Graphics Card

Function and Importance [Source]

The graphics card (or video card) is the part responsible for supplying images to the display monitor. Often referred to as the visual processing unit, it works in conjunction with input from the processor to generate 3D images binary data.

002_062815.jpg

[Source] – Above is the 3D representation of the internals of a dedicated graphics card. Like a processor it requires good cooling to dissipate the intense heat generated through operation, so the basic structure will consist of the GPU circuit board, a heatsink and fans. A backplate is an additional feature which is present on some higher end cards and aids with cooling and improving the appearance.

How important is it? Well, here’s where we go into detail about integrated versus dedicated, and, all the various terms thrown around including clock speed, bandwidth and VRAM.

[Source] – Firstly, integrated graphics are included in most CPUs and rely on system RAM for its processing; they’re cheaper, produce less heat and use less power. If the consumer is after a desktop to handle everyday applications, it’s perfect – if however you’re wanting a gaming machine, it won’t be enough. The integrated solution is just not powerful enough to allow for playing intensive games, it would likely handle older titles and perhaps even get a newer one to open – but at dramatically low resolution and visual settings.

Dedicated graphics cards are a separate unit and have their own memory – video RAM or VRAM. In contrast to the first type, they will use more power – requiring often a VGA cable from the power supply to supply the additional voltage needed, which also means they run hotter and are an additional cost to the build.

VRAM is the most common selling points in GPUs and generally the more you have, the better. There are however other factors which come into play and I will go through each you are most likely to encounter:

Clock – core, core boost and memory. Commonly displayed in MHz, in contrast to a CPUs GHz, these are the frequencies at which the certain part runs. Core clock is the base speed of the GPU, typically around 1000MHz – it isn’t as important when comparing cards to look at this, as other features will affect performance more however it’s useful when doing an overclock. Core boost is the “inbuilt” overclock, the base clock will increase to this value under heavy load to provide extra performance within the manufacturers safe limits. Memory clock is the speed of the VRAM, like in the RAM section – this will dictate how fast stored information is moved on.

Untitled.png

[Source] – Specification for the 980Ti

Memory Size – this is VRAM. Today, for 1080p maxed settings most games will require 4GB with some exceptions being GTA V which has a plethora of visual settings and an inbuilt indication of subsequent VRAM usage, which can peak near 6GB! Basically this dictates how much stuff can be dealt with, textures use memory – so if you have more, you can have higher quality textures – which will make the game look better. Playing at higher resolutions will also have the same effect.

Memory Bus – [Source] – it is “how much of the memory can be accessed per second” – these vary considerably between cards and again, it’s complicated – bandwidth, bus and speed determine performance. GDDR5 memory is currently the most used however HBM, high bandwidth memory, is now being adopted by both the major manufacturers and will lead to huge increases in performance when released.

CUDA cores / Stream Processors – [Source] – this is a complicated topic and the simplest I can put it is, the more you have the better… I’d advise having a read online if you’re interested enough but for the purposes of assembly, it’s not worth it.

Finally, you’ll see a list on the spec page of the various connections the card has – HDMI, display port etc… This is just the same as buying a TV, most cards now will use HDMI based connections as the older VGA and DVI leads are becoming less popular with enthusiasts.

hdmidvi

[Source]

Nvidia vs. AMD

Like Intel vs AMD – a topic of intensive ‘fanboy-ism’ from both sides which is incredibly hot right now, given the business position of AMD – unfortunately suffering from decreasing market share while it struggles to compete against Nvidia’s huge corporate resources.  Like the CPU page I’ll keep this as unbiased as possible and thankfully, or unfortunately for those looking for an easy decision, it’s an area you’ll have to research well to get the best for your money.

Basics:

  • Nvidia – run cooler and use less power – typically provide better performing cards for the enthusiast gamer.
  • AMDhotter, use more power – usually dominate the lower end of the market, where priceperformance is key.

There is simply no definitive answer for who is better, don’t ever set your mind on either manufacturer prior to deciding between cards. Like the other sections, set the budget, browse the web, compare benchmarks – that’s it.

Driver support is the topic of much debate amongst either ‘team’ – from the research I have done and reviews seen: AMD provide fewer, but more substantial driver updates resulting in bigger performance gains. Nvidia release more frequent updates, often resulting in smaller increases but often have the lead on releasing gameready drivers for new titles.

Also worth consideration is what sort of games you intend on playing, increasingly game companies are focussing more a certain manufacturer – with games such as Fallout 4 or The Witcher 3 choosing to optimise around Nvidia and DICE with Battlefront going AMD. Certain features will be entirely exclusive to either brand and although you will certainly run the game well with either branded card providing it’s of sufficient quality, unavoidably it will look better on some systems for this reason.

My Experience

I put the ‘EVGA GTX 980 Ti Superclocked+ ACX 2.0+’  in my rig [Source] – it is the pinnacle of gaming performance, though not as powerful as the Titan X it provides almost the same performance at half the cost giving one hell of a card for your money. Playing at 1080p is overkill for most things, but it ensures smooth gameplay at ultra settings – though some unoptimised titles such as Assassins Creed Black Flag do require settings to be turned down in order to prevent colossal frame rate drops present on every PC.

71Oq5+ZmYfL._SL1200_.jpg

For the other build – I put a ‘Palit GTX 750Ti StormX’ inside. My brother bought this for another computer which lacked a VGA cable, so needed a GPU which would run through the PCIE express slot power only. This card is great for mid-range 1080p gaming, it’s amazingly priced and runs cool. [Source]

51WnSVSzUkL

At the end of January this will be replaced with a GTX 970 to allow him to play at Ultra settings, since he lacked the funds at the time of the primary system build.