02/13/2009 | 12:07 PM
In November 2008 Intel started selling processors and chipsets for their new LGA1366 platform. WE have already posted a number of articles discussing the new systems and their peculiarities. This is definitely the article called First Look at Nehalem Microarchitecture with the theoretical discussion of the innovations in the new platform; then the article called New Hit from Remake King: Intel Core i7 Review that is devoted to the practical aspects of the new platform. At last, the article called Intel Core i7-920 Overclocking Guide tells you not only about the succession of actions during successful system overclocking, but also reveals the potential of its components in specific experiments. In particular, the article talks about ASUS P6T Deluxe mainboard. I strongly recommend to check out these materials if you haven’t done it yet, because there is a lot of useful info in them. Anyway, I will refer to these articles multiple times throughout our today’s review.
Our today’s article continues a series of reviews of the new platforms: we are going to talk about two new mainboards from Gigabyte: GA-EX58-UD5 and GA-EX58-Extreme. It makes perfect sense to combine these two mainboards within a single article because they have a lot of common traits and very few insignificant differences. This way we can avoid repeating the same things about them and at the same time focus on the distinguishing features of each of them. One of the most noticeable differences between them is their packaging.
When I worked on this review, the junior model -GA-EX58-UD5 – cost a little over $300, which makes it one of the most affordable solutions for Core i7 processors.
Gigabyte GA-EX58-UD5 mainboard ships in a pretty standard sized box:
There is a general view of the board on the back of the box and a bunch of details on its features and functionality:
Gigabyte GA-EX58-Extreme ships in a much larger package along all three dimensions and features a convenient carry handle:
The decorative front flip-cover opens to reveal the mainboard inside clear plastic casing. The double-page spread as well as the back of the box carries the detailed description of numerous advantages, peculiarities and features of the top model in Gigabyte’s Intel X58 Express based family.
However, the accessories bundle is almost the same by both mainboards and includes the following components:
According to the manuals, there should also be an additional rear panel bracket with two IEEE1394 ports, but we didn’t find any in the box. Maybe there wasn’t any because we got these boards directly from Gigabyte, and not from the retail channel that is why their accessories bundle may be a little different from the typical retail bundle. However, it could also be a mistake in the manuals and there shouldn’t be any bracket like that to begin with, especially since there is one IEEE1394 port on the mainboard connector panel already and the manuals list this bracket among the default as well as optional accessories.
Since we mentioned one small issue already, I would also like to add right away that I was very upset to discover that all versions of Gigabyte EasyTune6 utility including the one posted on Gigabyte’s web-site this year didn’t work on both mainboards. We got the following error message right after program launch:
The problem is in X58 based mainboards, the same exact utility worked just fine when we checked it out on Gigabyte GA-EP45T-Extreme under Microsoft Windows 7 Beta.
Another frustrating issue deals with the BIOS updating. Theoretically, you can use the BIOS version from the DVD disc with the drivers and software bundled with the boards. In reality, you can’t really do it, because the DVD disc version 1.0 for X58 based mainboards contains BIOS F9 for Gigabyte GA-EP45-DQ6.
I have to say that issues and mistakes like that immediately make you suspect that they were rushing the mainboard launch too much and didn’t do a thorough enough job preparing for it.
Gigabyte web-site has a very convenient tool for comparing mainboard features; however, it doesn’t reveal any functional differences between GA-EX58-UD5 and GA-EX58-Extreme mainboards. The only difference seems to be the cooling system:
They used only basic features during this comparison, but even if take a close look at the detailed specs of Gigabyte GA-EX58-UD5 and compare them with the detailed spec of GA-EX58-Extreme, you won’t find any differences. There is actually no secret here: the boards have identical base, as you can see from the photographs:
Moreover, the difference in the mainboards’ cooling systems that we have already mentioned before, is actually limited to the heatsink on the chipset North Bridge, because both mainboards use the same heatsinks on their chipset South Bridges and processor voltage regulator circuitry transistors. As for the North Bridge heatsinks, the difference between the two boards is in fact very significant, and we should dwell on each of them individually. Let’s start with Gigabyte GA-EX58-UD5.
The chipset South Bridge is topped with a very large heatsink. Moreover, it is part of a single cooling unit also covering the chipset North Bridge with a heatpipe running beneath the connecting segment for better heat transfer.
As for the chipset North Bridge heatsink, it is pretty ordinary. Its design is not idea from the cooling efficiency standpoint, but is quite traditional for the latest Gigabyte mainboards. With a heatsink like that it would be best to use a CPU cooler directing airflow down towards the mainboard PCB, so that it could go through the North Bridge heatsink.
There are two heatpipes coming out of the center of the heatsink. One of them ends in the first heatsink on the voltage regulator circuitry transistors. About half of the fins on this heatsink are of larger size and form an array on the mainboard back panel. Another heatpipe goes through the first heatsink and ends in the second one.
The cooling system of Gigabyte GA-EX58-Extreme mainboard is if more sophisticated design. The heatsinks on the processor voltage regulator circuitry transistors and chipset South Bridge are the same as on Gigabyte GA-EX58-UD5 mainboard. The North Bridge, however, has a complex hybrid cooler that can also become part of a liquid-cooling solution.
The water block on the chipset North Bridge can be included into a liquid-cooling contour:
The plate that serves as a water block base has special seats and retention holes for an additional heatsink.
We have already seen something like that when we reviewed Gigabyte GA-EP45T-Extreme mainboard. However, the additional heatsink used on this mainboard is even more complex. It consists of two arrays: the first one is connected to the chipset North Bridge heatsink with two heatpipes, while the second one is combined with the first one by another two heatpipes and is located on the mainboard back panel:
At this point we will not yet discuss the efficiency of these modifications made to the additional heatsink. Let’s just take a look at what it looks like when put together:
And this is the view from the other side:
The additional heatsink takes away one of the PCI Express x1 slots. Moreover, the pin-connector for the audio ports appears right beneath this heatsink, so it is pretty hard to reach. These seem like rather insignificant losses, don’t they? But in return you get an exclusively unique looking system that will please your eye on a daily basis and surprise your colleagues and friends. But that’s about it. This extremely interesting and, obviously, expensive solution has no practical value of any kind.
I will not deny the data provided by Gigabyte themselves stating that this system offers 30% more efficient cooling. We will not try to find out against what system and in what conditions they compared their solution. Instead, let’s try to understand why we might need a cooling system like that in the first place. Of course, it serves to ensure more efficient cooling of the chipset North Bridge. And when do we usually need better North Bridge cooling? Of course, when we overclock processors by raising the voltages. However, here is the tricky part: we needed to increase the chipset North Bridge voltage to ensure stability during overclocking of the previous two CPU generations. Successful overclocking of the new Intel Core i7 CPUs doesn’t require any NB voltage increase and it can stay at the low nominal setting of 1.1V. In other words, any cooling system will do just fine here, even a small heatsink like the base one on Gigabyte GA-EX58-Extreme. I checked.
I could come up with another use for this advanced chipset cooler. Any other mainboard employing additional PCI Express bus controllers could use this add-on heatsink. These controllers run very hot and are usually included into the chipset cooling system, because they heat it up seriously. However, I have never seen any heatsinks like that on the boards that could really use them. Gigabyte GA-EX58-Extreme is not one of them, because it uses no additional PCI Express bus controllers. So, what does Gigabyte GA-EX58-Extreme need an enhanced chipset North Bridge cooling system?
One of the possible explanations here is that the developers were just doing their job by inertia. They have first come up with a system like that not that long ago when they designed GA-EP45T-Extreme mainboard. Of course, the engineers continued their work trying to make the cooler even more efficient. How could they have known beforehand that it wouldn’t be necessary anymore? Another possible explanation is the reckoning of the marketing team betting on the users’ psychology. Every customer understands that a mainboard with an advanced chipset cooling system is better than a board with an ordinary one. It is evident to each and every one, isn’t it? The truth, however, is available only to those of you who have read our article called “Intel Core i7-920 Overclocking Guide”, but, I guess, it is too late anyway: you have already purchased Gigabyte GA-EX58-Extreme, although you could have gone with a much more affordable Gigabyte GA-EX58-UD5 that is just as good.
That is why we really do not care how efficient the hybrid cooler on Gigabyte GA-EX58-Extreme actually is. It can be 30% improvement as Gigabyte claims, or 50%, or maybe even 100%. All this could be really great, if there had been real need for that. So far, the two boards are absolutely identical from the functional prospective, so let’s continue with our review.
Both mainboards use 12-phase voltage regulator circuitry for the CPU, and two-phase voltage regulator circuitry for the DDR3 memory and chipset North Bridge. Due to DES Advanced power-saving technology, the number of active phases may vary depending on the immediate workload. The LED rows indicate how many phases are active. Numerous LEDs tell us not only about the number of active voltage regulator phases in the processor, memory and chipset North Bridge circuitries, but also about the current voltage on the CPU, memory, chipset North and South Bridges as well as about the processor and North Bridge temperature increase past 60°C. It is really hard to figure out what message the LED’s deliver, because there are too many of them. However, this multi-color glow makes the board look very festive and fun. Only a row of bright-blue LED lights reporting high bus frequency during overclocking may be a little too annoying.
There is enough room around the processor socket to accommodate large CPU coolers. At least, we could easily fit a pretty large Cooler Master GeminII in there. The power connectors are also very conveniently located. It is a little uncommon to have Power On and Reset buttons so high up in the upper right corner of the PCB.
As we got to the description of the lower part of the mainboards PCB we first of all have to dwell on the graphics cards operational modes. Even if we install two graphics cards into two upper PCI Express x16 slots, they will work at their full speed. But as soon as we put in the third graphics card, the work mode changes to “x16-x8-x8” – two lower slots will have to share 16 PCI Express lanes. Moreover, there is also one PCI Express x4 slot, one PCI Express x1 and two PCI slots. It is interesting that PCI Express x4 slot doesn’t have the back side of the slot, so you can actually install a PCI Express x8 or even PCI Express x16 card.
There are six main Serial ATA ports and four additional ports along the right side of the boards PCB. The main ports are implemented in the Intel ICH10R South Bridge. As for the additional ones, two of them are supported by Gigabyte SATA2 controller (it must be the well-familiar JMicron JMB363 chip) that also provides PATA support. Another two JMicron JMB322 controllers help split these two ports into another two providing us a total of four additional Serial ATA ports. There is a two-digit POST code indicator right next to them, which is a pretty rare feature on Gigabyte mainboards.
The mainboards back panels have the following connectors and ports:
Overall, the design and functionality of both mainboards make a very good impression. Of course, we can feel that there is barely any free space left, but this is only because the developers did their best to provide their products with maximum functionality. You can get a better idea of the Gigabyte GA-EX58-UD5 and GA-EX58-Extreme mainboards layout from these schemes. As we have expected, these layouts are almost identical:
While we still can find a few exterior differences between Gigabyte GA-EX58-UD5 and GA-EX58-Extreme mainboards, such as different chipset cooling system and different model names, the BIOSes of both boards are completely identical. Only by pressing F9 an info window will pop up telling you the mainboard model name. Everything else in the BIOS Setup, such as sections and their contents are exactly the same:
Moreover, many BIOS Setup sections boast the same looks and functionality as the same sections in the BIOS of other contemporary Gigabyte mainboards. Namely, in our Gigabyte GA-EP45T-Extreme mainboard review we have already discussed in great detail almost all the functions. So, today we are going to dwell only on the parts that are different and that of utmost importance. We would like to start with the first and most interesting section called “MB Intelligent Tweaker (M.I.T.)”, which contains almost all the settings that matter to overclocking fans and computer enthusiasts.
To show all the settings available in the “MB Intelligent Tweaker (M.I.T.)” section we had to put together an image out of three screenshots. But even in this case we don’t see all the available options, because there is the whole bunch of parameters that are hidden in additional sub-sections. The section is nevertheless very convenient to work with and has a lot of info, although we could point out a few changes for the better as well as for the worse.
As for the negative change, I have to mention the absence of any sort of parameters grouping. All settings used to be arranged into groups referring to the CPU, chipset, memory and voltages. These groups were also provided with small highlighted sub-titles for additional convenience. No there is no structuring of any kind and different parameters just follow one another. It is not a dramatic drawback, but I think it has definitely affected the ease of perception and use.
As for the positive changes, I certainly have to mention that there appeared a section called “Advanced CPU Features”. The extended options dealing with processor technologies used to be part of the “Advanced BIOS Features” section, while it would definitely be much more convenient to move them over into the “MB Intelligent Tweaker (M.I.T.)” section where all other important system configuration options are. It was practically the only request we had related to this section and it is very satisfying to see that our wish has been granted. Now you can adjust all important system parameters without leaving the “MB Intelligent Tweaker (M.I.T.)” section.
However, this is where you can notice that some of the parameters get duplicated. The first one is “CPU Clock Ratio” that allows changing the processor clock frequency multiplier, and the informational “CPU Frequency” parameter. We have already seen these exact options in the main window of the “MB Intelligent Tweaker (M.I.T.)” section. So why have they been duplicated? We observe the same parameter duplication in a few other sub-sections. It is not a very serious drawback and some users may even disagree with me; however, I believe that duplications like that make things a bit more complex. All duplicated parameters are interconnected, i.e. no matter where you make the change, the parameter will get a new value everywhere.
In the “UnCore & QPI Features” section we again see two options that we have just come across in the main section window:
“Advanced Clock Control” sub-section contains not only the chipset fine tuning options, but also duplicates the parameter changing the base clock frequency. Here we can also see the options for adjustment of the PCI Express bus frequency and control of the “C.I.A.2” automatic CPU overclocking function.
As for the extended options for memory configuration, we can’t really figure out what the informational “Profile DDR Voltage” and “Profile QPI Voltage” parameters stand for. Their values don’t change even if you change the corresponding voltage settings; however, both these settings are available in the main section window as well as in the “Advanced DRAM Features” sub-section. What for?
“Advanced Voltage Control” sub-section also contains a few duplicated settings. However, most importantly, it contains everything we might need for successful overclocking:
Next we are going to skip the features of the following familiar sections, because they haven’t changed: “Standard CMOS Features”, “Advanced BIOS Features”, “Integrated Peripherals” and “Power Management Setup”. As we have already mentioned, you can check out our Gigabyte GA-EP45T-Extreme mainboard review for more details on these sections.
Now let’s take a closer look at the “PC Health Status” section that seems to have barely changed. The list of controlled voltages is incomplete. WE know the rotation speed of only four fans out of five that can be connected to this board. The distinguishing feature of Gigabyte mainboards is their ability to adjust the rotation speed of fans with three-pin connectors.
Now we just need to check out the boards’ ability to save BIOS settings profiles. They used to be pretty good at it, and now they also allow saving profiles on external media, such as an external hard drive, floppy or USB drive.
The most evident advantage is that you won’t need to reenter all the carefully picked settings after a BIOS update; you can load them from a file at any time. Moreover, you can exchange settings profiles. If the BISO versions are different, you will receive a warning that you can ignore. However, you won’t be able to load the settings profile created for a different mainboard model.
So, despite a few drawbacks we have just pointed out, we are happy to conclude that Gigabyte GA-EX58-UD5 and GA-EX58-Extreme mainboards feature all fine tuning options that you need to maximally optimize the system performance or overclock. Now let’s see how good the boards will prove in some practical experiments.
Let me briefly remind you what Turbo Boost Technology actually is. All existing Core i7 processors fall within the predefined 130W thermal envelope even under maximum load, when all cores are fully utilized. Of course, when only one core is working, the power consumption will be way lower. However, this core will still work at the same frequency as the other three if the Turbo-mode is disabled. The picture below shows an example of a top Intel Core i7-965 Extreme Edition processor with the 3.2GHz default frequency:
What happens when we enabled Turbo Boost? The Power Control Unit (PCU) integrated into the processors evaluates the situation. If the power consumption and temperature are within the standards, it increases the core frequency to the next step or two steps up. The CPU remains within the 130W TDP, since some cores are idle, the temperature is normal, but the performance increases compared to the same CPU working without the Turbo Boost.
In fact, I believe that Turbo Boost is probably the most important innovation since AMD announced their Cool’n’Quiet processor power-saving technology. Quad-core processors have been selling for a while now, so why don’t we all have quad-core CPUs in our systems yet? Because not everyone needs them. We have to decide if we want to get a quad-core CPU or a dual-core CPU working at higher clock speed. There are quite a few tasks out there that can be split into parallel threads very effectively and a multi-core processor will definitely be faster in them. However, despite this fact, it is still better to have only one or two cores working at higher frequency. Turbo Boost eliminates the need to make this difficult “either this, or that” choice, because it makes any CPU universal. The CPU may work as a multi-core processor employing all computational threads; however, it may as well work as a single-core CPU operating at higher clock speed.
It is important to keep in mind that Turbo Boost may cause some problems during overclocking. Imagine that after a series of experiments you have finally come up with the combination of parameters with which the CPU remains stable under maximum utilization of its al cores. When only one core is utilized, Turbo Boost kicks in and the frequency goes up, which may cause some failures. If the settings choice is based on the maximum frequency we can obtain with enabled Turbo Boost, then our processor cores will work at lower frequency than they could during maximum CPU utilization. Disabling Turbo Boost is also no good, because we will lose all the advantages it has to offer…
Theory is great, because it is all so logical and clear and we can take all the time we need trying to put together an optimal solution. After that it is time for a practical experiment and we end up choosing not what looks best from the theoretical standpoint, but what works best in real life. It turned out that disabling Turbo Boost on Gigabyte GA-EX58-UD5 and GA-EX58-Extreme mainboards is not the best thing to do. There are several reasons for that, and primarily it is the fact that on these mainboards Turbo Boost works not the way it is supposed to, according to the theory we have just discussed above. For example, I have never seen a CPU clock frequency multiplier being increased two steps up at a time, from 20x to 22x, even if there was a single-thread application running. No, one time during the memory tests Everest did report a 22x clock multiplier. However, since I couldn’t reproduce it again, I consider it to be a program glitch.
Moreover, I am almost 100% certain that mainboards increase the clock multiplier to 21x for all working processor cores, and not only in case one or two cores are utilized. I determined the core frequency by performance readings. Namely, I measured the speed with enabled Turbo Boost and multiplier increased to 21x, and then disabled it and increased the base clock speed far enough to obtain the same resulting CPU frequency with a 20x multiplier. I got almost the same performance, which couldn’t have happened if there had been only core working with a 21x multiplier with enabled Turbo Boost. Of course, the memory frequency in this case was different, so to be 100% certain one needs to use an Extreme Edition CPU with an unlocked multiplier.
In our Intel Core i7-920 Overclocking Guide we mentioned that ASUS P6T Deluxe mainboard allows “disabling the processor’s ability to control its power-related parameters without deactivating Turbo Boost. This trick makes it possible to statically increase the processor clock multiplier by 1 over the nominal independent of the workload and its current level of power consumption”. We got the impression that Gigabyte mainboards disable this control automatically once you enable Turbo Boost. CPUID Hardware Monitor utility that we often use to monitor temperatures and voltages proves this point. There is an additional “Powers” parameter there for Intel Core i7 processors and its maximum is always at 130W with disabled Turbo Boost.
The utility doesn’t measure the actual power consumption. The maximum value will remain unchanged even if we increase the CPU Vcore, which is impossible. However, once we enabled Turbo Boost, this value jumps to 150W:
Well, if the clock frequency multiplier increases by 1 for all cores simultaneously, it makes overclocking on Gigabyte mainboards a lot easier. First, with a higher multiplier, you don’t need to push the base frequency that far up in order to obtain the same processor frequency. It automatically means that the rest of the frequencies connected to the base one will not go too high either. Moreover, even if the one of the processor cores frequency will go two steps up at once during single-thread load, we shouldn’t be concerned about losing stability. Unfortunately, this implementation of the Turbo Boost technology ruins all its advantages. Our processor can no longer turn into a high-frequency single- or dual-core CPU: it will always remain a common multi-core processor. What a pity…
The next reason why you shouldn’t disable Turbo Boost mode on Gigabyte mainboards is because they will not set the correct processor voltage in this case. The nominal voltage of our Core i7-920 processor is by no means just idle talk. I assumed that it was 1.225V, but Gigabyte mainboards often showed 1.21875V. The difference is not dramatic in this case, but very often they read 1.18, 1.16 or even 1.13V. I thought that the boards detected the current processor core voltage during system boot-up, so I disabled processor power-saving technologies. However, the voltage readings still remained lowered.
However, it didn’t cause us any trouble. Once Windows booted the core voltage increased to 1.2V under maximum CPU load. It can be considered normal taking into account possible voltage drop during maximum CPU utilization and some monitoring errors. However, it only was like that with enabled Turbo Boost. Once we disabled it, the CPU Vcore read only 1.1V. Our preliminary tests showed that this low voltage was nevertheless sufficient for the CPU to work fine at nominal speeds. However, it was definitely not enough for any overclocking. And even if we manually increased the CPU core voltage preventing the mainboards from lowering it, we could always get better overclocking results with enabled Turbo Boost.
Our Intel Core i7-920 Overclocking Guide claimed that systems on Intel Core i7 processors are fairly easy to overclock. I seem to have a different opinion on this matter. First of all, I have to say that a completely new approach to overclocking and the use of a totally new platform did make things a little complicated in the beginning. Moreover, we usually do not know the potential of only one system components out of three (CPU, memory, mainboard). We either investigate the overclocking potential of a new processor on a well familiar mainboard, or try to find out how well a mainboard can overclock our familiar CPU. This time I had to solve an equation in all unknowns by trial and error. I am not very well familiar with the overclocking potential of the new Intel Core i7-920 as well as with the overclocking abilities of both new mainboards.
Even though we have used OCZ memory modules for testing mainboards with DDR3 SDRAM support for a while now, we had to check out their overclocking potential all over again. We know what they are capable of at 1.95V voltage. However, it is not recommended to push the voltage over 1.65V during Core i7 CPUs overclocking and I haven’t yet tested our memory in this mode. Besides, it turned out that the third OCZ DDR3-1800 Platinum Series memory module that I took for the triple-channel configuration was slightly different from the pair I already had. Its SDP reported higher frequencies and more aggressive timings, which you could see in the BIOS screenshots above, because the settings for each memory channel are adjusted independently. The mainboard set 7-7-7-20-1T timings for the first two models and 6-6-6-16-1T timings for the third one.
Let me briefly remind you how we overclock Core i7 processors. We increase their so-called “base frequency” of 133MHz. All the other frequencies increase with it, because they are calculated by applying a corresponding multiplier to the base frequency.
The processor core frequency is calculated using base frequency and clock multiplier, which is in our case 133x20=2.66GHz. The frequency of UnCore unit that includes L3 cache memory, voltage regulators, QPI and memory busses equals 213MHz (133x16). The frequency of the QPI bus between the CPU and the chipset North Bridge equals 4.8GHz (133x36). Default memory frequency is set at 1066MHz (133x8).
You should lower the corresponding multiplier in order to prevent any excessive increase in one of the above listed frequencies. However, you can in fact lower only the memory and UnCore frequency, because QPI bus already uses minimal multiplier possible. So, we will have to increase the voltage to ensure that the system remains stable at higher frequencies. Gigabyte did a great job preparing a manual on voltages and their role in overall system stability (the illustration below has been taken from this manual):
Click to enlarge
At first, this is exactly what I did: lowered memory and UnCore frequencies and increased the voltages. However, then it turned out that I didn’t have to lower UnCore frequency, because a lot of processor units will work at a lower frequency and it will affect the overall performance quite noticeably. The system, however, tolerates UnCore frequency increase during overclocking just fine. Then it turned out that I didn’t have to increase the voltages, either. I didn’t have to increase any of the voltages that could theoretically improve the mainboard stability, namely QPI, VTT, PLL or chipset NB voltage. Although all frequencies got higher during overclocking, the system worked perfectly fine with the default voltage settings. At least, it was the case up until the base frequency hit 200MHz. However, I have to point out two things about the voltages. The first one is evident: you do have to raise the processor core voltage during significant overclocking. The second one deals with the memory.
Many manufacturers have already announced memory modules designed specifically for work in LGA 1366 platforms in nominal and overclocked modes. Namely, they can work at high frequencies and almost nominal voltage of 1.5V for DDR3. Our modules are not like that. They require a serious voltage increase during overclocking, up to 1.95V, which is unacceptable for Core i7 CPUs. Therefore, we increased the memory voltage to the maximum possible value of 1.64V. But even in this case we had to use the lowest x6 memory frequency multiplier. As a result, the memory frequency was around 1100MHz during our overclocking experiments. It is very low for DDR3 in LGA775 systems, but turned out quite ok for LGA1366. The memory controller is integrated into the CPU, so you have to b careful when you increase the memory voltage. However, it means that the memory subsystem latencies in LGA1366 platforms are considerably lower than in LGA775 platforms. Besides, the wider 192bit memory bus has its positive input, too. Since we couldn’t increase the frequency, we managed to lower the timings to 6-6-6-18-1T and reach acceptable performance level.
So, we managed to overclock our Core i7-920 processor to 181MHz base frequency without increasing its Vcore over the nominal value. The multiplier was raised to 21x thanks to enabled Turbo Boost, so the resulting CPU frequency equaled 3.8GHz. I think it is a very good result, because the CPU could process all 8 threads simultaneously under maximum load despite the low core voltage. No other voltages were increased, except the memory voltage. The only other thing we had to do to ensure stability was to enable “Load-Line Calibration” that prevents the processor Vcore from dropping under maximum workload.
The good thing about CPU overclocking without increasing its Vcore is that all processor power-saving technologies remain intact. As the load drops, they lower the clock multiplier as well as the core voltage.
I personally almost always overclock processors at their nominal Vcore settings and with all processor power-saving technologies up and running. However, not everyone shared this approach to overclocking. For example, it doesn’t make much sense to fight for power-saving if the CPU is loaded with work 24/7. It turned out that by simply increasing the processor Vcore to 1.3V, we could push the base frequency to 188MHz and hence the resulting CPU frequency – to 3.95GHz.
In this case the processor clock frequency multiplier will still be lowered I idle mode, but its core voltage will remain increased.
Unfortunately, I couldn’t get my CPU to work stably at a beautiful frequency of 4.0GHz even with its Vcore increased to 1.4V. I didn’t dare check out higher voltage settings because of extremely high core temperatures.
By the way, I am very surprised that some people claim Intel Core i7 processors overclock and remain stable at 1.4-1.45V Vcore. I wonder what type of cooling is used in this case? For example, when we overclocked our Intel Core i7-920 processor to 181MHz base frequency even without any voltage adjustments, its temperature increased to 76°C in Prime95 (the room temperature was around 22-23°C). 76°C is a pretty serious temperature, although I did use a highly efficient Cooler Master GeminII cooler with a 120-mm fan at 2500RPM inside Antec Skeleton system case, which is even better than an open testbed because it has an additional huge 250-mm fan at the top. Overclocking to 188MHz base frequency and 1.3V core voltage pushed the CPU temperature to 83-84°C. So, how much higher can you actually go?
Besides, a 120-mm fan rotating at 2500RPM becomes very noisy. This is actually why I replaced it with a quieter fan from the OCZ PSU. You can handle this noise for the time of tests, but it is hardly acceptable for long-term work.
As a result, I think that the only Core i7-920 overclocking that has any real practical value for us, is with the nominal core voltage settings. In this case we get more or less acceptable thermal characteristics and power-efficiency at an acceptable noise level. We lose all that the moment we raise the CPU Vcore, while the frequency increases just a little bit, by 150MHz, from 3.8GHz to 3.95GHz. I will try to back up my conclusions with a few benchmark results in the next chapter of this article.
When it was time to measure the platforms performance, I had to decide on the competition for a more illustrative comparison. It didn’t make sense to compare against LGA775 platform, because it would definitely lose. The quad-core Intel Core 2 Quad Q9300 CPU could overclock to impressive490MHz FSB frequency on Gigabyte Intel P45 Express based mainboards. However, even in this case its resulting frequency would only be 3675MHz, which cannot compete against 3.8GHz and especially 3.95GHz by Intel Core i7-920. As for the performance comparison at the same clock speeds, our article called New Hit from Remake King: Intel Core i7 Review has already discussed it in great detail. WE have already talked about memory subsystem performance in different modes, revealed the influence of SMT, Turbo Boost and other innovations in the new CPUs on their performance. It also doesn’t make sense to compare Gigabyte GA-EX58-UD5 and GA-EX58-Extreme with one another, because their results are absolutely identical and the difference will lie within the measuring error. So, I decided to check out the performance gain from CPU overclocking.
Our test platform featured the following configuration:
We used Gigabyte GA-EX58-UD5 mainboard instead of GA-EX58-Extreme simple because it was in the system already at the time. I swapped the boards forth and back, reflashed new BIOS versions, replaced CPU coolers – however, the results were identical on both of them.
I ran the tests in three modes. First, in the nominal mode. A lot of users out there do not adjust the system settings to their most optimal values at all. They simply select the default BIOS profile and that’s it. So, we didn’t change voltages, frequencies or timings. Since Turbo Boost was enabled in this case, the clock multiplier increased to 21x and this mode is marked as 133x21 in the charts. In fact, Gigabyte mainboards raise the base frequency by almost 2MHz in this case thus increasing all other frequencies as well. For example, our CPU frequency was 2.83GHz instead of the nominal 2.66GHz, while the memory frequency rose to 1079MHz instead of the nominal 1066MHz.
The second mode was overclocking without increasing the processor Vcore and it is marked as 181x21. The CPU frequency was 3.8GHz and the memory frequency was almost unchanged – 1086MHz. We couldn’t increase the memory frequency, but we set more aggressive 6-6-6-18-1T timings. We also calculated the percentage of the frequency gain compared to the nominal mode.
The third mode was further overclocking to 188MHz base frequency, i.e. to 3.95GHz CPU clock. The processor Vcore was increased to 1.3V. Again, let’s check out the frequency gain compared to overclocking the base frequency to 181MHz.
Well, here are the results:
Frankly speaking, I didn’t expect a triumph like that at all. Overclocking the processor by 43% brought us an average 25% performance increase. The performance improved by more than ¼! Another 4% frequency increase results in 3.5% performance improvement. Of course, the specific values depend on the applications. The results in Crysis were especially impressive, and we didn’t even touch the graphics card yet, only overclocked the CPU and set more aggressive memory timings!
Now let’s check out the system power consumption in all three work modes. We used Extech Power Analyzer 380803. The device is connected before the system PSU, i.e. it measures the power consumption of the entire system without the monitor. We used Fritz Chess Benchmark to load the system and recorded the maximum reading in the end of the test.
And the results are actually very interesting! Overclocking without the voltage increase raises the power consumption by 17.5% on average. It makes perfect sense to refer to an average value here, because the CPU is rarely fully idle or fully loaded. It appears that overclocking improved the processor power-efficiency and performance-per-watt, because its power consumption increased only by 17.5%, while performance got 25% higher. However, we can’t say the same about overclocking with increased core voltage, because there is a squared and not linear relationship between the voltage increase and the processor power consumption. The power consumption got about 11% higher than in the previous case, while the performance improved only by 3.5%.
So, the advantages of overclocking without increasing the CPU Vcore are evident. The performance boost is more than evident, over 25%, while additional voltage increase will only add another 3.5% and the price will be way higher power consumption, heat dissipation and noise. However, if your CPU is 100% utilized all the time, even a few additional percents will help and the noise doesn’t really matter then go all the way. It is your choice.
The obtained results should be very convincing even to dedicated overclocking antagonists. Although, there is another opinion out there, too. Namely, that “normal people” will be able to make enough money to go and buy the part they need in the time it will take poor overclockers to get to the desired speeds. You should just keep in mind a few things. For example, the official wholesale price of the top Intel Core i7-965 Extreme Edition processor is $999. And as you know, this processor works only at 3.2GHz. We, however, managed to get a 3.8GHz processor without much effort. And a slight Vcore increase will push its speed almost to 4.0GHz. There are no CPUs like that in the market today, no money can buy them. Even if they appear one day, overclockers will already be conquering new heights by that time. So, you don’t have to overclock, but why not do it, if you could really use some high speeds for work or gaming?
Our today’s review was devoted to Gigabyte GA-EX58-UD5 and GA-EX58-Extreme mainboards. At this time we do not have any worthy competitors available, so we can’t say once and for all how good these boards actually are. However, there is one thing I can say at this point: certain flaws, such as wrong BIOS version on the bundled DVD disc for instance, do spoil the overall impression a bit. BIOS issues, such as suspicious Turbo Boost operation and lowering of the CPU core voltage setting with disabled Turbo Boost, and non-operational EasyTune6 utility do not let us fully enjoy the advantages of the new platform. However, the competitor solutions may have even bigger problems, so we won’t draw any conclusions just yet.
Those of you who usually read only the introduction and conclusion shouldn’t really worry too much about the previous paragraph. What I am trying to say, Gigabyte GA-EX58-UD5 and GA-EX58-Extreme are good boards, although not completely flawless. But is there a product that is? Our tests proved that these platforms work well not only in the nominal mode but also during overclocking. They are quite functional and you will hardly need to buy any additional controllers, because these boards support all contemporary interfaces and technologies, except the wireless. If we tried to choose between the two, I personally would go with Gigabyte GA-EX58-UD5: it is just as good as its elder counterpart, but it costs about $35 less. Moreover, GA-EX58-UD5 is currently one of the most affordable platforms for Intel Core i7 processors.