Intel Smart Response technology may even slow down the disk subsystem altogether. The thing is that in Enhanced mode the write speed is determined by the writes of the slowest drive in the SSD-HDD combination, which can be either one. In Maximized mode the performance depends solely on the SSD.

Intel tried to resolve this issue by launching a special SSD developed directly for the use in Smart Response arrays. Please meet Intel Larson Creek or Intel SSD 311.

This 20 GB SSD is really great for the cache. Its major secret weapon is the use of 34 nm SLC NAND chips, which are more expensive than the popular MLC-flash, but boast tangibly faster writes and allow much greater number of rewrites. In other words, Intel SSD 311 has all the features that are important for the caching SSD drive.

Of course, it doesn’t break any performance records like OCZ Vertex 3 or Crucial M4 and supports only 3 Gbps SATA, but nevertheless, offers very good combination of features for its needs.

The only disappointment is the price: the 20 GB Intel SSD 311 costs around $100, i.e. about the same as the regular SSD drives with twice the capacity. This high price is the result of using expensive SLC-flash. However, it seems to be quite justified, because SLC NAND should ensure that Intel’s Larson Creek may continue functioning as cache for a much longer period of time than any drive with MLC-memory inside would.

So, what about real numbers? At first we checked out the performance of our Smart Response array in both available modes: Enhanced and Maximized, using a simple synthetic CrystalDiskMark 3.0.1 test. Smart Response array was built using Intel SSD 311 caching drive and a conventional Seagate Barracuda 7200.12 with 320 GB storage capacity. For comparison purposes we performed the same tests with Intel Smart Response technology disabled, where we only used a single Seagate Barracuda 7200.12 HDD or one OCZ Vertex 2 SSD with 120 GB storage capacity.

The results illustrated all the major features of Intel Smart Response Technology very clearly. Enabled caching raises the read speed to the SSD level, while the write speed increases only when we enable caching write-back.

At the same time it is important to understand that the results of the synthetic CrystalDiskMark test do not represent the whole situation fully. This test uses five test runs to generate the final score. Therefore, Smart Response cache gets successfully filled after the first run and after that the test is actually checking the performance of the caching SSD, which already contains all the necessary data. And the results on the graphs are the average of four test runs with SSD-caching and one “trial” run when all the data is taken from the HDD and then transferred to the cache.

Therefore, the use of synthetic benchmarks for Intel Smart Response testing is not very illustrative. The results will depend on the fact whether the benchmark executed completely random data requests or repeats the same route over and over again. So, we decided to also use some tests that measure the actual performance in real applications. In this case we compared the results obtained after the first “trial” run, when the performance of Intel Smart Response is practically the same as that in a system with a single HHD without any additional caching.

PCMark 7 test shows most definitely that Intel Smart Response plays its role very well. By adding a fast SSD cache, the system equipped with a traditional HDD really speeds up to the level of a platform with a large solid-state drive in it.

Besides the results of the disk benchmarks we also ran a few tests from the SYSmark 2007 suite. This suite gives us a better idea of the overall platform performance, which can be observed in real applications and tasks.

The results speak for themselves. Replacing a HDD with an SSD is a good way of speeding up your system. This measure improves the system responsiveness, load programs faster and opens files quicker. However, you may achieve a similar effect by enabling Intel Smart Response. At least you will be able to speed up access to frequently used programs and files, which in the end will translate into the same positive experience for the user.

At the same time it is important to understand that 20 GB of the disk cache space offered by Intel is quite enough to successfully speed up the average mainstream system. However, if you constantly work with a large number of diverse programs and apps, we would strongly recommend using a larger SSD than Intel SSD 311.

Conclusion

Although overall we didn’t have any serious issues with the new Intel Z68 chipset, our first encounter with it left us a little confused. It is totally unclear why this core logic set, which like its predecessors belongs to the Cougar Point family, appeared only now. There was nothing in it that could justify launching it almost 6 months after the release of the Sandy Bridge processors. In fact, Z68 doesn’t bring anything new to the table: it is merely a combination of P67 and H67 created as a universal “one size fits all” product, which should suite any type of user. We got the feeling that this late Intel Z68 launch is mostly a marketing move, which doesn’t actually do Intel a lot of good. It looks as if the manufacturer purposefully delayed the launch of a fully-functional chipset trying to make a profit out of its limited modifications.

However, in any case the users have finally received a Sandy bridge chipset that we can, hands down, recommend for any high-performance LGA1155 systems. This chipset has it all: you can use the processor integrated graphics core or not; you can overclock processor, memory and graphics; you can even use Intel Quick Sync without losing the discrete graphics accelerator.

At the same time Intel didn’t just unlock everything that was in Cougar Point chipsets right from the start. They have also introduced a few very interesting technologies. In our opinion, the most useful one is Intel Smart Response technology that can now be found in the Intel RST driver that allows to speed up the disk sub-system by adding an SSD cache. Our practical tests showed that by adding a small fast-caching SSD to the system, we could significantly increase its performance and improve the characteristics of the slow HDDs to the level only attainable by fast solid-state drives. In fact, for a modest price of around $100 Intel gives you the opportunity to significantly improve the quality of your disk sub-system, which is a great option for those who have limited budget but need high performance and large storage capacity.

Another useful advantage is Virtu technology developed by Lucid Company. This technology allows users of Sandy Bridge systems with discrete graphics cards to get access to one of the most interesting CPU features – Intel Quick Sync technology. This technology integrated into the graphics core inside the CPU delivers unprecedented HD video transcoding speed and now it automatically becomes an advantage of any Intel Z68 based system supporting Virtu without the need to compromise.

The only drawback we found about the new Intel Z68 chipset is higher power consumption of the systems with it. Our tests showed that the platforms built around an older P67 chipset, which doesn’t give you access to the integrated graphics core, consume considerably less power. However, keeping in mind much richer functionality of the new Intel Z68, it is hardly a serious problem.