** Official ** Gigabyte UD3 P45 Series -- EP45-UD3 / EP45-UD3R / EP45-UD3P

[Elektrosadist]

I have this Board now for 2 Weeks and tested it (imo) very extense during this time with lots of settings. I really enjoyed this board so i thought i write my expression down.

My System: Q9550, UD3P, OCZ Flex 9200 2048mb, GTX 280 702-1404-1161MHz, Vista 32 rig

10.20.2008:
Dang, 11:34 i jump into my car to drive 100 km to get my new CPU. While i was in the shop and chatted with the chief 0C there he showed me that UD3P and said that this board has a awesome performance, not to mention the price. I originaly thought about a Asus MF Formula II but with the performance in mind i thought what the heck and i took the board with me. Wnet home and downloaded bios f5a which i used then for a while.

System runs @ 3GHz with 1,1v. Poor Latency but awesome low Power Consumption.

System now runs @ 3,4GHz with 1,19v. Average Latency but stable with DDR2-1200MHz. Read: 8923 Write: 8350 Copy: 8350 Lateny: 56,1ns. First thing i notice: The Bios scales very good the gtl ref values. At the start of your 0C just set all voltages to the normal rating. Fe. if you have the vtt at 1,2 and the CPU and NB GTL refs at 0,76v you will notice that with every step you increase the vtt he will automaticly adjust the GTL refs with the 0,635 multiplier. So my estimated 0,889v for 1,4v vtt was automaticly set to 0,886v ref value. So for 0C starters this board helps you a lot to keep a Quad stable. The board is also able to run high FSB with low voltage on the mch.

System is becoming unstable at FSB 450. After some advice i switch back to Bios F4.

System now runs @ 4GHz with 1,344v. Lower Core Voltages @ FSB 471*8,5. Latency is good the system still is able to run a tRD of 7 while at DDR2-1178. Read: 9895 Write: 9857 Copy: 9248 Lateny: 56,0ns. All Benchmarks and stress tests are passed without errors.

System now runs @ 4,25 GHz with 1,383v = FSB 500*8,5. System is getting unstable. Even giving lots of vcore and vtt won't manage to keep the system under control. So we go down one half multi and loosen the core voltage a little to 1,36v with 1,38vtt and 0,881v for the GTL ref. MCH core voltage is set to 1,3v. The system now seems stable all benchmarks are passed and stress test showed no errors. So i switch to a memory multi which will put more stress on the mch controller and the system itself. We're now at 4GHz @ FSB 500 with DDR2 1250 @ 6-6-6-18 and a tRD of 8. This leads to astonishing Read: 10182 Write: 10528 Copy: 9966 Lateny: 51,3ns. Pretty darn good for a Quad on a P45 imo.

After having the system for a few days at the mentioned settings i thought i push it a lil more. I gave only 0,01v more core and upped the FSB to 510*8 = 4,08GHz with DDR2 1275 MHz (6-6-6-18 tRD:8). Ok now i get really close to what i have been looking for: Read: 10321 Write: 10747 Copy: 10006 Lateny: 50,4ns. But the system is getting unable to pass Prime 95. So i up the core voltage to 1,392v. This does the magic. Everything works now, Benchmarks and Stress tests can be all passed again. 3DMark 2006 produces 20019 points.




Maybe i can do a little more. I up the FSB to 515*8 and i set the vcore to 1,41v and a vtt of 1,6v. But the system now really becomes totally messy (prolly the tRD thats to low or the rams speed to high). I quit 0C for now and fall back to 500*8 at 1,3442v, vtt 1,36v, mch 1,3v and DDR2 1250. And thats a pretty stable setting. Passed various 6h prime tests as well as OCCT and the intelburntest. 3D Applications all work fine.

I will try later on to get to a super stable setting above 510 FSB but now i have no time. But i can say this board rox. The mosfets keep really cool compared to the Asus Boards i had until now but on the negative side i must say if you have this board running near its limits you hear the mosfets whistling. Well besides that i have sumed it all up. This board along with its default bios pumps the out of 45nm Quads if your ram works along with it.