by Doors4ever
09/30/2009 | 04:46 PM
All of us who are interested in computer technologies have definitely noticed the recent launch of the new Intel LGA1156 platform. The new CPUs turned out very interesting solutions, of course, not without drawbacks, but with a number of evident advantages. However, I am sure you know all that by now already, especially if you have already checked out our recent article called "Second Advent of Nehalem: Core i7-870 and Core i5-750 Processors in LGA1156 Platform". And, as it always happens with new systems, overclocking immediately becomes one of the hottest topics for discussion. The main CPU overclocking principles haven’t formally changed for a long time already. You must increase the base frequency trying to keep all other connected frequencies within acceptable intervals. You can also increase voltages if necessary monitoring the temperatures very carefully and it should also improve the final result. Everything is quite simple at first glance, but as a rule commencing overclockers quickly get lost and can’t find any similarity to their own system when you refer them to an Intel Pentium II overclocking guide, for instance. That is why it makes much more sense to explain everything using current examples, which is exactly what we are going to do today.
New Lynnfield processors are based on Nehalem microarchitecture that is why the basic overclocking principles as described in our article called “Intel Core i7-920 Overclocking Guide” are absolutely applicable in this case, too. However, there are a number of peculiarities connected not only with the integration of the PCI Express bus controller into the CPU as well as transition from dual-chip core logic structure to a single-chip one, but also with a different more advanced implementation of the Turbo Boost. With the help of Intel Core i5-750 and Intel Core i7-860 processors we will find out how well they can be overclocked using static and dynamic implementations of Turbo Boost technology. But before we get to that let’s talk about the features and functionality of Asus P7P55D Deluxe LGA1156 mainboard based on Intel P55 Express chipset, which we are going to use as a platform for our overclocking experiments.
We are very well familiar with boxes used for Asus mainboards on Intel chipsets. They primarily use blue color scheme, since it is the color of the Intel logo; feature a flip-open front cover, which increases the effective surface size containing detailed info about the mainboard features. This is exactly what the box of Asus P7P55D Deluxe mainboard looks like:
When we speak of the mainboard accessories bundle, we usually stick to enlisting the included accessories and provide a photo of all the items. I don’t think that many readers really take the time to check out every cable and every bracket, because they are all pretty common and have already been photographed many times before. This time we would like to offer you a large photo, because besides the standard set of accessories Asus P7P55D Deluxe also comes with a TurboV Remote control unit. It is an L-shaped panel with buttons that allow you to power on or turn off the PC, select automatic or manual power-saving mode, and most importantly, allows you to quickly switch between one of the three existing settings profiles. For example, you can switch from power-saving mode for Internet surfing to high-performance gaming mode in no time. Moreover, you can use this remote control unit to change the base frequency and even clear CMOS by pressing a special button on the back of TurboV Remote, but these particular features are very unlikely to become very popular. This remote control unit is not a wireless one, which is actually more of an advantage rather than drawback, because this way you won’t lose it. It is convenient to control some multimedia applications from the comfort of your sofa, but TurboV Remote will come in most handy when you are in front of your PC. Besides, its 1.5-meter cable will allow you to find a nice spot for it in close proximity of your system.
Besides the mainboard and TurboV Remote control unit, Asus P7P55D Deluxe also comes with the following accessories:
I would like to draw your attention to the additional bracket for the system back panel with an eSATA and two USB ports: we haven’t yet come across a ports combination like that before. The thing is that Asus P7P55D Deluxe also has an IEEE1394 (FireWire) port on the back panel, and no eSATA, that is why they included a bracket like that among the bundled accessories.
As a result, we can conclude that Asus P7P55D Deluxe mainboard has a nice set of useful accessories and a TurboV Remote unit. We can hardly consider this remote control unit a primary essential, but it is definitely a handy device: there will definitely be users working with it on a regular basis. Besides, it is one of the peculiar features that distinguish a Deluxe mainboard from the rest of the models in the lineup.
LGA1156 mainboards based on Intel P55 Express look a little unusual due to the fact that they use a single-chip core logic set without the North Bridge chip, which functions have been moved into the CPU. However, some of our upcoming reviews will also talk about more traditional design implementations, too. A lot of mainboards have the only chips of the Intel P55 Express core logic set – Platform Controller Hub – where the conventional North Bridge used to be. In this case it is usually equipped with too sophisticated cooling system involving heatpipes, just like before. The South Bridge spot is taken over by additional onboard controllers providing support for PATA and SATA storage devices and these chips are usually covered with a separate heatsink, just like the South Bridge chip. As a result, new mainboards look almost like the solutions based on previous chipsets, but I have to say that it is not true for Asus P7P55D Deluxe. You’d better get used to it, this is what a typical mainboard for Lynnfield processors should actually look like:
Since we came to speak about cooling solutions, let’s give due credit to Asus P7P55D Deluxe mainboard developers who have tackled this aspect extremely attentively. The only Intel P55 Express microchip is located where the chipset South Bridge used to be. It is cooled with a large but low-profile heatsink, which is more than enough even without any heatpipes. However, the heatsinks installed on top of the 16-phase processor voltage regulator circuitry use reliable screw-on retention not just for decorative purposes and are connected via a heatpipes for a reason. They bear the primary thermal load, besides the processor heatsink, of course, and this load increases substantially during overclocking. Therefore, the heat generated by the processor voltage regulator circuitry at the bottom of the PCB is dissipated through thermal interface to a couple of additional metal backplates.
In order to provide a better-balanced, better-grounded and more objective estimate of the Asus P7P55D Deluxe mainboard efficiency, we have also tested the new processors on Gigabyte GA-P55-UD3 for the sake of comparison (the review of this mainboard will follow shortly). Although this is one of the junior mainboard models in the lineup, it is also equipped with a couple of pretty large heatsinks on the components of the CPU voltage regulator circuitry. However, they are fastened using common plastic push-pins and have no additional backplates at the bottom of the mainboard PCB. We noticed that its heatsinks heated up significantly during overclocking. Moreover, the PCB textolite under the heatsinks changed its color and became noticeably darker because of extreme overheating.
We didn’t detect any serious heatsink heating like that on Asus P7P55D Deluxe. It could be the case due to more phases in the processor voltage regulator circuitry, but it could also be due to more reliable screw-on retention and heat-dissipating backplates at the bottom of the PCB. All in all, we would like to give Asus developers a well deserved check-mark for very efficient and not too excessive cooling system they implemented on their P7P55D Deluxe.
As you may guess even from the model name, the functionality of Asus P7P55D Deluxe is more advanced than that of a mainstream mainboard on Intel P55 Express/ to begin with, contemporary LGA1156 mainboards are equipped with one PCI Express 2.0 x16 slot or two slots, which will switch to PCI Express 2.0 x8 mode when two graphics cards are used. They are supported by the PCI Express controller, which is now integrated into the CPU. There is also the third graphics card slot on Asus P7P55D Deluxe, but it uses the remaining four lanes of the PCI Express chipset and a graphics card installed into it will not work any faster than PCI Express 2.0 x4.
The developers had to use an additional JMicron JMB363 controller in order to implement support for PATA storage devices, because Intel chipsets haven’t featured their support for a long time already. Note that there is one “native” SATA port with a black connector and another SATA port split into two more with the help of JMicron JMB322 controller (dark-blue and gray connectors). The devices connected to these two ports do not require any drivers and can be easily tied up into RAID 0 or even 1 arrays without any special knowledge. In Asus terms, this technology is called Drive Xpert. As a result, Asus P7P55D Deluxe mainboard allows connecting up to nine SATA devices: six ports provided by Intel P55 Express and three more – by additional controllers.
Asus P7P55D Deluxe mainboard design looks good not only in general terms, but also has some very nice minor advantages. Power On and Reset buttons are lit up during work, and so is the small MemOK! Button that should help on first system boot-up if the computer cannot start because of incorrectly configured memory settings. There are special switches right above the memory DIMMs that allow sending slightly higher voltage to the CPU, the integrated memory controller and the DDR3 memory modules. There are green LEDs next to the switches, which will turn orange as soon as the voltage increases. The memory DIMM slots have thumb-locks only on one side, farthest from the graphics card, so the installed graphics card will not complicate taking out or putting in the memory modules. We really felt the difference from the wide grips on the graphics card slot lock. Scythe Zipang 2 processor cooler that we used during this test session is very wide and comes very close to the graphics card installed into the first slot. We would inevitably have problems with any other mainboard, but not with Asus P7P55D Deluxe: by gently pressing the tip of the screwdriver against the grip we could easily free the graphics card and take it out.
The connectors on the mainboard back panel are also quite diverse:
Asus P7P55D Deluxe mainboard has not just a good layout, but a really excellent one. I managed to find only one, currently very insignificant drawback – inconvenient location of the COM port, which is way too far up, to the right of the memory modules. The components layout from the mainboard manual will give you a bird’s eye view of the components location on the PCB:
The table summing up all technical specifications of Asus P7P55D Deluxe mainboard will wind up our discussion of the board’s layout and functionality:
First look at Asus P7P55D Deluxe mainboard leaves an extremely favorable impression. The board features an excellent well thought-through design, superior functionality, a bunch of nice trifles that make the entire experience extremely pleasing. We hope that the impressions from the BIOS Setup won’t spoil our high initial score that currently looks as “9 out of 10”. We decided to take one point not for some drawbacks, which are hardly even there, but just in case. It is the first Intel P55 Express mainboard that we review in detail and maybe tomorrow we manage to get our hands on something even more functional and easy to work with but also less expensive at the same time, who knows? However, at this point we can’t even wish anything better than Asus P7P55D Deluxe. So, let’s move on.
We are very well familiar with the exterior looks and structure of the Asus mainboards BIOS based on AMI microcode that has undergone significant modifications:
Without going deep into details on all available functions, we would like to draw your attention to the most important BIOS Setup sections that are essential for proper system configuration. In this respect the primary section would be “Ai Tweaker”. Although it has some many parameters that they simply can’t fit into the first screen, I consider this layout to be much more informative and easy to work with than a nested chain of numerous sub-topics. When we configure our system settings we go through all the parameters from top to bottom, changing those settings that we need, and it is much more convenient than jumping from one sub-section to another. Only the memory timings are singled out onto a separate page, but it seems to be a smart decision, considering how numerous they are.
You can set “Ai Overclock Tuner” parameter to Manual, and in this case you will get full access to a bunch of other settings. You8 can select “D.O.C.P.” — DRAM OverClocking Profiles. In this case the mainboard will select the most optimal settings for the desired operational mode on its own. For example, if you have an Intel Core i7-860 processor and you wish to overclock the memory to 1800 MHz, the board will increase the base frequency from the nominal 1333 MHz to 150 MHz in order to get the desired memory frequency. At the same time it will lower the processor clock frequency multiplier, so that its resulting frequency could be as close to the default 2.8 GHz as possible.
If the memory modules in your system support X.M.P. (eXtreme Memory Profile) technology, the board will act in a similar manner. In order to switch our DDR3 Corsair Dominator GT CM3X2G2000C8GT memory modules to 2000 MHz frequency, we had to increase the base clock to 167 MHz and at the same time reduce the multiplier to 17x.
The examples above are true for the Intel Core i7-860 processor. With an Intel Core i5-750 CPU the board acts differently. It is not only because this CPU features lower nominal clock frequency and will require a different clock multiplier. As you know, Intel Core i5-750 turned out more feature limited than we have expected it to be. It doesn’t support 12x memory multiplier that we used before and can only go as far as 10x. In this case, we need to increase the base clock to 200 MHz and lower the processor frequency multiplier to 13x in order to hit 2000 MHz memory frequency.
Why do we pay so much attention to “D.O.C.P.” and “X.M.P.” settings for “Ai Overclock Tuner” parameter? These features are not really new, Asus mainboard have supported them for quite some time now. The thing is that when we changed the processor clock frequency multiplier before, it was locked automatically at a selected value, the multiplier would no longer drop in idle mode when the processor utilization was low. Of course, it had its negative effect on system power consumption and all related aspects such as heat dissipation and noise levels that is why these overclocking techniques weren’t taken seriously. Today, it has become an absolutely real way of increasing the system performance, because now the processor clock multiplier will still get lower in idle mode even if you change it manually. This new feature expands our abilities for system fine tuning. For example, you can increase the base frequency in such a way that your memory modules will work at their most effective frequency. At the same time, you can lower the processor clock frequency multiplier in order to avoid increasing its Vcore, and end up with a pretty fast and energy-efficient system.
Another new feature is the “OC Tuner Utility” built into the mainboard BIOS. If you activate it, the mainboard will reboot multiple times, each time increasing the base clock a little more. As soon as the board detects first errors at the POST stage, it will step back a little to avoid these errors in the future.
Of course, it is still a pretty primitive overclocking technique, but it doesn’t require user’s participation and is performed automatically. We don’t have many ways of influencing the work of “OC Tuner Utility”. All we can do is change the “OC Tuner Limit Value” parameter from “Good Performance” to “Better Performance”. However, it is nevertheless better than the older “CPU Level Up” overclocking option, when we used the preset CPU profiles, just like we did for the memory. This time the system doesn’t try to squeeze our CPU into a certain preexisting pattern, but tries to adjust for the potential of the existing processor unit.
As we have already said, the only sub-section in “Ai Tweaker” section is “DRAM Timing Control” that allows you to monitor the current timings settings and adjust them if necessary.
The next group of parameters in “Ai Tweaker” section controls the voltages. It is very convenient that there is the current parameter value displayed next to it.
In our Asus Rampage II Gene mainboard review we have already come across the option to set a relative voltage value (Offset) instead of absolute fixed voltage. However, it is for the first time that we see a feature like that on a regular mainboard that doesn’t belong to the elite Republic of Gamers series. I t is difficult to overestimate the importance of this feature. The formal advantage of all Asus mainboards for Intel processors that has long turned into a drawback, namely when mainboards increase the processor core voltage on their own during overclocking, didn’t go anywhere. However, from now on this peculiarity of Asus mainboards is no longer an issue for those users who prefer energy-efficient overclocking. Now, when you increase the CPU Vcore, all Intel power-saving technologies stay up and running dropping the core voltage in idle mode and increasing it under heavy load. Moreover, the adjustment increment for the CPU core voltage is extremely small, only 0.00625 V. So, you can increase the processor Vcore by this tiny number, so that it stays practically nominal and this way avoid automatic core voltage increase during overclocking. By the way, you can lower the voltage instead of increasing it, if you are after more energy-efficient and quiet system operation rather than maximum performance at any rate.
“Ai Tweaker” section is really good from all standpoints, but it only partially represents the processor related options. You have to check the “CPU Configuration” sub-section of the “Advanced” section for a full access to all processor technologies. I personally would prefer to see this sub-section moved entirely to the “Ai Tweaker” section.
Next we would like to take a look at “Hardware Monitor” sub-section of the “Power” section. Before, we would be pretty unhappy about scarce number of monitored parameters, but let’s not forget that now we know all the important voltages from the “Ai Tweaker” section. They are all listed directly next to each of the settings responsible for voltage adjustment. So, we should only check “Hardware Monitor” sub-section if we wish to enable automatic fan rotation speed control and set the appropriate mode. By the way, even when we overclock processors with pretty substantial voltage increase, Q-Fan system coped perfectly fine with the CPU cooling even in Standard mode.
The last section of the Asus P7P55D Deluxe BIOS that we would like to mention in our today’s review is called “Tools”. In fact, we are familiar with its functionality. The only new parameter here is “ID LED”. When we talked about the mainboard design, we mentioned a number of different LEDs, so if their glowing annoys you, you can simply disable them here.
The functionality of the “O.C. Profile” sub-section has also become quite more advanced lately. Now you can save several full BIOS settings profiles. You can provide each of them with a unique name, easily load the selected profile. You can even save the profiles not only in the internal memory but also on the external storage media.
Very convenient “EZ Flash 2” utility will help you save the current BIOS version and update it with the latest available one.
Summing up our experience with Asus P7P55D Deluxe mainboard, we can conclude that the BIOS structure and functionality hasn’t really changed that much compared with the solutions based on other chipsets. In fact, this isn’t surprising at all, because the BIOS of contemporary mainboards has been modified and polished off for years. At the same time, we can’t help pointing out a number of new features, like automatic processor overclocking tool or the option allowing to disable LED lighting. However, we were mostly impressed with the mainboard’s new ability to change the processor clock frequency multiplier and core voltage without disrupting the work of power-saving technologies. They provide enormous flexibility for optimal system configuration.
Moreover, we can’t disregard the fact that our tests were performed on the first officially available BIOS version 0504. Of course, we provided the screenshots from this particular BIOS version above and later on we are going to talk about the results of our overclocking experiments with it. However, now that the mainboards are available in retail and the users began to share their feedback about them, the developers focused their efforts on fixing the issues and expanding the BIOS functionality. The new BIOS versions support low-voltage DDR3 SDRAM, have even better optimized algorithms in the built-in “OC Tuner Utility. There also appeared “Turbo Profiles”, which allow to simultaneously overclock the CPU and the memory. It is also quite possible that by the time the review posts on our site even newer BIOS version will be available, with new features and functions, so do not forget to update the BIOS in your board if you want to get access to them.
Of course, the BIOS of Asus mainboards and P7P55D Deluxe mainboard in particular are not ideal. There are a few insignificant issues, which should make it even easier to work with this mainboard, once eliminated. We have already mentioned a few ones today, for example, it would be nice to move the “CPU Configuration” sub-section into “Ai Tweaker” section. We pointed out some of them in our previous Asus reviews, for example, it is much more convenient to monitor current memory timings if they are listed in a column, each next to the corresponding parameter, instead of a single-line presentation used now. However, we don’t even want to bring up these small trifles again. We can’t wait to get to the board’s actual abilities for CPU overclocking. However, first we need to get ready and learn how we should actually overclock the new Core i7 and Core i5 CPUs from Lynnfield family.
We anticipated the arrival of the new processors with mixed feelings. On the one hand, we were extremely curious to see what they are capable of in tests, find out how Lynnfield different in functionality from the junior Bloomfield and top Core 2 Quad. We were already excited about the new Turbo Boost implementation, because Lynnfield are the first universal processors that combine the advantages of multi-core and single-core CPUs. They work as multi-core processors in contemporary multi-threaded applications: run at not very high frequencies but execute multiple computational threads at the same time. They lower the number of active cores putting uninvolved ones into energy-efficient modes when there is no need for multi-threading, thus increasing the working frequency of the remaining active cores quite substantially. On the other hand, we had some logical concerns. How do we overclock processors, which frequency multiplier can increase by 4-5 points above the nominal setting? Keeping in mind that in idle mode the multiplier lowers to 9x and under heavy load increases to 24-27x, it seems barely possible to determine operational stability in all intermediate modes.
Luckily, it turned out that new processors can be overclocked just as easily as any other CPUs, and maybe even easier. Unlike LGA1366 platform, now we don’t have to monitor the frequency of the North Bridge part integrated into the CPU, which is called Uncore according to Intel or IMC (Integrated Memory Controller) according to Asus. Secondly, overclocking no longer requires serious IMC voltage increase. Before, we were offered to increase this voltage to 1.5-1.6 V in order to ensure that our memory could work at high frequencies. In reality we could often do by raising this voltage to 1.35-1.45 V, which was still pretty high. Now we don’t have to raise the IMC voltage at all for the memory to work at high frequencies, and for stability at 200 MHz base clock it should only push it to 1.2 V.
Just like with Bloomfield processors on LGA1366 mainboards, there are two ways of overclocking Lynnfield. The first way is the static implementation of Intel Turbo Boost technology, or even its complete disabling. In both cases we deal with a system where the processor clock frequency multiplier remains locked at a certain fixed value under heavy load. It is either at its nominal value if we disable Turbo Boost completely, or is a little above the nominal independent of the processor load. The second overclocking approach is dynamic implementation of Intel Turbo Boost, when the multiplier depends on the current processor utilization. The fewer cores are busy, the higher rises the multiplier, and the other way around.
It is clear that both these approaches have the right to exist. The static implementation is good for those users who work with well-paralleled applications – programs that can perform multi-threaded calculations that speed up the process a lot. Among them are applications for distributed computing, creation and processing of multimedia content: multi-threaded tools for work with models, sound, images and video. Dynamic overclocking approach will suit more for a home systems used for everyday work and entertainment. In this case we benefit most from the use of single- and dual-thread applications, which are still the majority these days and at the same time ensure very high performance level in multi-threaded tasks.
However, it all looks so simple only in theory. In reality we didn’t manage to find a universal LGA1366 mainboard that could have both versions of Intel Turbo Boost technology implemented equally successfully. We have most often seen mainboards only with static implementation, and more rarely – only with dynamic one. If we had a mainboard that allowed us to choose either of these, then again only one of the implementations was a preferred one. As for LGA1156 mainboards, looks like they simply do not have a problem like that at all. By default, all mainboards are set for static implementation of Intel Turbo Boost technology, and if you wish to enable the dynamic one you have to enable extended C3-C7 modes in the BIOS Setup section with processor settings.
Before any overclocking experiments we have to do some preparatory work. First of all, it would be really good if you could remove the “Auto” settings for all significant parameters in the mainboard BIOS. No one knows at what stage the mainboard suddenly decides to increase the voltages, change the memory frequency or timings, which may have a negative effect on system stability. That is why we lower the memory frequency right from the start - it will increase as the base clock frequency increases – and we will find out the end-value a little later when we make our decision regarding the CPU overclocking. As for the major timings, it would be best to set them to guaranteed operational values, for example, 8-8-8-22 or 9-9-9-24. All voltages should be at their nominal values, except the IMC voltage, which can be increased to 1.2-1.25 V right away (we will lower it later on if we don’t need this increase), and the memory voltage, which should be raised no higher than to 1.65 V. as for the processor Vcore, we can also leave it at its nominal, of you prefer to have a fast but at the same time energy-efficient system in the end. Do not forget to enable ”Load-Line Calibration” technology preventing the processor Vcore from dropping under heavy load. On the other hand, you can increase the voltage right from the beginning, but the increase really depends on the efficiency of your processor cooling.
As the first step towards overclocking, you can double-check that the mainboard can remain stable at high base clock rates. In fact, we do not expect any problems here, all LGA1156 mainboards we have worked with so far performed stably up until 210 MHz base clock frequency. However, it is best to make sure this is the case instead of guessing later on why the CPU doesn’t overclock any higher simply, to find out that the reason lies with the mainboard. Therefore, let’s lower the processor clock frequency multiplier to 12-14x, so that its frequency could be close to the nominal during maximum overclocking. After that we raise the base frequency to 200-210 MHz. at this point we have to check if the memory frequency really lies within the acceptable range for the installed memory modules. And finally we use any test program to run a stability test. If you have hard time deciding on a program like that, we can recommend Prime95. At this point you can lower the IMC voltage, if possible. If lower voltage setting will be acceptable at these high base clock speeds, then it should definitely work at lower base clocks as well.
Now let’s check out overclocking during static implementation of Intel Turbo Boost or with this technology totally disabled. If you overclock without increasing the processor core voltage, then you can expect the resulting frequency to be around 3.5-3.7 GHz. This is only an approximate number that we obtained during our overclocking experiments performed with only two processor samples. We will be able to offer you more exact data a little later when we collect a bit more statistical data. But anyway, you are the only ones who can find out the overclocking maximum for your particular processor sample. At first you have to make sure that the processor cooler that you work with is efficient enough to handle overclocking. Intel Linpack testing suite generated extremely high processor load. You can use LinX shell for your convenience. After that using Prime95 as a stability test we increase the base clock if the system passes the tests, and lower it if errors emerge. After a few attempts you will fina the maximum stable frequency for your particular CPU.
For higher results, you need to increase the processor Vcore and this is where temperature becomes of primary importance. The higher you set the core voltage, the higher overclocking results you can actually achieve. But excessive voltage increase may push the CPU temperature to extreme heights, which will only limit further overclocking. Our goal is to fine the optimal combination of CPU core voltage and temperature.
The eternal question is: what is the maximum acceptable CPU temperature? Strange as it might seem, but it is your call. Some try to keep the CPU temp below 60 °C, some believe that 95 °C is not the limit yet. I can say one thing with all certainty: it is best if the CPU core temperature doesn’t hit 90 °C. Moreover, overclocking with the CPU temperature exceeding 90 °C is meaningless and unreasonable. For example, when the CPU core temperature reaches 93-94 °C on Asus mainboards, protective technologies kick in and start lowering the frequency. Summer has come and the room temperature increased, or you turned the heater on, or maybe the processor heatsink sucked in too much dust and doesn’t cool well anymore – any, even the slightest change in the working conditions may lead to system instability and errors. Why do we overclock processors in the first place? To boast a record-breaking screenshot or to get higher performance in any conditions and under any type of workload?
You can use i7Turbo utility to monitor the CPU temperature. It will show if the clock frequency multiplier is dropping under maximum load. It doesn’t make sense to overclock the CPU if it can’t remain stable under maximum operational load and starts losing frequency. Therefore, 90 °C is the maximum temperature for processor cores, and would really be best to stay as far below it as possible. Therefore, during overclocking we can look for maximum CPU Vcore with which the temperature will remain within acceptable range, rather than maximum clock frequency. We will get the maximum frequency as a result of our core voltage increase.
It doesn’t matter if you tried to find the maximum setting at which the CPU remained stable with or without its core voltage increase. If you did, then leave the determined value, if not, then set an approximate base clock frequency that will produce resulting CPU speed between 3.5 and 3.7 GHz. After that, increase the voltage. At first let’s try and push it up to 1.27-1.3 V. Launch LinX and see how far away the core temperature is from the dangerous 90 °C or any other threshold you picked. You can monitor the core temperature using any tool: RealTemp, CoreTemp, HWMonitor, SpeedFan, Everest. If the temperature is too high, you should lower the voltage, if it is pretty low, you can increase the voltage a little more. In either case, it is important to remember that later on when we increase the frequency the temperature will continue to grow, even though not so greatly as during voltage increase.
Have you determined the approximate voltage setting when the core temperature remains within acceptable range? Now let’s repeat the familiar course of actions: increase the base clock if the system passed the stability test, or lower it if it didn’t. This way we will find the maximum processor frequency at the given voltage setting, which in its turn prevents us from getting beyond the acceptable temperature threshold. After that you can normally lower the processor Vcore a little without losing stability, which will also lower the maximum temperature. Now all we have to do is find the optimal memory frequency and timings for the determined base clock. Congratulations! You have just overclocked your system. It will be able to stably deliver higher performance for years in this safe mode (from voltage and temperature prospective).
Summing up, here is the schematic succession of actions we have just performed:
At first it seems that with dynamic implementation of Intel Turbo Boost technology it is much harder to find optimal overclocking parameters than with static implementation. In reality, everything turned out pretty simple. The difference is that besides the dangerous CPU temperature during maximum processor utilization, we should also take into account the frequency limitation when only one processor core is loaded with work and the clock frequency is at its highest. We have just found the maximum CPU overclocking when its multiplier is within 20-24x depending on the CPU model. Obviously, you can’t just enable dynamic overclocking, when the multiplier can increase to 24-27x. So, we must lower the base frequency beforehand. With the maximum processor clock frequency multiplier you can shoot for approximately 4.1-4.3 GHz. You can use the voltage setting that we have just got before. Since under maximum load the CPU clock frequency will be lower, we may even be able to increase it a little higher. If you started experimenting with the dynamic implementation right from the start, then you should first do the same thing as with the static Turbo Boost: determine the maximum voltage when the core temperature under maximum load remains within acceptable range.
After that we perform the familiar course of actions: test system stability. The only difference is that now the tests are performed not during maximum CPU utilization, but when only one processor core is loaded with one or two computational threads, so that the processor clock frequency multiplier could remain at its maximum. If the system passes the test, we increase the base frequency; if it doesn’t, we lower it or increase the processor core voltage. Just keep in mind that maximum power consumption and heat dissipation can only be reached when all four CPU cores are loaded to their utmost. So, once you increased the voltage, make sure that the temperature is still within acceptable range.
In brief, here is what we do step-by-step:
We performed all our experiments on the following test platform:
We used Microsoft Windows 7 Ultimate (Microsoft Windows, Version 6.1, Build 7600) operating system, Intel Chipset Software Installation Utility version 9.1.1.1019, ATI Catalyst 9.8 graphics card driver.
Algorithms, flow-charts, all this sounds very contemporary and very appealing, but sometimes you can’t see the wood for the trees – the picture doesn’t shape up from the succession of seemingly logical actions. Therefore, we decided to take an even closer look at Lynnfield CPU overclocking than we would normally do. Some specific examples could appear more illustrative than theoretical discussions and will depict the overclocking principles more clearly.
So, we have an Intel Core i7-860 processor. Its nominal frequency is 2.8 GHz, which means that it works at the default base frequency of 133 MHz with the multiplier of 21x. In reality, we haven’t seen this CPU ever work at its nominal speed at all. The static Intel Turbo Boost implementation is enabled by default and under any load the CPU clock multiplier increases to 22x, which results into 2.93 GHz frequency. If we enable dynamic implementation, we will see the same exact multiplier when three or all four cores are loaded with work. When only two cores are utilized, the CPU runs at 3.33 GHz with 25x multiplier, and with only one core working the multiplier hits 26x and the CPU works at 3.46 GHz.
We have determined in advance that Asus P7P55D Deluxe mainboard works at 210 MHz base clock frequency, which requires IMC voltage to be increased to 1.2 V. the memory voltage was set to 1.65 V, the frequency remained the same, we used the same 10x the memory frequency multiplier, because the nominal frequency of our Corsair Dominator GT CM3X2G2000C8GT modules was 2000 MHz and in reality this memory can do even better than that. We set the timings at 8-8-8-22-1T. All other voltages were at their default settings and the CPU Vcore was increased by 0.13125 V with enabled “Load-Line Calibration” protection against voltage drop under heavy load. There is a simple explanation of why we resorted to such an “unrounded” setting: the nominal Vcore for our CPU sample is 1.16875 V, which will provide us with a pretty “rounded” 1.3 V final Vcore.
At first let’s find out what our mainboard and processor are capable off with static Turbo Boost, when the CPU clock frequency multiplier increased only to 22x. We started with 175 MHz base clock, which produced 3.85 GHz CPU frequency. It is higher than 3.5-3.7 GHz recommended by our own methodology. However, we mentioned in the beginning of this article that we also tested the new processors in Gigabyte GA-P55-UD3 mainboard, so, we have already been through everything and know very well what our CPU is capable off.
When we launched LinX utility, its eight computational threads increased the CPU core temperature to 90 °C in just three test cycles. It was way too high. We stopped our tests and adjusted the processor core voltage increasing it by only 0.125 V instead of 0.13125 V. After that we launched the test one more time. Once again we hit 90 °C, but this time after ten test runs. However, it is still too high. Now we will only add 0.11875 V to the nominal Vcore setting but at the same time will increase the base frequency to 177 MHz. System passes the test but the temperature again hits 90 °C. We lower the voltage increase to 0.1125 V and this time the test finishes at 87 °C. Not bad, but could we increase the base frequency to 179 MHz at the same Vcore setting? No, we couldn’t, because the utility started to report errors, so we rolled back to 177 MHz base clock. Maybe we would be able to lower the core voltage even more then? Nope, we get errors again. So, this is how we found the maximum possible processor core voltage setting and maximum frequency at this voltage.
Finally, let’s optimize all other system parameters. We increase the memory frequency, launch LinX utility for a final test run and then a one-hour-run of Prime95 in Blend mode. This way, by adding 0.1125 V to the nominal CPU core voltage we overclocked it to 3.9 GHz. The memory also did very well and agreed to work fine at 2124 MHz with 8-8-8-22-1T timings. Asus P7P55D Deluxe mainboard reports slightly higher base clock that is why the processor and memory frequencies are even higher.
I believe it is a very decent result. We gained 1.1 GHz of frequency. At the same time, we retained all power-saving technologies, so the processor clock frequency multiplier and core voltage will lower in idle mode.
Now let’s try to overclock our processor using all advantages of dynamic Turbo Boost. Obviously, we can’t just switch to dynamics: at 177 MHz base clock with 26x multiplier the processor frequency will increase to 4.6 GHz and in our testing conditions the system will never be stable with these settings. So, we lower the base frequency to 161 MHz, but increase the voltage back to 1.3 V by adding 0.13125 V to the nominal. Our tests prove that under maximum load created by LinX utility the CPU core temperature remains within acceptable range. That is why at this point we get to tests with only one-two computational threads, when the processor clock multiplier increases to its maximum value of 26x.
All preliminary tests have been successfully passed, so we increase the base frequency to 165 MHz, but we see errors. We add 0.14 V and then 0.15 V to the CPU Vcore, but the errors are still there that is why we lower the base clock to 163 MHz. Unfortunately, we couldn’t achieve stability at this frequency, too, so we go back to 161 MHz. after a number of tests we find out that we need to increase the processor Vcore by 0.1375 V to ensure stable operation of our processor with 26x multiplier. Another run of LinX under maximum load – the CPU core temperature barely gets past 80 °C, so in this respect the voltage setting will work just fine. Now let’s increase the memory frequency, lower the timings and launch a one-hour run of Prime95 with maximum CPU utilization and use of eight computational threads. Another successful pass with 77 °C maximum core temperature. Now we repeat the test with only one computational thread: no errors and the core temp is only 60 °C.
As a result, the CPU will work with 22x multiplier at 3.55 GHz frequency under maximum load.
When only one core is loaded with work, the processor frequency will increase to maximum 4.2 GHz.
In idle mode both, multiplier and core voltage go down, due to power-saving technologies.
At this point I would like to answer several, possible questions that you might have.
Is it enough to run tests with only two utilities to claim that the system is stable?
LinX program warms up the CPU perfectly well, and Prime95 in Blend mode tests not only the CPU but also the memory. However, they cannot give you a 100% stability guarantee. Our experience suggests that successful pass in this two applications will allow the system to do well in any other tests. Besides, these are only preliminary overclocking results. Very soon we will replace our processor cooler and the results may be different (if not the overclocking results, then at least the temperature readings). We still have a lot of mainboard reviews lined up for tests in various applications. Later on we will be able to correct the data accordingly.
Isn’t 87 °C too high for an overclocked processor?
It is relatively high. However, you should keep in mind that it was registered during tests with special LinX utility that creates extremely high processor load. During work in regular applications, we can hardly get close to this value.
If LinX utility creates unrealistically high workload, then why should be use its temperature readings as a reference during our overclocking experiments?
True, to determine the maximum voltage setting you can use any other “heavy” applications from those that you work with on a regular basis or from time to time. In this case the maximum temperature will most likely be lower than with LinX and you will be able to push the CPU even further. However, this approach may be suitable only for you, but not for me. I have no idea what applications you intend to run on your system that is why I prefer to offer you a slight reserve loading the system to its maximum. Following the above described methodology, you will most likely get a well overclocked system that will work stably under any load, but these are just recommendations. You have the right to act as you see fit taking the responsibility for the consequences of your actions.
We have also overclocked an Intel Core i5-750 processor with the nominal clock frequency of 2.66 GHz and 1.225V Vcore. I am not going to go deep into details about how we actually overclocked it, because it was performed following the same exact methodology as Intel Core i7-860 processor. With static Turbo Boost when we increased Vcore by 0.1125 V we could overclock this CPU to 4.0 GHz.
During the dynamic implementation the CPU could work at 3.73 GHz and 0.1375 V Vcore under maximum load. Intel Core i5-750 doesn’t support 12x memory multiplier - it can only work with 10x maximum, therefore, we couldn’t increase the memory frequency, but we could lower the timings.
In case of single-threaded load the CPU frequency will increase to 4.26 GHz.
Now that I was summing up the obtained results, I noticed that both processors required absolutely identical voltage increase for successful overclocking. We increased the nominal core voltage by 0.1125 V with static Turbo Boost, and by 0.1375 V with dynamic Turbo Boost. It is a very interesting coincidence. We are going to check if it remains the case with other mainboards. And now let’s sum up everything we know about the overclocking potential of Asus P7P55D Deluxe mainboard:
Asus P7P55D Deluxe mainboard is the first LGA1156 solution that we have tested in detail in our lab. It doesn’t yet have any competitor we could compare it against that is why let’s see what performance gain we can get from overclocking Intel Core i7-860 processor. In nominal mode the board set all parameters on its own, we only enabled dynamic Turbo Boost implementation manually. At first let’s compare the performance numbers with those obtained during overclocking when the clock frequency multiplier also changes dynamically.
The numbers are quite impressive. If we disregard those cases when the performance is limited by the graphics card, the gain makes about 20-30%. And now let’s check the performance gain compared with the overclocking results obtained with static Turbo Boost implementation in place.
This time, the performance gain from overclocking reaches 40% in some cases. And finally, let’s compare the performance numbers obtained during overclocking using static and dynamic Turbo Boost implementation:
What we see is really puzzling. If graphics card becomes a limiting factor, the performance numbers are about the same, and in almost all other cases we see dynamics falling about 5-10% behind statics. And although on average the lag makes only about 4.5%, it doesn’t really work as a consolation, as we expected the dynamic mode to win! Formally there is nothing surprising about it. Our benchmark set is put together for the purposes of comparing mainboards, and not CPUs. Moreover, we have specifically selected mostly multi-threaded applications that can use the potential of multi-core CPUs. So, what equal conditions are we talking about if the CPU works at 3.55 GHz with dynamic implementation and at 3.9 GHz with static one? Of course, statics is faster. The only single-threaded application that we have added to our benchmarking suite in order to estimate the possible gain from dynamic Turbo Boost technology implementation is SuperPI. This is where we naturally see a logical advantage of the dynamic approach.
So, we started to look frantically for single-threaded applications that could convincingly demonstrate the advantages of dynamics over statics. To my great surprise, I didn’t find anything like that. Of course, we can run Cinebench or Fritz with only one thread and get the desired result, but it has nothing to do with the real state of things. I doubt that anyone will give up multi-threading and sacrifice higher performance only in order to raise the CPU frequency. We only care about maximum speed and it doesn’t matter how we achieve it: by increasing the frequency or the number of simultaneously executed computational threads. If the second approach is much faster, no one will resort to the first one. This is where a paradoxical at first glance conclusion comes to mind: static Turbo Boost implementation performs much higher during overclocking than dynamic one.
In fact there is nothing surprising here, dynamic Turbo Boost implementation shows all its advantages only when the CPU works in its nominal mode but not during overclocking. So what changes when we switch from statics to dynamics in nominal processor mode? Nothing, except the fact that in some cases we allow the CPU to increase its own clock speed. We have the same base clock of 133 MHz, which means that all connected busses, such as the memory bus, for instance, have the same frequencies. Of course, in this case dynamic implementation is preferable, as we can see from the results of our comparison. We see a convincing and logical advantage of the dynamic implementation when the CPU works in its nominal mode.
And when we switch from statics to dynamics during overclocking, everything changes. We had to lower the base clock and as a result all frequencies tied up to it lowered, too: by lowering the base frequency from 177 to 161 MHz we automatically reduced the memory clock from 2124 to 1932 MHz. Of course, more aggressive memory timings partially make up for this lowering, but nothing will be able to cover up the drop in the CPU frequency under heavy load. Yes, sometimes the CPU frequency will increase to 4.2 GHz, which is higher than 3.9 GHz with static Turbo Boost, but at the same time it will often be only 3.55 GHz instead of the same 3.9 GHz. Keeping in mind that there are barely any contemporary single-threaded calculations these days, because the CPU anyway has to pay some attention to OS and other applications’ requests, it appears that we only get maximum performance during overclocking with static Turbo Boost implementation. Of course, we could occasionally run Pi calculations just for maintaining our self-esteem with dynamic Turbo Boost used during overclocking, but it is hardly practical. We could also dig out some old single-threaded games, where we will also see a performance increase, but the performance of contemporary processors and graphics cards is more than enough for old games even without the Turbo Boost. All in all, dynamic Turbo Boost implementation turns out to be less useful during CPU overclocking than static one.
We measured the power consumption using Extech Power Analyzer 380803 device. This device was connected before the system PSU, i.e. it measured the power consumption of the entire system without the monitor, including the power losses that occur in the PSU itself. When we took the power readings in idle mode, the system was completely idle: there were even no requests sent to the hard drive at that time. We used LinX program to load the Core i7-860 CPU. For more illustrative picture we created a graph showing the power consumption growth depending on the increase in CPU utilization as the number of active computational threads in LinX changed.
System power consumption with Intel Core i7-860 processor working in its nominal mode is practically the same in case of static and dynamic Turbo Boost. The only thing I could point out here is slightly higher power consumption in idle mode with statics in place. During CPU overclocking, this difference becomes even more obvious.
Independent of the Turbo Boost implementation, the system power consumption in both cases is very similar. However, in idle mode the power consumption of an overclocked system with dynamic Turbo Boost is almost the same as with the CPU in the nominal mode, while with static Turbo Boost it is way higher. The thing is that when we enable C3-C7 states for dynamic implementation, we allow the CPU to switch to deep power-saving modes by disabling more units. That is why the difference between dynamic and static modes in idle mode is quite logical. However, I didn’t expect it to be so significant. Since more than 90% of the time the system is not really loaded heavily, those users who decide to go with static overclocking should keep in mind that in idle mode their system will be more energy-hungry.
The comparison between power consumption of LGA1156 and LGA1366 platforms is even more impressive. Besides two versions of Intel Core i7-860 CPU overclocking, we also added the results obtained during Intel Core i7-920 overclocking to 3.8 GHz using Gigabyte GA-EX58-UD3R mainboard.
During CPU overclocking, the difference lies between 30 and 60 W and although the graph only shows results during overclocking, the same is true for the nominal processor mode. Moreover, I have to say that Gigabyte GA-EX58-UD3R is a very energy-efficient mainboard for LGA1366 standards. It has very effective processor voltage regulator circuitry, doesn’t have additional PCI Express bus controller. Besides, we overclocked Intel Core i7-920 without increasing its Vcore, unlike Intel Core i7-860. All in all, LGA1156 and LGA1366 platforms are totally incomparable in terms of power consumption.
Our today’s article has discussed several hot topics that is why our conclusion is going to be longer than usual. At first we would like to say that LGA1156 platform makes an overall great impression, but the Intel Core i5-750 processor turned out a real disappointment. Not only doesn’t it support Hyper-Threading and 12x memory frequency multiplier, but it also overclocks just a little higher than Intel Core i7-860 despite our expectations. It turns out that Core i5-750 can only compete against old Intel Core 2 Quad and quad-core AMD processors. However, Intel Core i7-860 is a fully-fledged high-performance processor with good overclocking potential. But this is where a very interesting question pops up to mind: which CPU is better – Intel Core i7-860 or Intel Core i7-920? The answer to this question will depend on the conditions, in which you are going to use these CPUs, and on the parameters that matter to you most.
In terms of performance, if you do not support overclocking or encourage playing with the BIOS settings, then you should go for Intel Core i7-860. It will be faster in nominal mode due to higher clock speed and memory frequency, besides, do not forget about more flexible Turbo Boost implementation. However, Intel Core i7-920 CPU will be faster during overclocking. Both these CPUs overclock almost to the same frequency, but the base clock and all frequencies connected with it, such as memory frequency, will be higher by Intel Core i7-920 due to lower clock frequency multiplier, provided that both CPUs have been overclocked to the same speed. Besides, do not forget that Intel Core i7-920 works with triple-channel memory. Moreover, when you use high-speed DDR3 memory the frequency of the integrated memory controller will be higher than by Core i7-860. In terms of pricing, these CPUs cost about the same, but the total cost of an Intel Core i7-860 platform will be lower due to fewer memory modules and not so expensive mainboards. as for the power consumption, LGA1156 and LGA1366 platforms cannot even compare, because the latter is way more resource-hungry.
My personal choice would definitely be Intel Core i7-860. Only once I worked with this processor for the first time, I thought that maybe it is time for me to finally switch to a quad-core CPU. And despite the obtained results, you don’t have to give up dynamic Turbo Boost implementation during overclocking. The static implementation is overall faster, no doubt, since we have to lower our overclocking when we switch to dynamics. However, we get a more flexible and more economical system in the end, which is also very important. Not all of you have used up the entire overclocking potential of their processors for different reasons. Some of you probably overclocked using phase cooling systems (Freon based) in order to get everything the CPU can offer. Some may have specifically searched for the most optimal system components in order to get maximum gain with minimal financial investments. Some may have just overclocked their processors as far as their cooling, mainboard and other system components allowed. When Lynnfield processors came out, some started to choose static and some dynamic implementation of Intel Turbo Boost technology.
But let’s get back to where we started this review – Asus P7P55D Deluxe mainboard. It is simply amazing. Asus Company primarily stresses the new functions such as “OC Tuner Utility” that allows automatic CPU overclocking. I personally do not really care much about features like that for understandable reasons. At this point there is no utility that can provide the same results as manual overclocking, although I have to admit that technologies like that will be extremely helpful for overclocking beginners. I mostly like the fact that even when you change the processor clock frequency multiplier and core voltage, all power-saving technologies remain up and running. Now the only thing that may limit our success is the CPU itself and its cooling system. And of course, you shouldn’t forget about the typical peculiarities of Asus mainboards: good accessories bundle, well-balanced design, high-quality electronic components, numerous brand name functions and technologies, excellent overclocking potential, long warranty periods. We have just started our review series devoted to LGA1156 mainboards, but I can hardly imagine that there is any mainboard out there that could supersede Asus P7P55D Deluxe. At best, the potential rival may prove just as good. It is for the first time in a while that I am totally raving about an Asus mainboard and I really hope that the company will continue at the same pace.
As a result of our today’s test session we would like to award Asus P7P55D Deluxe mainboard with our Editor’s Choice title as the best enthusiast LGA1156 mainboard:
