Gigabyte EP45-DS3R P45 Motherboard Review

Overclocking Results

Well that is about enough admiring the looks of this board, now let’s get to what you have been waiting for: how the Gigabyte EP45-DS3R overclocks!

Starting right from initial setup of the board a slight issue arose. Upon installing all of the hardware, we shorted the power switch pins and the board powered up but did not POST. After some investigation it was determined that the default BIOS settings were not set up to allow our Ballistix PC8500 to POST with two sticks installed. Removing one stick of memory quickly solved this problem although it did get quite annoying to remove one stick of memory after every CMOS reset. Other than that, the board worked great at stock.


Moving on to the actual overclocking of the EP45-DS3R, we must say it was much much easier than some P45 boards we have previously tested. Auto settings are tuned well and the BIOS has all of the options we could ask for. With little effort and most auto settings for voltages we booted right up at 500FSB and an 8x multi for 4000Mhz. This was even stable for the duration of stress testing (as shown above). BUT, against our better judgement we allowed all of the stress tests to run even though the auto settings of the board set VTT and PLL voltages at very dangerous levels, 1.60v and 1.85v respectively. We should add here that it has been commonly shared and concluded in the PC industry that 1.4v is about the maximum safe 24/7 voltage for VTT. However, it has come to our attention that some motherboard manufacturers have changed this limit to closer to 1.9v VTT. We will ask Gigabyte for an official answer and discuss this in the comment thread. All hardware appears undamaged but unless otherwise advised, we highly recommend against using auto settings for PLL and VTT voltages.


After the close call with the high auto voltages, it was time to get down to business and set the board up manually. After extensive testing and tweaking we were able to achieve an overclock of 4203Mhz. This is about the max our Xeon 3110 will allow as above this much higher voltage is required for 24/7 stability and temperatures go through the roof. Due to the variation between BIOS set vcore and vdroop the vcore needed was 1.51875v (BIOS) giving between 1.43v and 1.44v under load. We has a bit of trouble with long-term stability at first, but after the correct combination if tRD, VTT, GTL, and PLL was found it came easy from there.

A point of contention with many P45 boards is the strap and memory divider settings and whether or not the settings work as they should. Well, we are happy to report we had no issues using the 333 and 400 straps and several dividers from each. Testing memory independently, the board was on par with other P45 boards. Using Crucial Ballistix PC8500 (the new single sided stuff), 500Mhz (DDR 1000) was achievable at 4-4-4 and 600Mhz (DDR1200) at 5-5-5 at 2.15 to 2.25v without much trouble at all. These findings are consistent with other P45 board tests so it seems the chipset is pretty good for clocking RAM in general.

Jumping straight to maximum FSB tests, the DS3R was decent. 500FSB was possible with safe voltages for VTT and PLL but to get any degree of long-term stability at higher FSB it required quite a bit higher voltages. Again, since there is some confusion on what VTT and PLL voltages are safe, we decided to play it safe for our CPU and run the board at 467FSB for daily use. We should add however, we were able to bench SuperPi without much trouble upwards of 550FSB.

All in all we have to conclude that the EP45-DS3R is a great clocker for its mid to low price point. It took some time to learn the board and determine appropriate voltages and GTLs, but after that it should become fairly easy for an experienced overclocker. A novice user should not have much trouble with the board either if they know P45 board overclocking basics and safe voltages. We do have concerns about the VTT and PLL voltages that are set if the settings are left on auto and FSB of 500+ is used, but this can be avoided by setting these voltages manually or booting under 500fsb. In the end we achieved a 40% overclock for 24/7 use, and only the CPU was limiting us from going further!