Test Setup and Testing Methodology
Gigabyte NL was so kind to lend Madshrimps a review sample of their EP45T-UD3LR. This board is a cheaper variant off the UD3P series. Yet they are both based on the same 2oz copper design and are part of the Ultra Durable 3 family. Major features lacking are the 2nd PCI-E lane (so no Crossfire here), a 2nd LAN port, 4phase power, 4pin vs 8 pin power connector, no heatsinks at the PWM's and the bios is a more simplified version.
Here are some pictures of the board :
4phase power design (note the lacking heatsinks) , a very clean socket layout and the 4 pin power connector up left. Bulky NB cooler, no heatpipes on this board, making it far easier to upgrade the cooling. OCZ PC12800 sticks ready for actionPlenty of USB ports, 6 channel audio and still 2 PS2 ports for keyboard and mouse.6 S-ATA ports at your disposal, IDE and Floppy connector
Nice feature on this board is the double bios (should really become a standard these days) You can clearly see the single PCI-E 16X, 3 PCI 1X and 2 PCI slots. The board has proven to remain rock stable at 500FSB for days without a hiccup.
Since most Intel P45 based chipsets are having issues with some Micron based rams I knew I was up for quite a challenge. With the previous Battle of the caches
article I used the EP45T-UD3P motherboard in combination with I7 Low voltage rams. To obtain 1600Mhz ram speed was absolutely no problem then. With the Micron rams now I ran into a wall at 1590Mhz. 495FSB was a no post, no matter what I tried. Popped in the I7 kit and bam she posted at 500FSB and the rams at 1600mhz. Sadly I urgently needed that ram in another rig and decided to continue testing with the 2Gb Micron kit. Bios F5 was the latest available bios and was used during these tests. So I finally decided to back down on the ram speed and wanted to see how low I could go on the Performance level setting with this budget board. Too bad that there doesn't seem to be a workaround here for this long known issue.
A golden tip : if you intend to buy a P45 DDR3 based board, it's best to go straight for low voltage I5/I7 rams.
For the 266FSB tests, I was running both CPU's (E6300 and E7400) at their rated 2800mhz ( 10.5X ) stock speeds. Using the rams at 1066Mhz with 5-5-5-15 2T timings and a PL of 5.
Since I opted to re-include the E8600, we then tested this trio at the stock clock speed of Intel's fastest Dual core CPU. Selecting the 10X multiplier with a FSB of 333Mhz. This made the ram run at 1333Mhz with 6-6-6-20 2T timings and a PL of 7.
To make things more interesting we then opted to test the CPU's at 4ghz (very common clock speed under gamers and light overclockers) with 2 different Front Side Busses. For 400FSB and 10X, the rams at 1280mhz with 6-6-6-18 2T timings and PL 8 set. 500FSB made the rams run at 1200Mhz with 5-5-5-15 2T with the same PL setting of 8. Will be interesting to see if the higher FSB, but lower ram clocks with tighter timings, can improve overall system performance.
Here's a quick sum up chart :
Leeghoofd's E6300/E7400/E8600 Test Setup
|CPU||E6300/E7400/E8600 clock speeds @ 2.8, 3.33 and 4Ghz|
|Cooling||Thermalright eXtreme + 1600rpm Fan|
|Mainboard||Gigabyte EP45T UD3LR F5 Bios|
|Memory||2Gb OCZ PC12800 C7-7-7-24 2T|
|OS||Windows XP with Service Pack 3|
|PSU|| OCZ ModStream 700W|
|Video Card||Geforce GTX 285 190.38 Forceware|
Benchmarks and applications tested :Futuremark
3Dmark2001 and 2006Futuremark
Superpi v1.5 1mb and 32Mb test
Wprime v1.55 32 and 1024 testLavalys
Everest Cache and Mem benchmark
Cinebench R10 single and multi CPU testWinrar
500Mb test via integrated benchmark
Game tests run at 1280 x 1024 resolution with the integrated benchmark tool, High detail settings selectedFarcry 2Tom Clancy's HawXLeft 4 Dead
We took the average FPS from the game tests. Maximum FPS means nothing if your minimum FPS drops too much, making a game unplayable. Since I'm used to playing First person shooters my own personal limit is 60 FPS. If I can't keep that steady I'm lowering detail levels.
Let's move on to the results