Nice article there. IMHO this is suppose to have a drastic effect on cooling but none the less it is still a great brand of mother. If only they could invent a cooling gel that could cover the resistors, transistors etc...


BTW this is my first post here in Tom's upon my visit here for a long time.... :-)

I bought a UD3P on eBay when MS was giving 30% cashback ($84.50). It's my first enthusiast board and I gotta say it runs great. It's pushing my Q8200 (I know, but it was only $119) to 3.3ghz (7x475).

IMHO thicker copper on the TOP and BOTTOM layer would have helped more that thicker inner layers. Also Exposed copper areas (stanium plated areas with no solder mask)would have been good for improoving copper to air heat transfer. Thicker inner layers can improve the overall heat distribution on the PCB which can be a plus for the lifetime of the board (uniform heating=less mechanical stress IMO).
But I don't know if it's realy OK to put thick copper on the routing layes of a high speed application....

"(...) and support, but you still need to be an enthusiast to appreciate most of these."

No need to be an enthusiast to experiment the lower than poor support from most motherboard manufacturers. I have a collection of pathetically pointless replies from GigaByte's "support" that is growing almost large enough to be turned into a book.

All p45 or other chip sets got similar performance. Most of new boards over clocks grate. I don't think 1% in a benchmark will be a decision maker for me in any direction.
Build quality, RMA rate, and good layout comes first for me. After i sort that out i look in to over clocking and futures.
After that was said, i am really happy with Giga-byte build quality, they defiantly come a long way. Time to few others to catchup.

arkadi:

P45 not p45, great not grate, I not i, Gigabyte not Giga-byte, definitely not defiantly, 'for a' not to, 'catch up' not catchup (a condiment made from tomatoes?)

"Gigabyte says that it uses 2 ounces instead of 1 ounce of copper for each of the inner layers. One ounce equals 28.35 g, which means that an Ultra Durable 3 motherboard comes with almost 57 g of additional copper for the two inner layers."

Actually, 1 ounce copper weight for a printed circuit board typically means 1 ounce of copper per square foot area. So depending on the size of the motherboard, 57g additional copper may be inaccurate. It is safer to say that Gigabyte doubled the amount of copper on the inner layers.

I'm sure if their replies are pointless, then your questions must be as well.
I've made a few help petitions to gigabyte by now, and I've always gotten a rather helpful reply. In fact some of the replies I got from them qualify to call them the best company I've received support from online.
When I'm asking them about something, it's usually the complicated stuff that isn't put in the manuals. And yet they manage to almost always actually answer my questions. I like their support. So far anyway.


As for the article - a very important thing is missing imo!
Overclocking. I don't mean this moderate oc of the quadcore to check the heat claims, but a real overclock. Using water cooling or peltier or something to see how far you can push the system before it becomes unstable. You can get past 500fsb on a regular P45 - can you get past 650 with a ds3? or just how far can it be pushed? That's what interests me.
I don't give a fuck if the vrm's are running at 60C or 65C or if the total power draw is 3W lower than comparable boards. I want to know if it can provide any more performance than the regular deal. At stock there won't be much of a difference between boards.

Now all they have to do is put the processor socket on the other side of the motherboard so the case can be used as a heat sink...

PS I would like my share of the profit from my idea.

patent a working prototype and sell it to gigabyte.

Nvidia and ATI should hire Asus, Gigabyte and MSI to help them design videocards; the reference designs are crap. Gigabyte managed to shrink 9800GTX+ card by 2 inches, utilize only one 6 pin connector and decrease power consumption by 10 and 20 watts at idle and load. This is on top of higher overclock ceiling.

n3m is correct about the amount of copper actually in the raw material used to make these circuit boards. The terms "1oz" or "2oz" refer to the amount of copper per square foot of circuit board area. What he/she didn't point out (nor does the article) is that nearly all that extra copper is etched away during the manufactuting process. It would be accurate to say that the amount of copper had doubled, assuming the circuits were otherwise identical, however the amount of extra copper is miniscule, IMHO and the net result of this advertising gimmick is expensive copper settling to the bottom of the etch tank. Lately, IMHO, Gigabyte seems to have a penchant for improvements that are more form than function.

I like what this company is doing. I dont think people realistically expect 50C lower temps, alot of it is hyped up marketing. This board offers great build quality which is very important to people like me. It has plenty of features and offers a great plaform for overclocking.

I've been a fan of the GA-EP45-UD3R an will be using that one in my next build.

Too much marketing over a single product. I feel like I have just sat through a Gigabyte Marketing Seminar. The article mentions lowering the impedance of the circuit with more copper. If you lower impedance you increase voltage. You can lower the voltage by adding resistance, which in turn causes heat and raises the impedance back up. It sounds like the "extra copper" was added to help distribute heat though, not to change the characteristics of the electrical circuits.

So, like, you say "and it may represent one of the best platforms for overclocking"...Any reason you didn't bother to test this? What's the point in testing "the best platforms for overclocking" if you don't test its crapacity to be overclocked?

This review is kinda pointless.

Good review, but where is the overclocking section? That is what these boards are best known for.

1 or 2 oz. Cu for PCB's means actually 35 microns (1.4 mils), respectively 70 microns (2.8 mils) thickness for the copper layer. The weight in the designation is the copper cladding amount on one square foot, as pointed out by n3m.

Actually, I rarely ask them questions, most of the time I write to report BIOS bugs.

From my latest contact with them informing of a bug in a new BIOS for one of my mobos:

(ME)
Under "Advanced Chipset Features/HT Link Control" (CTRL+F1), changing the HT Link Frequency from AUTO to 1 GHz will cause the BIOS to set it to 400 Mhz. Changing it to AUTO sets the value correctly at 1 GHz.

(GB)
1. The HT Link is the datapath that information uses on an AMD system, it's stock speed is 1Ghz, suggest you always to keep it around 1Ghz otherwise it gets unstable. As you overclock your system, your HT Link speed goes up along with your RAM FSB, so to keep the HT Link stable at higher speeds you lower the multiplier, on a 600Mhz overclock , run it at 4x (800) which works out to just over 1Ghz HT Link speed.The 1000=5x

(ME)
What part of the following did you not understand?

"Under "Advanced Chipset Features/HT Link Control" (CTRL+F1), changing the HT Link Frequency from AUTO to 1 GHz will cause the BIOS to set it to 400 Mhz. Changing it to AUTO sets the value correctly at 1 GHz."

If you change the setting to ONE (1) GHz in the F9 BIOS, WITHOUT any overclocking AND with Cool'N'Quiet disabled, the BIOS will set the HT Link to 400 Mhz (FOUR HUNDRED MEGAHERTZ). It ONLY stays at 1 GHz if the setting is kept at AUTO.

(GB)
In regards to your issue, we have duplicate this issue and passed it to RD for further confirmation this issue in F9 BIOS , we will update this issue as soon as possible.

* * *

And this is just a minor example... GB used to be my favorite brand after DFI, but support has gone downhill, period. I have two options for you:

1 - Fanboy.
2 - GigaByte employee.

I own both the UD3P and DS3R, both paired with a Q6600 quad core chip. The DS3R overclocks the chip easily to 3.0. With the UD3P, (the newer, and supposedly better board) I only got it to 2.7 MHZ. Anything above 2.7 and Prime95 would crash within 15 minutes. Now I don't know if this is because of a poor overclocking chip but I have my doubts about the UD3P. It sure looks nice though, with the blue heatsinks.

they lost me when I purchased a ga-965p-ds3 rev. 3.3 All the sites phraised this motherboard for having solid state capacitors etc. It was an overclocking marvel etc. What they failed to mention is the fact the motherboard once restarted goes to default settings. I contacted gigabyte and got the typical we are looking into the issue.

By the way I have been building with Gigabyte boards since 1998. I am going to give Asus a try again, they at least don't sell you something with a HUGE problem.