Motherboard Deconstructed – Cooler Analysis
CPU VRM Cooler
The upper portion of the VRM cooler consists of three parts. The upper portion is the heat dissipation portion of the heat sink with lots of surface area. The lower portion is the flat mating surface for contact with the VRM chips. Sandwiched in between the two is the heat pipe, sitting in a pre-formed channel. The parts are held together with several small screws that sit flush to the surface of the bottom of the heat sink. GIGABYTE uses a minimal amount of thermal paste on the heat pipe for better heat transfer, but not quite enough in my opinion.
The fan and water barb site in close proximity to one another in the upper VRM heat sink. The fan is partially protected by the plastic overlay near the top lower portion of the heat sink.
The heat pipe channel imbedded in to the pieces of the heat sink form a square channel for added support of the delicate heat pipe channel. This prevents kinking or crushing damage to the heat pipe since the heat sink parts are supporting the contact pressure between the heat sink and board. The height of the square channel also give the top portion of the heat sink enough clearance to overhang the chokes.
GIGABYTE used an interesting design in implementing the heat pipe for the VRM cooler. The water barbs are directly connected to the heat pipe, using the heat pipe copper to transfer the heat sink's heat directly to the water medium. However, if you are not using water, thermal transfer between the two heat sinks relies on air and the copper tubing itself to transfer the heat along the heat pipe. Without water, the integrated fan in the upper heat sink is the sole heat dissipater for the system. Most water systems rely on surface areas and disruptive channels to optimize heat transfer to the water. So while you do have a good amount of surface area along the tubes, the tube pass-through method is not the most effective means for water-based heat dissipation.
Similar to the upper heat sink, the side VRM heat sink is made up of three different parts. The upper and lower portions of the sink sandwich the heat pipe. At the end of the heat pipe is the other 3/8" water barb. The heat sink is held together with screws through the bottom plate of the heat sink that are flush with its surface.
The included fixed water barbs are nickel-plated copper, matching the material used in the heat pipe. The water barbs are ridged to hold the tubing in place, but require a plastic or worm-gear type hose clamps to fix the tubes in place. The barbs themselves accept 3/8 inch inner diameter tubing.
Chipset Cooler
The heat sink responsible for cooling the PCI-Express bridge chip is a one-part heat sink with the heat pipe soldered to the aluminum heat sink. The heat pipe and bottom of the heat sink were sanded and polished to make a consistent and flat surface for mating with the heat spreader on the bridge chip. While you can see polishing ridges in the heat sink's bottom, the bottom was smooth and flat.
The Z87 chipset heat sink is also a fixed heat sink with the heat pipe embedded in between the chipset interface layer and the upper aluminum block housing the integrated fan. GIGABYTE uses thermal tape to interface the chipset surface with the bottom of the heat sink. The black lower surface of the heat sink is flat, providing a good mating surface for the chipset die.
The integrated 40mm fan sits almost directly over the die portion of the heat sink, providing maximum cooling potential to the chipset. The heat pipe effective transfers heat from the bridge chip to this larger sink for the fan to dissipate its heat as well. Unfortunately, GIGABYTE did not integrate this cooling path into the VRM cooler, negating the more effective cooling to be offered by using the water cooling option for the VRM cooler. In testing, this cooler was found to get much hotter than the VRM cooling loop with both using fan-only dissipation and would have benefited greatly from integration with the CPU VRM cooling loop and its water cooled potential.
I bet those little fans will
I bet those little fans will go before the board…are they easily sourced ?
I prefer to just have big case fans turning slow as possible.
Does having “ultra durable”
Does having “ultra durable” being plastered on it actually mean anything ?
Ultra Durable is just the
Ultra Durable is just the GIGABYTE branding for the power circuitry and PCB design used. They're current iteration is dubbed Ultra Durable 5….
For that price it should come
For that price it should come with a real audio card…sheesh.
I think you know im a Huge
I think you know im a Huge PcPer Fan.
i used to watch live when i could staying until 5-6AM not to miss when i could, and your reviews are industry FACTS.
I also admit to you, that when ever i reach this part in a review “we would like to thank our friends at XYZ”
it bugs me.
it shouldn’t I know, & i know
readers interest is in your mind first & Co-operations last.(as it should)
This days PCper is significant and important enough to the hardware world so you don’t have to display this public symbolic bow down in thank to the conglomerate.
(it should be the opposite, as you market what they sell)
i would wish PCper to consider the option to Omit that small yet Symbolic enough line.
i know its really not what is important
but Symbolism has its powers..
your fan
Panta
Not sure how many samples you
Not sure how many samples you have tested. However in regard to the disapproval regarding the BIOS assisted overclocks assigned voltages being off, I not only think the opposite (they are literally SPOT ON) but have hard data to back that up.
I have experienced with my own cpu, and read posts time and time again… where people have cited being able to increase voltages by xx and get stable up to 4.6, however anything past 4.6 takes HUGE jumps and some cant get stable past 4.6 no matter what.
I personally use the board you reviewed (which is why I was here) but wish I had waited for the next round of 1150 chips to launch as haswell isn’t all it cracked up to be imo.
to get stable at 4.6 It takes me 1.28 (which is considered above average, and not too far off from your 1.20-1.25 however that I would consider FAR above average) and for 4.7 it takes me an additional .15v to get stable. HUGE jump. 1.43v to be exact. Now my temps don’t even hit 70c via IBT at that voltage (water) and the bios assigned 1.4v with the 4.7 auto overclock setting that you cited as having too high of a voltage, wouldn’t even have me stable.
Before tossing tossing that out there as fact and deter’ing a bunch of people…. see the numbers it takes a few cpu’s to make those last few jumps. Depending on cooling it may not even be possible (god forbid the people that wont or cant delid, haswell is complete garbage then. Atleast on an enthusiast level). Or atleast read around, and remember until there is a hard-line agreement on what is “stable” you have the people with a brain that realize most people assume/use Prime95 and that by calling their pc/numbers stable via whatever other benchmark they decide only harms others and the community as a whole.
that is all.