The EVGA GeForce GTX 580 Classified 3 GB looks special compared not only to the reference GeForce GTX 580 but also to other graphics cards in general. The chrome letters of the product name are sunken into the cooler’s black plastic casing. A plate with the word Classified is affixed below.
The card’s key features are printed in small white letters on the casing around the cooler's fan. There is another “Classified” plate at the top.
The card measures 280 x 40 x 137.7 millimeters. The large dimensions are due to the cooler which has an additional vent grid in its top part:
The weight of the card is an impressive 1361 grams but we don’t think this can scare off true enthusiasts.
The EVGA GeForce GTX 580 Classified is equipped with two dual-link DVI-I outputs that support high resolutions. There is no HDMI or mini-DisplayPort here, obviously due to the overclocker-friendly positioning of the product. Take a look at its mounting bracket:
There are large vent grids here which are meant to ensure better cooling at less noise (they do not succeed in achieving the latter goal, though, as you will learn shortly). You can also see a connector for an EVGA EVBot device that can be used to overclock a mainboard together with three graphics cards.
The MIO connectors in the top part of the PCB allow building 2-, 3- and even 4-way SLI configurations.
The EVGA GeForce GTX 580 Classified is equipped with one 6-pin and two 8-pin power connectors.
This doesn’t mean the EVGA card consumes 50% more power than a reference GeForce GTX 580. Its specs mention the same peak power draw of 244 watts. The extra power connector is only implemented to ensure stability if the card is overclocked further. A 600-watt or better power supply is recommended for a computer with this graphics card.
The EVGA GeForce GTX 580 Classified offers three connectors for measuring voltages:
It also has LED indicators and a switch that changes its operation mode between Normal and OC:
Here is a photo of the LEDs shining:
The PCB of the EVGA card is completely different from the reference one in size as well as components.
First off, we’ve got a 17-phase power circuit here with three phases for the graphics memory and controller and 14 phases for the GPU. Then, the card uses high-quality DirectFETs which are low-resistance and low-inductance MOSFETs with superior thermal properties.
Three Proadlizer NEC/TOCIN capacitors are installed on the PCB, one of them on the reverse side.
These capacitors feature improved stability at high frequencies and voltages and are employed by several makers in their top-end products.
Dated the 24th week of 2011 (mid-June), the Nvidia GF110 revision A1 chip was manufactured in Taiwan on 40nm tech process. It’s covered with a heat-spreader.
The GPU has 512 unified shader processors, 64 texture-mapping units and 48 raster operators, in full compliance with Nvidia's reference specs, but its 3D frequencies are pre-overclocked to 855/1710 MHz at a voltage of 1.143 V. This is the same factory overclocking as with the Gigabyte GeForce GTX 580 SuperOverclock we tested recently. The GPU clock rates are lowered to 51/101 MHz at 0.962 volts for 2D applications.
This GeForce GTX 580 has twice the amount of the reference card’s onboard memory. The 12 FCFBGA-packaged chips from Hynix Semiconductor, located on the face side of the PCB, sum up to a total of 3072 megabytes.
The barely readable marking says H5GQ2H24MFR T2C which means an access time of 0.4 nanoseconds, a rated clock rate of 5000 MHz and a voltage of 1.5 volts (1.35 volts for 2D applications). The memory frequency of the EVGA GeForce GTX 580 Classified is 4106 MHz, which is a mere 100 MHz above the reference card’s. That’s not much of overclocking, but the low rated access time of the chips is indicative of high overclocking potential.
Here is a summary of the card’s specs: