Learn About Computer Hardware – Ultimate Guide 2
The world of GPUs can be a scary place fraught with big words, bigger numbers, and lots of confusing nomenclature. Allow us to un-confuse things a bit for you
The amount of memory a GPU has is also called its frame buffer (see below). Most cards these days come with 1GB to 3GB of memory, but some high-end cards like the GTX Titan have 6GB of memory. In the simplest terms, more memory lets you run higher resolutions, but read the Frame Buffer section below for more info.
GPUs nowadays include compartmentalized subsystems that have their own processing cores, called Stream Processors by AMD, and CUDA cores by Nvidia, but both perform the same task. Unlike a CPU, which is designed to handle a wide array of tasks, but only able to execute a handful of threads in parallel at a high clock speed, GPU cores are massively parallel and designed to handle specific tasks such as shader calculations.
They can also be used for compute operations, but typically these features are heavily neutered in gaming cards, as the manufacturers want their most demanding clients paying top dollar for expensive workstation cards that offer full support for compute functionality. Since AMD and Nvidia’s processor cores are built on different architectures, it’s impossible to make direct comparisons between them, so just because one GPU has more cores than another does not automatically make it better.
The memory bus is a crucial pathway between the GPU itself and the card’s onboard frame buffer, or memory. The width of the bus and the speed of the memory itself combine to give you a set amount of bandwidth, which equals how much data can be transferred across the bus, usually measured in gigabytes per second. In this respect, and what generally stands with all things PC, more is better.
As an example, a GTX 680 with its 6GHz memory (1,500MHz quad-pumped) and 256-bit interface is capable of transferring 192.2GB of data per second, whereas the GTX Titan with the same 6GHz memory but a wider 384-bit interface is capable of transferring 288.4GB per second. Since most modern gaming boards now use 6GHz memory, the width of the interface is the only spec that ever changes, and the wider the better. Lower-end cards like the HD 7790, for example, have a 128-bit memory bus, so as you spend more money you’ll find cards with wider buses.
This technology is available in high-end GPUs, and it allows the GPU to dynamically overclock itself when under load for increased performance. GPUs without this technology are locked at one core clock speed all the time.
The frame buffer is composed of DDR memory and is where all the computations are performed to the images before they are output to your display, so you’ll need a bigger buffer to run higher resolutions, as the two are directly related to one another. Put simply, if you want to run higher resolutions—as in fill your screen with more pixels—you will need a frame buffer large enough to accommodate all those pixels.
The same principle applies if you are running a standard resolution such as 1080p but want to enable super-sampling AA (see below): Since the scene is actually being rendered at a higher resolution and then down-sampled, you’ll need a larger frame buffer to handle that higher internal resolution.
In general, a 1GB or 2GB buffer is fine for 1080p, but you will need 2GB or 3GB for 2560×1600 at decent frame rates. This is why the GTX Titan has 6GB of memory, as it’s designed to run at the absolute highest resolutions possible, including across three displays at once. Most midrange cards now have 2GB, with 3GB and 4GB frame buffers now commonplace for high-end GPUs.
High resolutions require a lot of RAM, which is embedded in the area around the GPU just like on this 6GB GTX Titan.
All modern GPUs use PCI Express power connectors, either of the 6-pin or 8-pin variety. Small cards require one 6-pin connector, bigger cards require two 6-pin, and the top-shelf cards require one 8-pin and one 6-pin. Flagship boards like the GTX 690 and HD 7990 need two 8-pin connectors. Most high-end cards will draw between 100–200W of power under load, so you’ll need around a 500–650W PSU for your entire system. Always give yourself somewhat of a buffer, so when a manufacturer says a 550W PSU is required, go for 650W.
These are what connect your GPU to your display, the most common being DVI, which comes in both single-link and dual-link. Dual-link is needed for resolutions up to 2560×1600, while single-link is fine for up to 1,200 pixels vertically. DisplayPort can go up to 2560×1600, as well. HDMI is another connector you will see: versions 1.0–1.2 support 1080p, 1.3 supports 2560×1600, while 1.4 supports 4K.
PCI Express 3.0
The latest generation of graphics cards from AMD and Nvidia are all PCIe 3.0, which theoretically allows for more bandwidth across the bus compared to PCIe 2.0, but actual in-game improvement will be slim-to-none in most cases, as PCIe 2.0 was never saturated to begin with. Your motherboard chipset and CPU must also support PCIe 3.0, but most Ivy Bridge and older boards do not support it in the chipset, even though the CPU may have the required lanes. In general, every GPU has PCIe 3.0 these days, but if your motherboard only supports version 2.0 you will not suffer a performance hit.
GPU coolers fall into several different categories, including blower, centralized, and water-cooled. The blower type is seen on most “reference” designs, which is what AMD and Nvidia provide to their add-in board partners as the most cost-effective solution typically. It sucks air in from the front of the chassis, then blows it along a heatsink through the back of the card to be exited out the rear of your case.
Centralized coolers have one or two fans in the middle that suck air in from anywhere around the card and exhaust it into the same region, creating a pocket of warm air below the card. Water-cooled cards are very rare, of course, but use water to absorb heat contained within a radiator, which is cooled by a fan. Water cooling is usually the most effective (and quiet) way to cool a hot PC component, but its cost and complexity make it less common.
This is Nvidia technology baked into its last few generations of GPUs that allows for hardware-based rendering of physics in games that support it, most notably Borderlands 2, so instead of just a regular explosion, you will see an explosion with particles and volumetric fog and smoke. Typically, AMD card owners will see the PhysX option grayed out in the menus, but the games still look great, so we would not deem this technology a reason to go with Nvidia over AMD at this point in time.
Different GPUs offer different types of antialiasing (AA), which is the smoothing out of jaggies that appear on edges of surfaces in games. Let’s look at the most common types:
Full Scene AA (FSAA, or AA):
The most basic type of AA, this is sometimes called super-sampling. It involves rendering a scene at higher resolutions and then down-sampling the final image for a smoother transition between pixels, which appears like softer edges on your screen. If you run 2X AA, the scene will be calculated at double the resolution, and 4X AA renders it at four times the resolution, hence a massive performance hit.
Multi-Sample AA (MSAA):
This is a more efficient form of FSAA, even though scenes are still rendered at higher resolutions, then down-sampled. It achieves this efficiency by only super-sampling pixels that are along edges; by sampling fewer pixels, you don’t see as much of a hit as with FSAA.
Fast Approximate AA (FXAA):
This is a shader-based Nvidia creation designed to allow for decent AA with very little to no performance hit. It achieves this by smoothing every pixel onscreen, including those born from pixel shaders, which isn’t possible with MSAA.
This is specific to Kepler GPUs and combines MSAA with post-processing to achieve higher-quality antialiasing, but it’s not as efficient as FXAA.
Morphological Antialiasing (MLAA):
This is AMD technology that uses GPU-accelerated compute functionality to apply AA as a post-processing effect as opposed to the super-sampling method.
Though the basic functionality of Wi-Fi routers has remained relatively unchanged since the olden days, new features have been added that help boost performance and allow for easier management
The band that a router operates on is key to determining how much traffic you will have to compete with. You would never want to hop on a congested freeway every day, and the same logic applies here. Currently there are two bands in use: 2.4GHz and 5GHz. Everyone and their nana is on 2.4GHz, including people nuking pizzas in the microwave, helicopter parents monitoring their baby via remote radios, and all the people surfing the Internet in your vicinity, making it a crowded band, to say the least.
However, within the 2.4GHz band you still have 11 channels to choose from, which is how everyone is able to surf this band without issues (for the most part). But if everyone is using the same channel, you will see your bandwidth decrease. On the other hand, 5GHz is a no-man’s-land at this time, so routers that can operate on it cost a pretty penny since it’s the equivalent of using the diamond lane, and a great way to make sure your bandwidth remains unmolested.
This stands for multiple-input, multiple-output and it’s the use of multiple transmitters and receivers to send/receive a Wi-Fi signal in order to improve performance, sort of like RAID for storage devices but with Wi-Fi. These devices are able to split a signal into several pieces and send it via multiple radio channels at once. This improves performance in a couple of ways.
When only one signal is being sent, it has to bounce around before ending up at the receiver, and performance is degraded. When several signals are sent at the same time, however, spectral efficiency is improved as there is a greater chance of one hitting the receiver with minimal interference; it also improves performance with multiple streams of data being carried to the receiver at once.
Channel bonding is something that’s done by the router and the network adapter whereby parallel channels of data are “bonded” together much like stripes of data in a RAID. This technology is most prevalent in 802.11n networks, where channel bonding is required for a user to utilize the full amount of bandwidth available in the specification. The downside to channel bonding is that it increases the risk of interference from nearby networks, which can reduce speeds. Since each channel is 20MHz, “bonded mode” operates at 40MHz, so check your settings to see if you can enable this.
Every router adheres to a specific 802.11 standard, which governs its overall performance and features. In the old days, there was 802.11a/b, then 802.11g, then 802.11n, which is the most widespread specification in use today since it’s been around for a few years and is relatively fast. Waiting in the wings is 802.11ac, which by default broadcasts on the uncongested 5GHz band, but is also backward compatible with 2.4GHz.
Whereas 802.11g had a peak throughput of 300Mb/s, 802.11n has a peak of roughly 500Mb/s, and 802.11ac doubles that to an unholy 1.3Gb/s. It achieves this speed increase by supporting up to eight channels compared to 802.11n’s four, and through increased channel width, using 80MHz and an optional 160MHz channel.
Quality of Service (QoS)
QoS is a common feature on today’s routers, and it lets you dictate which programs get priority when it comes to network bandwidth. You could theoretically slow down uTorrent while giving Netflix and Skype or Battlefield 3 more bandwidth. One crucial point is that the QoS setting is most important for outgoing traffic such as torrents, since incoming traffic is usually already prioritized by your ISP.
High-end 802.11n routers are able to broadcast dual networks on both 2.4GHz and 5GHz bands, though the new 802.11ac standard uses the 5GHz band by default.
System RAM, or memory, seems like such a basic thing, but there’s still much to know about it
The clock speed of RAM is usually expressed in megahertz, so DDR3/1866 runs at 1,866MHz, at a certain latency timing. The only problem is that modern CPUs pack so much cache and are so intelligent in managing data that very high-clocked RAM rarely impacts overall performance.
Going from, say, DDR3/1600 to DDR3/1866 isn’t going to net you very much at all. Only certain bandwidth-intensive applications such as video encoding can benefit from higher-clocked RAM. The sweet spot for most users is 1,600 or 1,866. The exception to this is with integrated graphics. If the box will be running integrated graphics, reach for the highest-clocked RAM the board will support and you will see a direct benefit in most games.
Modern CPUs support everything from single-channel to quad-channel RAM. There isn’t really a difference between a dual-channel kit and a quad-channel kit except that the vendor has done the work to match them up. You can run, for example, two dual-channel kits just fine. The only time you may want a factory-matched kit is if you are running the maximum amount of RAM or at a very high clock speed.
Voltage isn’t a prominent marketing spec for RAM but it’s worth paying attention to, as many newer CPUs with integrated memory controllers need lower-voltage RAM to operate at high frequency. Older DDR3, which may have been rated to run at high frequencies, could need higher voltage than newer CPUs are capable of supporting.
Heat is bad for RAM, but we’ve never been able to get any vendor to tell us at what temperature failures are induced. Unless you’re into extreme overclocking, if you have good airflow in your case, you’re generally good. We’ve come to feel that heatspeaders, for the most part, are like hubcaps. They may not do much, but who the hell wants to drive a car with all four hubcaps missing?
Capacity, Registered DIMMs, and Error Correction
It’s pretty easy to understand capacity on RAM—16GB is more than 8GB and 4GB is more than 2GB. With unbuffered, nonregistered RAM, the highest capacity you can get to run with a consumer CPU are 8GB modules. Registered DIMMs, or buffered DIMMs, usually refers to extra chips, or “buffers,” on the module to help take some of the electrical load off the memory controller.
It’s useful when running servers or workstations that pack in a buttload of RAM. ECC RAM refers to error-correcting control and adds an additional RAM chip to correct multi-bit errors that can’t be tolerated in certain high-precision workloads. If this sounds like something you want, make sure your CPU supports it. Intel usually disables ECC on its consumer CPUs, even those based on the commercial ones. AMD, on the other hand, doesn’t. For most, ECC support is a bit overkill, though.
We’re not sure what RAM heatsinks do today except look cool.
Power Supply Unit
The power supply doesn’t get all the attention of, say, the CPU or the video card, but disrespect the PSU at your own peril
The actual wattage of the PSU is the spec everyone pays attention to. That’s because 650 watts is 650 watts, right? Well, not always. One maker’s 650 watts might actually be more like 580 watts or lower at the actual temperature inside your case on a hot day. Despite all this, the wattage rating is still one of the more reliable specs you can use to judge a PSU.
How much you need can only be answered by the rig you’re running. We will say that recent GPU improvements have caused us to back away from our must-have-1,000W-PSU mantra. These days, believe it or not, a hefty system can run on 750 watts or lower with a good-quality PSU.
After wattage, efficiency is the next checkmark feature. PSU efficiency is basically how well the unit converts the power from AC to DC. The lower the efficiency, the more power is wasted. The lowest efficiency rating is 80 Plus, which means 80 percent of the power at a load of 20 percent, 50 percent, or 100 percent is converted. From there it goes to Bronze, Silver, Gold, and Platinum, with the higher ratings indicating higher efficiency.
Higher is better, but you do get diminishing returns on your investment as you approach the higher tiers. An 80 Plus Silver PSU hits 88 percent efficiency with a 50 percent load. An 80 Plus Platinum hits 92 percent. (Efficiencies for the higher tiers vary at different loads.) Is it worth paying 40 percent more for that? That’s up to you.
Single-rail vs. Multi-rail
A single-rail PSU spits out all the power from a single “rail,” so all of the 12 volt power is combined into one source. A multi-rail splits it into different rails. Which is better? On a modern PSU, it doesn’t matter much. Much of the problems from multi-rail PSUs were in the early days of SLI and Pentium 4 processors.
PSU designs that favored CPUs, combined with the siloing of power among rails, proved incapable of properly feeding a multi-GPU setup. Single-rail designs had no such issues. These days, multi-rail PSUs are designed with today’s configs in mind, so multi-GPUs are no longer a problem.
Intelligent vs. Dumb
A “dumb” power supply is actually what 99 percent of us have: a PSU that supplies clean, reliable power. An “intelligent” PSU does the same but communicates telemetry to the OS via USB. Some smart PSUs even let you adjust the voltages on the rails in the operating system (something you’d have to do manually on high-end units) and let you control the fan temperature intelligently, too. Do you need a smart PSU? To be frank, no. But for those who like seeing how efficient the PSU is or what the 5-volt rail is, it’s pretty damned cool.
Modular vs. Non-modular
Modular PSUs are the rage and give you great flexibility by letting you swap in shorter cables, or cables of a different color, or to remove unused cables. The downside is that most high-end machines use all of the cables, so that last point in particular is moot—what’s more, we think it’s too easy to lose modular cables, which sucks.
Modular power supplies are the rage today—just don’t misplace the cables.
How to dole out system advice like a pro
Warning: As a PC expert, you will be called upon often by family and friends for system-buying advice. After all, purchasing a new PC retail can be a daunting task for the average consumer. Remember, you might know the difference between an AMD FX-8350 and FX-6100, but will Aunt Peg?
This machine is probably too much PC for Aunt Peg to handle.
No, Aunt Peg will walk into the local Big Box with the goal of spending $750 on a basic all-in-one and end up walking out with a $3,000 SLI rig. We’re not saying that Aunt Peg doesn’t like getting her frag on as much as the rest of us, but let’s face it, she needs some basic buying tips.
Peg, what level of CPU you require depends on your needs. If your idea of a good time is Bejeweled, email, and basic photo editing, a dual-core processor of any model except Atom is more than enough. If you’re looking for more performance, the good thing is that Intel and AMD’s model numbers can mostly be trusted to represent actual performance. A Core i5 is greater than a Core i3 and an A10 is faster than an A8. If you are doing home video editing, Peg, consider paying for a quad-core CPU or more.
There are three known levers pulled when convincing consumers to buy a new PC: CPU, storage size, and amount of RAM. You’ll often see systems with low-end processors loaded up with a ton of RAM, because someone with a Pentium is really in the market for a system with 16GB of RAM (not!). For most people on a budget, 4GB is adequate, with 8GB being the sweet spot today. If you have a choice between a Pentium with 16GB and a Core i3 with 8GB, get the Core i3 box.
Storage is pretty obvious to everyone now, and analogous to closet space. You can never have enough. What consumers should really look for is SSD caching support or even pony up for an SSD. SSD caching or an SSD so greatly improves the feel of a PC that only those on a very strict budget should pass on this option. SSDs are probably one of the most significant advances to PCs in the last four years, so not having one is almost like not having a CPU. How large of an SSD do you need? The minimum these days for a primary drive is 120GB, with 240GB being more usable.
There’s a sad statistic in the PC industry: Americans don’t pay for discrete graphics. It’s sad because a good GPU should be among the top four specs a person looks at in a new computer. Integrated graphics, usually really bad Intel integrated graphics, have long been a staple of American PCs. To be fair, that’s actually changing, as Intel’s new Haswell graphics greatly improves over previous generations, and for a casual gamer, it may even finally be enough. Still, almost any discrete GPU is still faster than integrated graphics these days. Aunt Peg might not play games, but her kids or grandkids might and not having a GPU will give them a frowny face. A GeForce 650 or Radeon HD 7770 is a good baseline for any machine that will touch games.
If you have any question, Please leave a comment.