X58 Motherboard Roundup Reviewtacohunter52 -
» Discuss this article (6)
Testing Setup i7:
- Processor: Intel Core I7 920 150x20
- Memory: Mushkin 3x2GB DDR3 1200 MHz 7-7-7-20 1.65v
- Video Card: nVidia GTX260 Core 216
- Power Supply: Mushkin 800watt Modular Power supply
- Hard Drive: 1 x Seagate 1TB SATA
- Opticals: LG DVD-RW
- O/S: Windows Vista x64 Ultimate Edition SP1
- Processor: Intel Core i7 920 215x19 1.38 volts
- System Memory: Mushkin 3x 2GB 9-9-8-24 1700 MHz 1.62 volts
The P6T is the lesser featured brother of the P6T Deluxe OC and as such the expectations for how well it will overclock were not really as high as the higher end boards. With the reduction from a 16+2 power design to the 8+2 design on the P6T it did not look good . As it turns out, that thought could not have been further from the truth. What I found was that the P6T has got some game. The 920 that I am using is a retail version and not the ES that I looked at back in November and really was an untested commodity. After checking the stepping against a few overclocking databases it looked like it may well be an average chip. Average chip, lower end board, things did not look good. However, both the 920 and P6T put on a little show and came up big. The maximum baseclock I could achieve with stability was 222MHz, but this was pretty much bench stable. 220 was stable enough to be Prime stable at 220x18, or 3.96GHz, pretty sweet thus far. To get there I needed just 1.38 volts on the CPU, a QPI voltage of 1.39v, IOH of 1.36v, CPU PLL voltage of 1.86 with the memory at 1.62 volts. Not a whole heckuva lot to do to bust out that kind of speed. Of course, 3.9GHz is not enough and I wanted to see if the little 920 was indeed a bona fide 4GHz chip without big volts. To push higher, I dropped the base clock down to 215MHz from 220MHz with a multiplier of 19 to get to 4.1GHz. I was expecting a fight to get there and was surprised at the fact that all it took to get this clock speed stable was an increase in the CPU voltage to 1.395 volts with no other changes to the previous voltage settings. Now the one thing I found that you have to do though is keep air flowing over the X58 chip to keep it cool to maintain stability at the high base clock levels I was playing at. Other than that, nothing really special was needed to push the clock speeds on the P6T. I have got to say that the board delivered the overclocking goods with an untested CPU.
- Processor: Intel Core I7 920 185x20
- Sytem Memory: Mushkin XP3 12800 3x2GB 740MHz 9-8-7-20 1t
When overclocking, having a known commodity is helpful when working with a new piece of hardware. The Core i7 920 used in this review is capable of 4.1GHz Prime stable, so what better chip to use than one already capable of a 222 base clock in several other boards? Overclocking the ECS X58B-A started out pretty well with a boot at 200x20 but I could not gain any stability at 200, 195, 190MHz base clock speeds. Every board I have used has been capable of at least a 200MHz base clock. I went back and started from scratch and inched up 5MHz at a time to see just where the cut off point for stability was. I kept the memory divider low to eliminate that as a concern and concentrated strictly on the CPU to see just where the top base clock would be. After playing with the QPI and Vcore voltages I was able to reach 185MHz using a 20x multiplier. One more MHz resulted in Prime 95 failing in less than a minute; 186 = fail in one minute, 185 = 24 hour prime stability. One thing I noticed while trying for 4GHz with the ECS X58B-A was instability of the Vcore. While trying for a higher clock speed I could set the voltages to where they were required to get this chip to perform on other boards; 1.41 on the CPU and 1.4 on the uncore voltage and the voltage droop from the set voltage in BIOS versus the voltage read in Windows were under load were startlingly different. The CPU voltage would droop as low as 1.35 volts, causing the CPU to fail at speeds it has been proven to run at. If anything, I feel that this point alone causes the board to lose tremendous overclocking potential. My final clock speed on the ECS is 3.7GHz reached by running a combination of base clock and multiplier of 185x20. Not bad but not great either as there was 400MHz worth of lost potential left on the table. When you get down to it though, a 1.1GHz overclock is nothing to sneeze at.
eVGA X58 3X SLI:
- Processor: Intel Core i7 920 206x18 1.40 volts
- System Memory: Mushkin 3x 2GB 9-9-9-24 1648 MHz 1.65 volts
The i7 runs at 2.66GHz and only needs around 1.10v to work at this frequency, a pretty impressive feat. Getting it up to 3GHz was simple enough, all it took was a bump to the base clock from 133 to 150MHz. Then things started to get a bit complicated and required quite a bit of tweaking. Since I am using a processor with a locked multiplier, I started by decreasing it in order to find the highest stable base clock. The highest I could hit with full stability is 215MHz, which required the QPI voltage to be set at 1.55V. The eVGA board had trouble dealing with failed overclocks, it would usually not even get to the BIOS and I had to clear the CMOS every time. The fact that the board takes at least ten seconds from power on to POST is quite annoying as well, especially when you are overclocking.
Now that I knew how high the board could clock and already knowing the limits of the RAM, I set off to find out how high the eVGA X58 could drive the i7 920. Properly tweaking voltages was an absolute requirement to make the overclock stable. After having worked on it for a while, I finally settled for a 206MHz BCLK with the processor's multiplier set at 18, which translates into 3.711GHz. The memory ran at 1648MHz 9-9-9-24 at 1.65V. With voltages above stock settings, it's important to make sure the voltage regulators have adequate airflow, otherwise the passive heatsink will let them heat up to 100 Celsius. Of course, you will also need proper cooling on the processor since i7s tend to get toasty.
For those of you who are wondering why I chose to go for a lower multiplier and a higher base clock instead of using the stock multiplier and lower BCLK, I have a simple explanation. Even though I could squeeze out another 50MHz out of the processor using 19 and 20 multipliers, memory was forced to run about 100MHz slower due to dividers, so it just wasn't worth it. Running a higher BCLK also made the QPI link faster, which helps a bit with performance. Finally, even though I settled for a measly 3.7GHz, it's possible to reach over 4GHz, but the i7 920 required far too high voltages to make it a realistic overclock that you would run every day.
- Processor: Intel Core i7 920 200x17 1.25 volts
- System Memory: Mushkin 3x 2GB 9-9-8-24 1600 MHz 1.6 volts
Overclocking the Gigabyte board to 200 BCLK was very easy. Leaving everything on auto except for CPU and Memory voltage allowed the system to operate stably. Using a custom cooler at the last minute, I was able to raise the CPU multiplier and bring it to 4GHz, but the CPU was not fully stable. 3.8GHz was however, and thus this board can take users far if the cooler and chip are up to par. A few minor tweaks and it was running smoothly. Anything further required a little more tweaking, but for 200 BCLK and passive northbridge cooling there is nothing to whine about here.
MSI Eclipse Plus:
- Processor: Intel Core i7 920 200x20 1.45 volts
- System Memory: Mushkin 3x 2GB 9-9-9-24 1600 MHz 1.6 volts
Overclocking to 200 base clock on the MSI Eclipse Plus was fairly easy, although it did take some tweaks in the BIOS. Anything over 200 got difficult but there is still headroom. A fan is definitely necessary to keep the temperatures down, although the new thermal paste included stock was better than the older purple stuff on the MSI X58 Platinum. Using the OC Dial I was able to get a suicide run up to around 4.6GHz, while the system was tested at 4GHz. In the BIOS the extra multiplier for Turbo Mode can be forced to be enabled and it worked fine for pushing the overclock to the maximum.
MSI X58 Platinum:
- Processor: Intel Core I7 920 202x18 1.4125 volts
- Sytem Memory: Crucial 3x 2GB 7-7-7-20 606 MHz 1.6 volts
The i7 runs at 2.66GHz and only needs around 1v to work at this frequency, a pretty impressive feat. Getting it up to 3GHz was simple enough, all it took was to bump the base clock up from 133 to 150MHz. Then things started to get a bit complicated and required some, actually a lot, of trial and error. Since I am using a processor with a locked multiplier I started by decreasing it and finding the highest stable BCLK. The highest I could hit was 215MHz, although it was only stable up to 204MHz. The MSI board dealt reasonably well with failed overclocks, it would turn on for about ten seconds and then shutdown and so on until I manually shut it down and turned it back on myself. Never did I have to make use of that nifty clear CMOS button on the back.
Now that I knew how high the board could clock and already knowing the limits of the RAM, I set off to find out how high the MSI X58 Platinum could drive the i7 920. Tweaking core and QPI voltage properly was absolutely necessary to make the overclock stable. I needed +0.18V on the processor and another +0.15 on the QPI link, any more than that and temperatures got out of control, but it was enough to get a not too shabby 3.646GHz overclock. The memory modules were kept at 1212MHz 7-7-7-20 and the other voltages were left on the Auto setting.
Now that we've seen how well each board overclocks, let's take a look at how they perform.