On top of the PCB is a metal bracket that supports the heatsinks and fan, and provides further cooling to the memory modules and VRM circuitry. It also serves to provide structural rigidity to ensure that the PCB does not flex given its length and the weight of the heatsink at the end. Attached to it is a plastic shroud that can also be removed separately. The bladed fan is different to the blower type ones found on the GTX580/GTX570.


One of the ways Nvidia have attempted to tackle the heat generated by the core is a vapour chamber cooler. The copper heat plate contains liquid in the base (the raised middle part of the plate that is in contact with the GPU). The liquid evaporates and absorbs further latent heat through its phase transition to vapour. This rises and condenses on the outside upper part of the heat plate releasing heat which is then transmitted to the overlying fin array whilst the condenser fluid circulates around the outside of the heat plate. This was first introduced on Nvidia GTX570 & GTX580 where they worked very efficiently. In total, the heatsink on top of the vapour chamber is constructed using 33 aluminium fins oriented vertically to allow air through. Each GPU has its own heatsink and both are shaped slightly differently. The actual size of the individual heatsink is smaller than those on Nvidia’s single GPU high end cards due to space limitations.


At the base of the metal bracket are raised areas covered with thermal pads to cool the memory modules and MOSFETs. The base of the vapour chamber is made of copper and has a large enough surface area for heat dissipation.


A small LED strip on the inner side of the shroud lights up the GeForce branding on the side of the graphics card. This small touch goes a long way at raising the appeal of the card, especially when on display in windowed cases.


The PCB is jammed packed with components but the dominating features are the two GF110 GPUs with their 520mm squared die size packing over 3 billion transistors on each. Both are surrounded by a further 6 memory modules to add to the 6 on the back of the PCB. The graphics card follows an asymmetrical design but the VRM circuitry is situated in the middle to minimise trace length. Hardly any PCB real estate goes unused.


The first GPU is clearly labelled GF110 so we can expect it to support the same features and specifications as the GTX580 albeit with lower clocks. The memory modules are also covered by thermal pads for additional cooling.


At the other end, we find the second GPU surrounded by its own bank of memory modules. A 4 pin fan header is present on the side for fan PWM functionality and a 2 pin header at the edge of the PCB is for the LED strip.


There are a total of 24 128MB memory modules for 3GB of on-board GDDR5 memory although 1.5GB are assigned to each GPU. The modules have been sourced from Samsung and carry the model name K4G10325FE-HC04, the same as those found on the GTX580. These are rated at 0.4ns for a rated speed of 5Gbps. The memory on the GTX590 only come clocked at 3414MHz, suggesting high overclocking headroom.


The VRM circuitry on the GTX590 uses a 5 phase PWM design for each GPU and a further 2 for each bank of 1.5GB memory as denoted by the number of chokes and associated MOSFETs. This is not as robust as the GTX580’s 6 phase design but it is one better than the GTX570. Although voltage tweaking is disabled, there is still slight but limited headroom for overclocking without risk of the components blowing up.


The GTX590 relies on the Nvidia bridge chip for on-board SLI so users don’t need an SLI compliant motherboard for the functionality. This is ideal for current high end AMD motherboard based around AMD’s own chipset that lacks the capability without a Lucid Hydra chip. Since the board already relies on SLI, a multi GPU setup is limited to another single GTX590 for Quad SLI.


The graphics card features two CHiL CHL 8266 voltage controllers for voltage monitoring, the same used on the GTX580s. The first controller is found amongst the VRM circuitry.


The second voltage controller is closer to the edge of the PCB just below SLI bridge chip.