ASUS ENGTX580 DirectCU II Reviewccokeman -
» Discuss this article (0)
All it takes is a quick look to see that the ENGTX580 DirectCU II is quite a bit different from the reference design. The large dual-fan DirectCU II cooling assembly is the first clue. The second clue appears when the card is flipped over. Usually there is not much to talk about here, but the DirectCU II version has a large back plate that is used to help stiffen the PCB and is part of ASUS' own GPU Guard build process — part of the ASUS Xtreme Design philosophy. ASUS uses this process to increase the lifespan of the card by eliminating PCB flex with massive improvements in rigidity (238%). Additionally, a NEC "Super Cap" is located on the back side of the GPU core. This Proadlizer allows for a cleaner power supply to the GPU core to drive stability at higher overclocking levels. Now back to the front of the card. The large shroud that houses the two 100mm fans will physically occupy only one slot on your motherboard, but will cover three slots just due to the sheer size of the cooling solution. Comparing the ENGTX DirectCU II to a reference dual-slot card shows the dimensions of the card more clearly. Measuring 11.5 inches in length, the ENGTX580 DirectCU II should not provide any drama when installing it into most large mid tower and bigger chassis.
Connectivity options on the ENGTX580 DirectCU II include a pair of DVI ports (one Dual Link the other Single Link), Native HDMI, and a DisplayPort for added functionality. Running a surround or 3D Surround setup will still require the use of a second GTX 580 in an SLI configuration. The three-slot cooling solution vents a large percentage of the thermal load out of the chassis. The back end of the ENGTX580 DirectCU II shows the power and monitoring connections for the dual 100mm fans, the dual 8-pin power connections, and a pair of fuses under the power connections.
The ENGTC580 DirectCU II supports the ability to run an SLI configuration with up to four cards on a supported motherboard. The key to this is running them under liquid cooling or liquid nitrogen, as the large DirectCU II cooling will prevent anything more than a pair of these cards running together in the vast majority of systems. By running two or more of these cards from ASUS in SLI, you open up other options in the NVIDIA ecosystem, including running a surround setup with three monitors. Couple the surround setup with NVIDIA's 3D Vision kit and you move up to 3D Surround for a fully immersive gaming experience, allowing the end user to use the full spectrum of enhancements from the use of an NVIDIA GPU. Power above the 75 watts supplied by the PCIe slot is supplied from a pair of 8-pin PCIe connections, adding up to 300 watts of additional capacity.
Surprisingly, the large direct contact heat sink assembly is held to the PCB by four spring-loaded screws. Once pulled off, the black PCB can be seen. The rear plate/heat sink cannot be removed until the main assembly is removed, so the screws can be accessed. The back plate has a thermal pad that is used to pull additional heat from SAP 8 phase power circuit for dissipation via the aluminum backplate.
One of the reasons this card is a favorite of extreme enthusiasts is the robust power system and the components used to build the card. The ENGTX580 DirectCU II uses ASUS Super Alloy Power technology to drive the current to the GPU and memory with an eight phase power design, an improvement over the six phase design on the reference design. The components consist of SAP chokes that are made with a concrete core and special alloys so that they do not buzz under load and run up to 35 ºC cooler than traditional designs. Super Alloy capacitors are used to increase the lifespan of the card. Super Alloy MOS are smaller and more efficient and allow a 30% upswing in voltage for increased overclocking. The Super Hybrid Engine optimizes the power profiles in real time to increase performance while the Super Cap on the back side of the GPU is used to feed the core a steady diet of low noise power for increased overclocking ability. The last item is the series of fuses used as a means to fully protect the core from over current situations even when OCP is used on board — a good thing considering the DirectCU II does not have the hardware and software current limitations imposed on the reference design.
Held on by four screws, the heat sink assembly is easily removed for inspection. The cooling solution is what gives the card the DirectCU II name. The heat sink is a direct contact heat pipe design that is more efficient when implemented properly than a series of heat pipes that mount to a base plate. Stripped to the core, the cooling solution consists of three parts: the shroud, the heatsink, and the fan assembly that holds a pair of 100mm fans.
The heat sink uses five large heat pipes to carry the thermal load from the GPU core to the two fin arrays. Two go to the front and three to the rear. The airflow through the fin arrays is used to cool the memory and heat sink on the SAP mosfets. The base is much like many of the heat pipe direct contact designs out on the market. It is smooth for this style design without large gaps in the surface that need to be filled with "TIM". Based on the load temperatures seen in the testing, this heat sink does the job of cooling this high performance GPU down much better than the reference design. The SAP MOS get cooling in the form of a single aluminum heat sink along with the backplate, so the entire power circuit is cooled effectively.
An integral part of the success of the DirectCU II cooling package is the pair of 100mm fans used to push 600% more airflow through the heat sinks than the reference solution is capable of. All the while, running quieter to deliver excellent cooling. Looking at the specifications of the fan, it runs with a maximum speed of 3450 RPM, pushes 67.9 CFM, and has a static pressure rating of 5.5 mmH20 at just over 40 dBA.
Much like the reference GTX 580, the ENGTX580 DirectCU II uses a GF110 core with four graphics processing clusters, 16 streaming multiprocessors (instead of 15), 512 CUDA cores (instead of 480), 16 Polymorph engines, 64 texture units, and 48 ROP units. Instead of the reference clock speeds, the core sees a bump to 782MHz on the fixed function units and 1564MHz on the CUDA cores. The ASUS ENGTX580 uses the reference specification 1536MB of GDDR5 memory running through a 384-bit interface with 768K of shared L2 cache. Manufactured by Hynix, the 1.5GB of GDDR5 memory is rated to run at 1250MHz using 1.5v.
Now that there is a better understanding of the build quality of the ENGTX580 DirectCU II, it's time to find out just what kind of performance it delivers on air.