Intel DX58SO: a Completely Different Mainboard

We are going to talk about numerous features of the flagship LGA1366 mainboard from Intel that belongs to the Extreme product series. Read our review for details!

by Doors4ever
07/02/2009 | 12:15 PM

We first learned about the features of the Intel DX58SO mainboard back in early November 2008, right after the NDA lift and before the official launch of the new Intel platform. Our article called “New Hit from Remake King: Intel Core i7 Review” contain a short chapter devoted to this mainboard. However, this solution turned out very interesting and quite unique that is why we decided to devote the whole separate article to the numerous features and functionality of the flagship LGA1366 Intel board from the extreme series.

 

We have already reviewed quite a few LGA1366 mainboards by now and they all have a lot in common. It is quite natural because all of them are based on the same exact Intel X58 Express core logic set, support Intel Core i7 processors and DDR3 SDRAM. Each mainboard has its own peculiarities, but they all have a lot of features in common. The main impression we have from Intel DX58SO is that it is totally different from all other solutions. The differences are present in every aspect: PCB layout, BIOS structure and functionality, even the accessories bundled with the board are different. A different mainboard doesn’t necessarily mean a bad mainboard, but it always implies unusual and therefore interesting.

Package and Accessories

The box Intel DX58SO mainboard ships in doesn’t strike you with large size or extravagant design, flip-open front covers or clear windows. The company believes that the packaging should serve two primary purposes: protect its contents from transportation damages and inform the potential user of what is inside. The box copes perfectly well with both these tasks, and everything else is excessive luxury.

The back of the box contains a photo of the Intel DX58SO mainboard, a list of its features and peculiarities and a description of included accessories:

If we mention everything that we found inside the box, it will be a pretty long but not very interesting list. That is why we decided to mention only those accessories that have certain practical value for us, as users. The bundle includes the following items:

I have to say that it is a pretty modest bundle for a flagship mainboard, not taking into account the whole bunch of marketing booklets and materials also included with the board. As you may have already noticed, the traditional paper manual is not included with the board. Its electronic version is available on the DVD disk with drivers. It also contains a rather long list of additional software:

So, what do we have here in the end? When you buy Intel DX58SO mainboard, you don’t pay extra for pretty box design – that’s a plus. We don’t spend money on excessive amount of cables and adapters. We can easily get all that, if and when we need it. Instead, the board is bundled with a fan and a retention kit for its installation on top of the chipset North Bridge heatsink. Usually we have to arrange additional cooling ourselves, but here it has been taken care of for us, so it is an advantage as well. And finally, it is really cool to get so much free software. The list above contains a few pretty useful tools that are not available for free. Some of you wish they included a bracket with a few extra USB ports, some may be upset to see no connecting bridges for the multi-graphics card configurations, but overall, our first impressions are highly positive with the exception of marketing materials that are rarely useful or interesting.

PCB Design and Functionality

When we said that Intel DX58SO mainboard struck us as unique and unusual, we implied its layout as well. This is the only mainboard of all LGA1366 solutions we have tested so far that has memory DIMM slots located above the processor socket and not to the right of it. If you take a closer look at the mainboard PCB, you will, however, notice that the relative location of the processor voltage regulator circuitry, processor socket, memory DIMMs and chipset North Bridge didn’t really change compared to the traditional layout. It actually looks like they simply turned the upper part of the PCB by 90 degrees counterclockwise:

It is hard for me to explain why they decided to do it this way, but I am sure Intel engineers had their reason, because dramatic modifications like that are never performed without a really good one. However, you can easily notice the negative effects this “turn” had on the layout, the biggest one being fewer memory DIMM slots: four instead of six. The length of the mainboard PCB is limited by existing standards that is why they would have to cut down on the expansion slots, if they decided to keep all six DIMMs onboard. Besides, they also had to move the 8-pin ATX12V power supply connector into a very inconvenient spot: a little below the chipset North Bridge. We can’t say that these are very serious drawbacks, but they are indisputable, while the advantages of this particular design solution still remain unknown to us.

The photograph shows very well the heatsinks over the processor voltage regulator and chipset components. Intel as always uses very secure retention for chipset heatsinks that consists of metal brackets and clips. As for the heatsinks over the voltage regulator MOSFET, they are attached using push-pin spring clips. By the way, Intel DX58SO only has solid-state capacitor with polymer electrolyte in the voltage regulator circuitry for the processor, chipset North Bridge and memory. Otherwise, we see mostly regular liquid electrolytic capacitors.

The lower part of the PCB looks more traditional, except a lonely Power On button for those users who like to experiment in open testbeds. However, I have to correct myself: there are a few pretty unique things here as well. For example, they usually recommend using a special four-pin power connector on the PCB, the same as on PATA hard disk drives or optical drives, to provide additional power to PCI Express x16 slots when there are high-end graphics cards installed into them. There is a connector like that on Intel DX58SO, but you can also use a second power connector just like the one on SATA HDD together or instead of it. The pins for front panel buttons and indicators are located not in the traditional lower right corner, but close to the center of the bottom edge of the PCB. We did connect all of them, but it required some effort on our part, because the cables were at their maximum stretch.

You can install two graphics cards into PCI Express x16 slots. Besides, there are two PCI Express x1 slots, one PCI and one PCI Express x4 slot. The mainboard officially supports ATI CrossFire as well as Nvidia SLI graphics configurations. It is pretty interesting that PCI Express x4 slot has an open connector end and there is a special retention lock a little farther, which means that if you need, you can use this slot for the third graphics card.

As you can see, Intel DX58SO doesn’t support PATA drives and FDD. There are also no PS/2 keyboard and mouse connectors on the back panel. Instead there are two eSATA ports implemented via Marvell 88SE6121 controller. Besides, there are eight USB ports and one IEEE1394 port implemented via Texas Instruments TSB43AB22A. Gigabit network is delivered by Intel WG82567LM chip, while eight-channel sound is provided by Realtek ALC889.

Intel DX58SO mainboard is equipped with four small LEDs. One of them indicates that the board receives power, other two will report CPU or processor voltage regulator overheating. The fourth LED shows HDD activity status.

It is easier to estimate the quality of PCB design with the schematic components layout, but it seems to me not too informative (click on the image to see components description):

The photograph on the back of the box is not so detailed, but much more illustrative:

We are going to wind up our Intel DX58SO PCB design discussion with a detailed list of technical specifications taken from the manufacturer’s web-site.

Overall, the functionality of Intel DX58SO mainboard meets contemporary requirements quite well. You just have to remember that it doesn’t support FDD and PATA and keep in mind certain design drawbacks that we have already pointed out above.

BIOS Setup

You can update the BIOS on Intel DX58SO mainboard in several different ways. You can use a floppy (if you have a USB floppy drive), CD disk, USB drive or just launch a special executable file with the BIOS image and updating utility right from Windows. The latter method is the most popular one, because it requires minimum effort and experience from the user. There appeared one more way recently: reflashing tool integrated into the BIOS that can be launched on mainboard boot-up by pressing F7 key. The utility sees the partitions on all connected hard drives even if they have been formatted in NTFS, USB drives and optical drives. Just select the BIOS image on one of the media to complete the BIOS update. In this part of our article we are going to refer to the latest BIOS version available at the time of tests – BIOS 4014 from 05.07.2009.

When we talked about unusual PCB design of Intel DX58SO mainboard, we tried to avoid any sharp evaluations. Yes, there are drawbacks, but they are not critical, they may even be regarded as advantages at some point. Fewer memory DIMM slots are not a good thing. However, it did shorten the signal lines, so we could hope for higher system stability or even higher memory overclocking results. During our tests we use a pretty unique processor cooler – Cooler Master GeminII. Theoretically, longer heatsink wing should spread to the right over the memory modules thus delivering additional cooling to them. However, our Kingston HyperX DDR3-1866 KHX14900D3T1K3/3GX modules are equipped with very tall heat-spreaders, so we have to position the cooler with the longer side facing down, where it will be cooling the chipset North Bridge heatsink instead of the memory modules. This time we could easily install the cooler in a standard way. Even the location of the ATX12V processor power supply connector that will be pretty inconvenient for most users out there, can still please the owners of system cases with the power supply at the bottom of the case, because they will no longer need an extender cable.

As for the BIOS of Intel DX58SO mainboard, we can be more explicit here: it is not only unusual, but also very inconvenient to work with. The problem is that all meaningful parameters are spread out over different sections and numerous sub-menus, so you have to do a lot of extra surfing in order to make all the necessary changes and adjustments. Sometimes we can’t even get the logics behind the placement of certain parameters in specific sections. For example, why did they include the options for changing the number of active processor cores and enabling/disabling Intel Hyper-Threading technology into the very first section called “Main”? Wouldn’t it have been better to keep them in the “CPU Features” section together with all other CPU-related settings? The problem is that there is no section like that at all in the BIOS of Intel DX58SO mainboard.

“Advanced” section contains several sub-sections, but despite their simple and logical names, the actual contents of these sub-sections may surprise anyone who has never dealt with Intel mainboards before.

For example, the very first sub-section is called “Boot Configuration”. Instead of the preferred order of boot-up devices, we see parameters for fan rotation speed management.

During the boot-up procedure the board displays a picture that can be replaced but cannot be disabled altogether. In the “Boot Configuration” sub-section you can enable a number of reminders that will be displayed in the startup window, such as: press F2 (not Del) to access the BIOS, press F7 to update the BIOS, press F10 to load the menu where you could change the order of boot-up devices, press F12 to boot via network. There are POST codes displayed in the lower right corner.

The next sub-section of the “Advanced” section that is of interest to us is called “Hardware Monitoring”. This sub-section can’t boast very rich functionality; there are only informational parameters there: no adjustable settings. Among the advantages I could only point out temperature monitoring in four spots. Besides system and CPU temperature that any mainboard can monitor, Intel DX58SO also reports the temperature of the chipset North Bridge and processor voltage regulator circuitry.

At first, as usual, the “Performance” section tries to scare us away by enlisting five different dangers awaiting us inside.

However, if you are brave enough to respond “Yes” instead of the suggested “No”, there won’t be anything scary there: only two parameters and three sub-sections, while everything else is simply the info about the nominal, expected and current system settings.

All information there is correct except for the processor clock frequency multiplier. On the previous screenshot the board for some reason decided that the multiplier will be 26 after system reboot, which is simply impossible for Intel Core i7-920 processor, which nominal clock multiplier is 20 and can only increase to 22 maximum due to Intel Turbo Boost technology.

Let’s check out the first sub-section called “Processor Overrides”. There are very few parameters, but all of them are pretty interesting. “Static CPU Voltage Override” allows setting the desired CPU Vcore in the interval from 1.0 to 1.6 V with 0.0125 V increment. “Dynamic CPU Voltage Offset (mV)” will only add certain voltage to the nominal with variable increment that for the most part equals 6 mV (0.006 V) and maximum value of 489 mV (0.489 V). “Enhanced Power Slope” parameter corrects the processor voltage drop under heavy load, the so called “VDroop”. The interesting thing is that all other mainboards with such functionality of the solutions we have tested so far, have only two possible settings for this parameter. By default VDroop is disabled, which corresponds to “100% Slope” value of the “Enhanced Power Slope” parameter; in all other cases it is enabled, which corresponds to “No Slope”. Intel DX58SO also has an intermediate value of “50% Slope”, which indicates that the voltage drop is partially corrected.

Then I checked out the last sub-section in the “Performance” section called “Bus Overrides” and got very upset that there was no option for adjusting the Uncore frequency of the North bridge part integrated into the CPU.

It turned out that I simply wasn’t paying enough attention. There is a “Memory Overrides” sub-section that allows configuring all parameters connected with the memory subsystem, such as Uncore frequency, memory frequency, timings, Uncore voltage and memory voltage.

It is a good idea to gather all parameters connected with the memory subsystem configuration in a single sub-section, however, the implementation of this idea could have been better. By default, all these parameters are set automatically. If the memory modules support “X.M.P.” technology (Extended Memory Profile technology that records advanced memory settings into the modules SPD, such as frequency, timings and voltages), then you can select a corresponding profile and all parameters will again be configured automatically. Otherwise, you must set all parameters manually. Even if you simply have to set the nominal voltage or frequency for your memory modules, you will have to adjust all timings as well. Most other mainboards in this case use information from the modules SPD, but Intel DX58SO seems to be able to use it only in Auto mode. It is extremely inconvenient!

The next section where we come across processor technologies one more time is called “Security” for some reason. And while it makes sense to some extent in regards to “XD Technology”, it has absolutely nothing to do with virtualization technology.

The “Power” section, which we skip for the most part in our reviews, is worth checking out this time, because it also contains several processor related technologies, namely, power-saving ones. This time it is pretty clear what they are doing in this particular section, but it would be much more convenient if all important parameters were gathered in one place and we didn’t have to look for them in all BIOS sections and sub-sections.

The “Boot” section finally offers us to change the order of boot-up devices. We have no problems with this section. Moreover, it is extremely functional. It is one of the few great examples of corporate orientation of Intel mainboard.

The “Exit” section contains several logical functions and also allows to quickly restore the only complete BIOS settings profile. You can’t add a name or description to it, just “Custom Defaults”. Of course, it is nice that there is at least one profile, but it definitely not enough, especially since you have to navigate between different sections to adjust the necessary BIOS parameters.

There is one more inconvenience: the need to press “extra” keys. Most mainboards allow you to use cursor keys to find the desired parameter in order to start changing it right away. Sometimes, you can press “Enter” to open a new window where you can use arrow keys to select the value or just enter it from your keyboard. Intel DX58SO mainboard doesn’t allow entering parameter values from the keyboard, and you always need to press “Enter”, just to enter the editing mode for the selected parameter. Very rarely you will get a window with supported values, but in most cases you just have to use “+” and “-“ keys to go through all the values until you hit the right one. After that you have to press “Enter” again or use cursor keys to exit the editing mode. That’s way too much extra work!

I believe it is obvious, that Intel absolutely has to change the way their BIOS is organized. Other mainboard makers are constantly modifying their BIOS Setup trying to make configuring your computer system as simple and quick as possible. Intel developers, however, seem to be still held captive by the habits and traditions of their “corporate” past. The company seems to believe that only system administrators will use their mainboards and they will only welcome additional difficulties. The only consolation for us in this case is that the BIOS of Intel DX58SO mainboard has everything necessary for successful overclocking and optimization of the system. The increments are sometimes too big, some parameters are often very inconvenient to adjust, but it is doable, which is the most important thing here.

Testbed Configuration

All practical tests were performed in the following testbed:

We used Microsoft Windows Vista Ultimate x86 Service Pack 2 OS and ATI Catalyst 9.5 graphics card driver.

Performance in Nominal Mode

As usual, we didn’t discover any problems with Intel DX58SO mainboard’s performance in nominal mode. No wonder, since the problems usually occur only if the board is defective, and if it is well put together, there should be no issues. By default the memory works at 1066 MHz frequency with 8-8-8-19-1T timings. The clock multiplier and core voltage of our Intel Core i7-920 processor lower in idle mode.

If the CPU utilization is pretty high, its clock multiplier increases to 21 due to Intel Turbo Boost technology.

In those cases when the CPU utilization is relatively low and only one processor core of the four is loaded, its clock frequency multiplier may increase even to 22.

I have to point out that I noticed the clock multiplier go to 22 much more often on Intel DX58SO mainboard than on most other board we tested in our labs. Looks like only Asus P6T could easily push the multiplier to 22 under relatively low workload, which helped it to outperform the competitor – Gigabyte GA-EX58-Extreme in synthetic Everest benchmarks. It was then that we even brought back the single-threaded SuperPI utility into our benchmarking suite in order to see how big this advantage was. But strange as it might seem, Asus mainboard lost to Gigabyte in this test and the advantage proved to be virtual after all. This time, as we will see from our performance comparison charts, Intel DX58SO mainboard has a pretty real advantage in SuperPI due to frequently used maximum processor clock frequency multiplier.

Today Intel DX58SO will compete against the recently reviewed MSI Eclipse Plus. Both these boards set the processor and memory parameters to defaults. The only exception was the power-saving technologies that we enabled manually on both boards.

We don’t see any significant performance differences between the two boards, besides the considerably higher memory read and write speeds on Intel DX58SO, which might be explained by more frequent use of maximum processor clock frequency multiplier.

All mainboards set the memory frequency at 1067 MHz in nominal mode. Only sometimes the manufacturers cheat and increase the frequency to 1333 MHz to get a small advantage over the competitors. However, it is very low frequency for DDR3 SDRAM. For instance, our Kingston HyperX DDR3-1866 KHX14900D3T1K3/3GX memory modules can work at 1867 MHz frequency and support X.M.P. (Extended Memory Profile) technology, which allows to easily get the desired settings. Advanced memory settings, such as frequency, timings and voltages, are recorded in the modules SPD. So, all you need to do is select the corresponding profile in the mainboard BIOS and you will easily get the desired result without any effort and manual adjustment of the parameters.

Unfortunately, not all mainboards work with X.M.P. technology correctly, but Intel DX58SO is not one of them. When we selected the first profile, the board set 1867 MHz frequency, increased the memory voltage to 1.66 V and Uncore voltage – to 1.5 V. By selecting the second profile we increased the memory frequency to 1600 MHz and Uncore voltage – to 1.45 V. Both these settings are high enough, especially keeping in mind that the default Uncore voltage on Intel DX58SO mainboard way lower than on most mainboards: only 1.15 V. As a rule, lower values will do just fine without threatening the system stability. And this is exactly what we saw: our memory worked at 1867 MHz frequency with “QPI/Uncore Voltage Override” set to 1.275 V.

In conclusion I would only like to repeat once more time that when the CPU was working in its nominal mode, Intel DX58SO mainboard didn’t reveal any issues or specific operational peculiarities. However, all the interesting things started happening during overclocking, which is exactly what we will discuss next.

Turbo Boost Mode and CPU Overclocking Specifics

Let’s start with the fact that Intel DX58SO mainboard proved just as good as many other mainboards during work at high base frequencies. When we lowered the processor clock multiplier, the board performed stably and passed short tests successfully at 215 MHz frequency. It is a very good result. It guarantees that the board’s ability to reach high frequencies won’t stall our overclocking success.

Now we have to say a few words about the specific peculiarities of Intel DX58SO mainboard. First, all Intel processor power-saving technologies stay up and running, even if we increase the processor core voltage. You may have noticed that there are two parameters in the mainboard BIOS for adjusting the processor core voltage. If we use “Static CPU Voltage Override” to set the CPU Vcore the board will lock the voltage at this value independent of the current system load. If we add a certain value to the nominal voltage setting using “Dynamic CPU Voltage Offset (mV)”, Intel processor power-saving technologies continue working and lower the processor Vcore in idle mode in proportion to the initial setting increase. This is truly great news! Unfortunately, the acknowledged overclocker mainboards from Asus and Gigabyte, can’t boast anything like that.

The next thing worth mentioning has to do with Intel Turbo Boost implementation during Core i7-920 CPU overclocking. So far we are familiar with two implementation approaches. We come across the first one during our tests of Asus and Gigabyte mainboards. Even under very heavy CPU utilization its clock multiplier still increases to 21. We only saw it drop to the minimal value of 20x when the CPU temperature exceeded 94 °C and overheating protection kicked in on Asus P6T board. Another implementation of Intel Turbo Boost was revealed during MSI Eclipse Plus tests. Only under relatively small load the processor clock multiplier increased to 21. When the load was fairly high, the processor power consumption and temperature started growing, and the board returned the clock multiplier back to its initial value of 20x. Intel DX58SO mainboard also behaves differently, although I believe I should change the term here. Intel should know best of all what the right implementation of Turbo Boost technology is, shouldn’t they? So, we should call this particular implementation “correct”, and another one “different”.

In fact, “correct” is not necessarily the best, each approach has its advantages and drawbacks. Obviously, the boards that always increase their clock frequency multiplier to 21x will perform better in heavy multi-threaded applications, because their rivals with “correct” technology implementation will have their multiplier at the nominal value. However, as our recent tests of MSI Eclipse Plus mainboard showed, this advantage is not always indisputable. Since the board lowers the clock multiplier to its nominal value under heavy load, the resulting CPU frequency also lowers, which allows us to raise the base frequency a little higher during overclocking. As a result, the performance difference under heavy load is not that dramatic anymore. Moreover, under moderate load in a number of applications MSI Eclipse Plus would even take the lead due to higher base frequency.  In those cases when the multiplier is increased to 21 the CPU frequency is higher and so is the memory frequency. Since heavy multi-threaded applications are still not so widely spread as applications creating moderate single- or dual-thread load, the “correct” Intel Turbo Boost implementation appears of bigger benefit to “typical” users.

Overall, our updated terminology again doesn’t really depict the actual state of things, because the “correct” implementation is not always good, and the “incorrect” implementation is not that bad and can provide even higher performance in a number of cases. Therefore, I suggest that from now on we should use the term “dynamic” for the “correct” implementation of Intel Turbo Boost technology, because in this case the multiplier changes (increases or lowers back to its nominal value) depending on the workload. As for the “incorrect” implementation, we will call it “static”: the multiplier is always one step higher, no matter what the workload is. Just in case I would like to remind you that we are talking only about overclocking here. In the nominal CPU mode almost all mainboards we have tested so far employ “static” approach: no matter how heavy the load is, the multiplier is always set one step higher. At least, this is true for Intel Core i7-920 processor, because its power consumption never exceeds the acceptable limits in the nominal mode under any workload. And in both cases the multiplier will go down to the minimal value of 12 (passing through a number of intermediate lower values), if the workload lowers or disappears completely in idle mode.

But what if you are not a typical user and there are quite a few heavy-duty multi-threaded applications among your everyday tasks? Does it mean you will have to avoid mainboards with dynamic implementation of Intel Turbo Boost technology, if you create and process multimedia content, work with models, sound images or video? In case of MSI Eclipse Plus mainboard, it might be not the most optimal choice for you. However, the BIOS of Intel DX58SO mainboard has a beautiful parameter called “CPU VR Current Limit Override”. Once it is set to “Enable” the board stops reacting to high current received by the CPU, its multiplier no longer lowers to 20x in case of heavy load. Intel DX58SO mainboard allows easily switching from the dynamic implementation of Intel Turbo Boost technology to a more convenient and beneficial in this case static implementation!

Frankly speaking, by this time all my initial skepticism about Intel being able to make a good overclocker mainboard was gone. Intel DX58SO mainboard seemed to be an ideal universal overclocker mainboard. See for yourselves:

Unfortunately, I forgot that there is no such thing as an ideal mainboard, but one more peculiarity of the Intel DX58SO solution reminded me of that. During our experiments, I discovered that “Enhanced Power Slope” parameter that should prevent the processor core voltage from dropping under heavy load, works too well. We have already seen many times that enabling parameters like that on different mainboards caused the voltage to increase instead of going down, and sometimes this increase appeared quite significant. For example, DFI LanParty JR X58-T3H6 or EVGA X58 SLI Classified. To our great disappointment, Intel DX58SO is also one of them. We easily overclocked our system to 185 MHz base frequency using the dynamic implementation of Intel Turbo Boost technology. We have already demonstrated the effectiveness of this overclocking during our tests of MSI Eclipse Plus mainboard. We spend the rest of our time trying to achieve stability during overclocking with a locked processor clock frequency multiplier, but without luck.

The most important thing during CPU overclocking is to ensure stable power flow. Trying to prevent processor core voltage from dropping under heavy load we set “Enhanced Power Slope” to “No Slope”. In this case the processor core voltage would constantly increase under heavy load even if we hadn’t increased it in the mainboard BIOS left it at the nominal value. As a result, after 2-3 LinX cycles the core temperature increased to 96-98 °C and then we saw the blue screen of death or terminated the tests ourselves. By the way, I believe that Intel engineers are very well familiar with the problem of voltage increase under heavy load that is why they introduced this strange intermediate “50% Slope” value of the “Enhanced Power Slope” parameter. In other words, let the voltage increase, but not as dramatically as in case we set this parameter to “No Slope”. We have never seen anything like that on any other mainboard, but it didn’t really help: the voltage increased a lot anyway.

Ok, we gave up the unsuccessful “VDroop” implementation and left “Enhanced Power Slope” parameter at its default value of “100% Slope”. To make up for the anticipated voltage drop under heavy workload, we had to increase it right from the start in the BIOS. It actually got totally absurd: power-saving technologies were formally working and increased core voltage did get lower in idle mode, but it dropped only to the level where it would normally go up to under heavy load. But even in this case, either the voltage wasn’t increased high enough, which caused the system to hang as soon as the voltage dropped down, or the voltage turned out too high and BSOD appeared after the temperatures jumped to 96-98 °C.

You don’t have to use Intel Turbo Boost technology to ensure that the CPU remains stable when overclocked with a fixed multiplier. Intel DX58SO mainboard can work fine even at 215 MHz base frequency. We only had to increase this frequency to 195-197 MHz to overclock our particular processor sample to the desired 3.9-3.95 GHz even without increasing the multiplier to 21. The multiplier is set to its nominal value of 20x, the load is pretty high that is why it won’t go any lower than that, but the system is again very unstable for the same exact reasons. Either the temperature is too high because of increased core voltage, or the voltage is insufficient to guarantee stability.

I personally am a big fan not only of overclocking, but also of different power-saving technologies. I like it when the CPU works at the maximum of its potential only when it is really necessary, but is energy-efficient and quiet in idle mode. However, I am sure that there are people among our readers who mostly care about hitting the maximum processor frequency and do not really bother about high power consumption in this case. That is why we decided not to resort to any power-saving technologies and set the processor core voltage using “Static CPU Voltage Override” parameter and not “Dynamic CPU Voltage Offset (mV)” as we have just done before. Unfortunately, even this approach didn’t help us avoid CPU Vcore drop under heavy load or its extreme increase when e enabled VDroop function.

The next two graphs will show in a more illustrative way what happens when we change the “Enhanced Power Slope” value. The first graph shows a system overclocked to 185 MHz base frequency with dynamic implementation of Intel Turbo Boost technology. To ensure processor stability during this overclocking attempt, we used “Dynamic CPU Voltage Offset (mV)” parameter to add another 0.055 V (55 mV). We are not going to protect ourselves against processor Vcore drop. “Enhanced Power Slope” parameter set by default to “100% Slope”. The preliminary system stability test was performed using ten runs of LinX test.

First of all, we are interested in checking out the dynamics of processor voltage changes: its graph is green and is built along the left axis. Before the bests started, the CPU was idling, Intel processor power-saving technologies were up and running and the CPU received only 1.11 V. Right after we launch LinX test utility the voltage increases, but then drops immediately to 1.15 V under heavy load. Later on we see the same things repeat all over again: in the intervals between calculations the voltage increases to 1.27-1.29 V, but under heavy load it rapidly drops to 1.15 V. The processor temperature graph is in red color and built along the right axis. In idle mode the core temperature is about 40 °C, during the tests it increases almost to 80 °C and then goes down again when the test is complete.

The amplitude of voltage fluctuations is relatively big, because we are not protected against processor voltage drop under heavy load. But luckily, even a low voltage of 1.15 V is enough to ensure stability of our Intel Core i7-920 processor overclocked to 3.7 GHz. When we use dynamic Intel Turbo Boost implementation, the processor clock multiplier remains equal to 20x under heavy load created by eight computational threads of LinX utility. With 185 MHz base frequency it produces 3.7 GHz resulting CPU speed. If we decide to switch to static Intel Turbo Boost, then with 21x multiplier the CPU frequency will increase to 3.9 GHz and 1.15 V Vcore will no longer be enough. This is when we need to protect ourselves against voltage drop that is why we set “Enhanced Power Slope” parameter to “No Slope” and run our LinX test one more time without making any other changes.

The startup voltage remained the same: 1.11 V, however the system started acting differently during the test. Fluctuations amplitude got much smaller; the protection against voltage drop is obviously working fine. The minimal processor core voltage is 1.22 V in the beginning of the very first computational cycle, but even before the end of it the core voltage increases to 1.23 V and then to 1.24 V, and later on almost never drops below 1.25 V. the maximum Vcore between the test cycles is at first 1.3 V and then gets even higher than that. The temperature increases gradually reaching the maximum of 97 °C. This time we failed the test even without switching to the static Intel Turbo Boost implementation when the multiplier increases to 21. We saw the blue screen of death during the sixth test cycle.

If you have been reading our review attentively enough, you will remember that processor Vcore has been increased right from the start to ensure stability during overclocking when no protection against voltage drops is in place. However, now when we enable VDroop function, there is no need to increase the voltage anymore and additional volts will only heat up the CPU excessively. Everything is absolutely correct: we haven’t changed a thing except for the “Enhanced Power Slope” parameter for the sake of comparison. In fact, even if we leave the processor core voltage at its nominal of 1.225 V, it will still increase eventually beyond 1.3 V when we enable protection against voltage dropping.  And that is way too high. It causes the temperature to hit 98 °C and prevents us from using VDroop function. That is why we couldn’t get our overclocked processor to work stably with a constant multiplier under heavy loads.

Unfortunately, Intel DX58SO proved to be no universal mainboard. It is a real pity. However, it doesn’t diminish any of its advantages and strengths. As you remember, dynamic implementation of Intel Turbo Boost technology did allow us to achieve very good results during overclocking.

Performance during Overclocking

I believe that you understand well by now why we picked MSI Eclipse Plus mainboard for our performance comparison against Intel DX58SO: they have a lot in common, for example, overclocking results obtained at 185 MHz base frequency using dynamic Intel Turbo Boost implementation. Both mainboards have Intel processor power-saving technologies up and running even though the voltage is increased, so not only the multiplier, but also the CPU core voltage gets lower in idle mode.

During work in heavy-duty multi-threaded applications the processor clock multiplier is at its nominal value of 20x.

Under moderate load Intel Turbo Boost technology increases the multiplier to 21x.

This is pretty much all the boards have in common, because during CPU overclocking on MSI Eclipse Plus mainboard we can no longer control “C State” parameter, because it becomes unavailable to us, while Intel DX58SO mainboard has no restrictions like that. As a result, under low workload when only one processor core was active, its multiplier increased to 22x.

The next graph shows the dynamics of the Intel Core i7-920 processor multiplier change during the tests performed with SuperPI utility. Before the tests started the CPU was idle and its multiplier was lowered to the minimal value of 12x. Right in the beginning of the test the multiplier increased to 21 and got as high as 22 from time to time. Upon completion of the test, the multiplier went back to its minimal value. Therefore, the record-breaking results obtained during the calculation of 8 million PI digits on Intel DX58SO mainboard are not surprising at all, because the CPU frequency sometimes exceeded 4 GHz!

We see a completely different picture when the test is performed on a system built around MSI Eclipse Plus mainboard. Disabled “C State” parameter wouldn’t let us even try increasing the multiplier to 22: it is always at 21x under heavy load.

The differences between mainboards are even greater in multi-threaded applications creating much heavier load. The graph below shows the changes in the clock frequency multiplier of Intel Core i7-920 processor during a one-hour stability test in Prime95. You can clearly see how uneven the load created by this test program is. At first it is relatively low, which allows Intel DX58SO mainboard to increase the multiplier to 21x pretty often. Then we see a long interval of higher load when the CPU is almost always working with its nominal multiplier of 20x. After that the load gets variable again and the multiplier is switching between 20 and 21 all the time.

Now it is clear why you most often see errors only at least 15 minutes into the Prime95 stability test, if there were none during the very first cycles. This is when the load increases and the overclocked system is no longer able to handle it.

And now let’s see how the same exact test went during CPU overclocking on MSI Eclipse Plus mainboard. Back then, when we tested MSI Eclipse Plus mainboard, we mentioned that the only case when the clock frequency multiplier of the Intel Core i7-920 processor would increase to 21x guaranteed, is under relatively low working load created by approximately two threads of Prime95 utility. Unlike Intel DX58SO, MSI Eclipse Plus mainboard is simply unable to increase the processor multiplier when all eight threads of Prime95 are working: it stays at 20x throughout the entire test.

Therefore, it is not surprising that Intel DX58SO mainboard is often ahead of its competitor. Although both mainboards employ dynamic Intel Turbo Boost technology, the specific implementation is different in each particular case. Intel mainboard allows the CPU to work with an increased clock multiplier much more often.

However, I have to make one very important comment about MSI Eclipse Plus as well as Intel DX58SO mainboard. We value our reputation of a credible information source that is why we never report any unconfirmed facts. That is why if we claim that the CPU was stable during overclocking, this is the way it is. It isn’t hard to check out the systems stability with static Intel Turbo Boost implementation. In this case, the multiplier for the Core i7-920 processor increases to 21 under any load, even the heaviest one. Obviously, if the CPU copes with it fine, then stability won’t be an issue under lower working loads.

Things look completely different when dynamic Intel Turbo Boost implementation is employed.  The system decided on its own what multiplier to use and changes the processor core voltage accordingly. In fact, the stability tests we use in our suit perfectly for that. LinX utility performs calculations in stages; the load created by Prime95 also changes dynamically that is why the CPU switches the operational modes as well as voltage and multiplier combinations during the same test cycle. Even SuperPI tests checking the ability of our overclocked processor to work with 22x multiplier have been all completed successfully. However, I can’t exclude the probability of finding a certain voltage+ multiplier combination that may cause an error. During dynamic Intel Turbo Boost implementation common tests may be not enough. You may need to work in different applications for quite some time before it would be fair to claim that the overclocked system is in fact stable.

Conclusion

Well, let’s recall what we know about Intel DX58SO mainboard. It has a not very rich but sufficient accessories bundle and a long list of additional free software. It boasts unique PCB design, though it cannot be considered a “PC builder’s nightmare”, because some of the design peculiarities of this board turn into unique advantages at some point. The BIOS of Intel DX58SO mainboard is its obvious weak spot. It is not ergonomic or convenient to work with that is why it is hard to use this board as a testbed for anything else. You have to make a lot of extra moves to make more or less serious configuration changes. However, it shouldn’t be an issue for a mainstream user: you don’t really change your home system configuration on a daily and even monthly basis. You take the time and configure your system once and then live happily ever after. Yes, no matter how strange it may seem, but Intel DX58SO BIOS has all parameters and settings necessary for successful overclocking and configuration optimization. Moreover, in some aspects it even surpasses some leading overclocker solutions. Unfortunately, as it often happens, not everything has been done impeccably: the voltage adjustment increments are sometimes too bit, some of the parameters could have been implemented better. However, despite a number of issues we pointed out in our review, we managed to achieve very good results during overclocking. This board is not just as good, but is even superior to some of its competitors.

The mere fact that Intel DX58SO mainboard was brought into this world already means a lot. It is the absolute “strongest“ Intel mainboard lately. The company no longer disregards the needs of overclocking fans. Now they announce openly that their mainboards are designed for overclocking, and most importantly, these words are backed up by real good results. When we talk about mainboards, we first of all imply the solutions from the leading Taiwanese manufacturers, such as Asus and Gigabyte. Intel DX58SO mainboard is the first warning shot fired at them. Looks like they got so carried away by competition against one another that they completely forgot about a huge and powerful corporation like Intel, which can actually prepare some unexpected surprises.

Frankly speaking, I don’t hope to see the new BIOS versions for Intel DX58SO to have new revised structure and interface, and to be free from the issues bothering overclockers and advanced users. They will obviously first of all focus their efforts on adding support for the new processors and eliminating compatibility problems. Moreover, I don’t expect to see any dramatic changes even in a couple of next upcoming products. Firstly, it is hard for a large corporation like Intel to change their vision all of a sudden; secondly, they are still responsible for providing proper support for their already existing mainboards. In this case it makes more sense not to give up the solutions that have already proven worthy, not to destroy everything and then build from scratch, but to maintain continuity and introduce improvements gradually. But if they decide to launch a new Intel DX58SO mainboard revision or a totally new mainboard for LGA1156 processors tomorrow, even if it has the same inconvenient BIOS, I will prefer it to any other solution, unless the competitors can offer an adequate alternative. And all that because Intel DX58SO mainboard convinced us: from now on you can actually overclock on Intel mainboards.