Table of Contents
With the advancement of computer technology, the number of specifications for a video card has become overwhelming for those not deeply involved in the computer industry. We often get questions such as "how much video RAM do I need?" or "what is the maximum resolution of HDMI?" In this article, we will explain the major specifications of video cards and what they mean for the end user. While these specifications are mainly tailored towards discrete video cards, many also apply to the onboard video found on motherboard.
If you are interested in learning about what the specifications mean on other components, also check out out Specs Explained: CPU article.
Most modern video cards will fall under one of five different product lines. Of these five lines, three are desktop lines meant for general use and gaming, while the other two are meant for workstations. For desktops, the thee most prevalent lines are Intel HD (onboard) AMD Radeon HD (onboard and discrete GPU) and NVIDIA GeForce (discrete GPU). For workstations, the two most prevalent lines are AMD FirePro and NVIDIA Quadro.
The main difference between desktop and workstation cards comes down to speed vs. precision. In everyday use, precision is not really needed since one small artifact every 1000 frames is not noticeable to the human eye. When you get into 3D modeling or simulation, however, one small artifact can cause big problems. In a professional environment, you want to be 100% sure that everything was completed properly the first time and that there are no small artifacts in the results. So in that case you would want to use a more precise – but usually slightly slower – workstation card.
The other difference between desktop and workstation is that many workstation cards include features (whether hardware-based or driver-based) that allow for better performance in some professional software suites from companies like Adobe and AutoDesk.
The motherboard connection is simply the type of slot on the motherboard that the video card is designed to be used in. Most cards will be either PCI Express 2.0 x16 or PCI Express 3.0 x16. While you want to match the PCI Express revision (2.0 versus 3.0) of the card to the motherboard slot if possible, you can use a PCI Express 2.0 card in a 3.0 slot or a PCI Express 3.0 card in a 2.0 slot. You may have a very slight performance decrease, but several benchmarks have shown that even the fastest video cards available today are not capable of using all the bandwidth available from PCI Express 2.0, let alone PCI Express 3.0.
The x16 refers to the number of PCI lanes the card requires. This gets a little confusing as there are often slots on a motherboard that are the same size as x16 slots, but actually operate at x8 speeds. In addition to this, on some boards if you use multiple slots (for SLI or Crossfire for example) at once, even if the two slots you are using are rated for x16 speeds, they will actually only run at x8 speeds. Unlike the PCI Express revision, you will likely see a small loss in performance when using mid to high-end video cards in an x8 slot, so be sure to check the manufacturer's documentation to determine which slots actually operate at x16 speeds.
If documentation is not available, you can usually tell the actual speed of the slot by the number of pins in the slot itself. In the blue slot above, the pins go all the way across the slot, so it is a full-speed X16 slot. On the black slot, the pins stop about half-way, so this slot is actually an X8 slot in an X16 size.
The maximum resolution is simply the highest resolution the card is capable of outputting. Often a video card will have multiple ports that will not be able to output at this maximum resolution so you always want to confirm the maximum resolution for the port you are intending to use. Sometimes this resolution is listed by the manufacturer (DVI Maximum Resolution, HDMI Maximum Resolution, etc.), but other times you will need to check the revision of the ports to determine what their maximum resolution should be.
Note that in order to achieve this maximum resolution, both the port and the display device you are using need to support it.
HDCP (high-bandwidth digital content protection) is a protection scheme designed to eliminate the possibility of data being intercepted between the video card and the display. While at first this may sound like a security feature, it is in fact a method of preventing HDCP-protected content from being played (or recorded) on unauthorized devices.
For example, many Blu-ray movies use HDCP so you would run into problems if you tried to play a Blu-ray movie from your computer with a video card that does not support HDCP. Results will vary, but either the movie will simply refuse to play or it will play at a lower resolution. However, almost all video cards and displays made in the last few years support HDCP, so HDCP support is rarely an actual issue.
The core speed (commonly referred to as the frequency) of a video card is the speed at which the GPU core operates and is typically reported in MHz or GHz. Due to variances in GPU architecture, two video cards with the same clock speed will not necessarily perform the same job in the same amount of time. The number of cores, memory size/type/speed, and overall architecture plays an important role as well. However, with all other things being equal, a higher clock speed card will always be faster than a lower clock speed card. The downside to a higher core speed is that the higher it is, the more power (and thus heat) is required.
Boost is a fairly new feature found on some video cards and is similar to Intel's Turbo Boost. Simply put, Boost is a temporary overclock that increases the video card's core frequency when additional graphical power is needed. This can only happen as long as it the GPU is below a certain power, current and temperature threshold. Due to this, your video card may not always achieve the maximum boost frequency listed in this specification.
Stream processors (also known as CUDA cores on cards that support CUDA) are the number of cores available on the video card. Each core contributes to the overall power of the video card, so by adding more cores the overall power of the GPU increases. Due to how cores work on video cards, there is pretty much no downside to having more cores.
One very important thing to note is that you cannot compare the number of cores across manufactures. For example, an AMD Radeon HD 7970 3GB has 2048 stream processors, while a NVIDIA Geforce GTX 670 2GB only has 1344 CUDA cores. But in terms of gaming performance, these two cards are very similar.
Video Memory is much like the main system's RAM in that it acts as a temporary storage area for data. Typically, the video card manufacture will use the amount of video RAM that is appropriate for the power of the card, but sometimes multiple versions are available. While more RAM is better, having more RAM than the software can use does not yield much (if any) performance advantages. So depending on the software, cards with more RAM may not always have a performance advantage over cards with less RAM. For some examples, check out our article Video Card Performance: 2GB vs 4GB Memory to see the gaming performance difference between a card with 2GB of RAM versus one with 4GB. Note that using multiple monitors or professional software may increase the benefit from having more video RAM.
The memory type of a video card is the type of RAM that is used on the video card. Most modern cards use GDDR5 which is essentially DDR3 RAM that has been optimized for graphical operations. Some basic video cards simply use DDR3 RAM, but sacrifice a decent amount of memory performance by doing so. Until a new version of video RAM is released, GDDR5 and DDR3 are the only two versions that should be found on a modern video card.
The speed of the video card's RAM is typically reported in MHz and is basically how fast the video card can access the data that is stored on the RAM. Obviously, the faster the card can access the data, the less time it has to wait. So in pretty much all instances, faster memory speed is better.
Memory Bus Width
While fast memory is important, the video card needs to actually be able to process the data from the memory quickly. Technically speaking, the bus width is the amount of data the video card can access from the memory each clock cycle. So if you are doing something that uses a lot of video memory, you want to have a large bus width in order to efficiently transfer data to and from the video card's memory. Just like the size of the RAM, more is better but you will see diminishing returns after a certain point.
Memory bandwidth is actually a calculation of several other memory specifications and can be used as an overall indication of how fast the video card's memory is. The higher the memory bandwidth, the better the video card's memory performance should be.
|VGA (Video Graphics Array) is one of the oldest types of video connector still in use and is most commonly found in servers and low to mid-range video cards. The connector itself consists of fifteen pins in three rows of five. This is one of the last analog connectors still in use, and as such has the lowest maximum resolution of any modern video port.|
VGA ports on modern hardware have a maximum resolution 2048×1536 at 85 Hz, but in reality most monitors that utilize VGA max out at much lower resolutions so it is rare to actually run VGA at this high of a resolution.
DVI connector types.
Courtesy of Wikipedia
DVI (Digital Visual Interface) is one of the most commonly video connectors used today and comes in many different flavors. DVI-A is an analog only version of DVI, but is nowhere near as common as DVI-D and DVI-I. The main difference between DVI-D and DVI-I is that DVI-D transmits digital signals only while DVI-I can transmit both analog and digital signals. DVI-I is really only useful if you want to use a DVI to VGA adapter to connect to an older VGA displays. Otherwise, DVI-I and DVI-D are exactly the same. You can tell the difference between DVI-D and DVI-I by the two pairs of additional pins above and below the horizontal tab.
In addition to the types of DVI, there is also single and dual link versions of the DVI-D and DVI-I connectors. Dual Link simply allows for more data to be transmitted at once, so it allows for higher resolutions. Almost all modern video cards and motherboards will be dual link.
Unless otherwise noted, the following are currently the most common maximum resolutions for DVI sorted by manufacturer:
|AMD (Motherboard & GPU):||2560×1600 @ 60 Hz|
|Intel (Motherboard):||1920×1200 @ 60 Hz|
|NVIDIA (GPU):||2560×1600 @ 60 Hz|
HDMI (High-Definition Multimedia Interface) is similar to DVI except that it allows for the transmission of audio as well as video. Currently, this is the preferred method for connecting devices to televisions or monitors with integrated speakers. HDMI is most commonly found as a full-sized port, but on some cards where space is at a premium, a mini-HDMI port is used instead.
There are a number of versions of HDMI, but most modern hardware will use a reversion of 1.4. The base 1.4 revision supports resolutions up to 4096×2160 @ 24 Hz, support for some 3D formats, and the rarely used HDMI Ethernet channel. The 1.4a revision adds increased 3D support, while the 1.4b revision added support for 1920×1080 video at 120 Hz. Currently, HDMI 1.4a is the most common revision found on video cards.
The difficult part with HDMI is the supported resolutions. While 1.4 goes all the way up to 4096×2160 (otherwise known as 4K), it can only do so with a refresh rate of 24Hz. If you want to use the more standard 60Hz refresh rate, you are actually limited to 2560×1440.
|HDMI 1.4 Resolutions||Max Refresh Rate||Max Color depth|
|1920×1080||120Hz with 1.4b||unlisted|
*Not an official specification, but confirmed in our own testing
DisplayPort is a newer video port than HDMI, but has not had a revision since version 1.2 in December of 2009. As such, almost every DisplayPort on modern hardware is the latest 1.2 version. The difference between DisplayPort and HDMI/DVI is that it uses packetized data transmission much like USB, SATA, or ethernet. The main advantage of this is that it can use a fewer number of pins to achieve higher resolution. A secondary advantage is that since the information is in packet form, features can be added to DisplayPort without any changes to the physical port.
DisplayPorts are typically found in two sizes: a full-sized port and a mini port. One little-known feature of DisplayPort is that you can daisy-chain supported DisplayPort monitors together so that a single DisplayPort is powering multiple monitors at once. To do this you either need monitors that support daisy-chaining or get a DisplayPort hub. Unfortunately, there are very few released products that can take advantage of this feature.