Jump to content

[GUIDE] Deciphering Computer Hardware Model Naming

Hardware companies like to throw down model numbers as their product names like you're supposed to know what any of this stuff means. I mean, why is there a GeForce GTX 1080 when there was a GeForce 9800 GTX? Doesn't the higher number mean it's better? Nevermind the GTX swapped places!

 

Fortunately though, there is a method to their madness. Most of the time. This guide is to lay out what each major hardware manufacturer's numbering schema means for parts such as the processor, the motherboard, and the video card. I'll also try to go over some other pieces of hardware, like storage drives and monitors, but beyond that, a model name and number usually doesn't mean anything.

 

Please note: This only covers products made from about 2010 and after, and thus, is by no means a comprehensive guide.

 

A note about model numbers. I will be repeatedly use the term GPX format to describe model numbers. Most companies have a standard way of assigning a model number and for the hardware being described, the manufacturers are good at using this. The GPX format are numbers in the model number that describe the following:

  • G stands for generation of the model series.
  • P stands for the performance or feature tier in the model series of that generation.
  • X are SKU numbers. While they usually mean a sub tier for performance and/or features, other manufacturers use them just because.

Note that there can be any number of G, P, or X numbers in a model number. For example, a GPXX format model number means there's two SKU numbers at the end.

 

Processors

Processors from Intel

Spoiler

Intel is both good and bad about their naming scheme. Good in that, usually at a first glance, you can tell what kind of performance tier the processor is even to the layman. Bad in that Intel doesn't make it obvious about the smaller details and if you're not paying attention, you're going to get screwed over because though this Core i7 was a quad-core part when it was really a dual core part.

 

For the most part, Intel follows this scheme:

[Model Name] [Model Subname]-[Model Number] [Suffix]

Example: Atom Z3750, Celeron J1800, Core i5-4670K, Core i7-5600U, Xeon E5-2686 v4

 

Model Name is the branding that Intel uses when marketing their parts. Examples of these include Core, Pentium, Celeron, Atom, and Xeon.

 

Model Subname represents the performance tier in that model name, if it has one. The Core (m3, m5, m7, i3, i5, and i7) and Xeon (E3, E5, E7)  models have subnames, whereas the Atom, Celeron, and Pentium models do not.

 

Model Number represents the SKU number of the part.

For consumer parts, it's a four-digit number following the GPXX format. Note that the generation number is not indicative of the architecture being used. The Pentium G4000 series uses both Skylake and Kaby Lake architectures.

 

For the Xeon lineup, the most significant digit of a model number is the number of processors you can install using that particular model in a multi-socket board. For example, an E5-1603 means you can only use this Xeon in single socket boards. An E5-2603 means you can use them in two socket boards. Other than that, a model number is a model number. Intel has provided no clear pattern on what the model names mean. Confusingly, they recycle model names as well and append it with a version number, for example, the Xeon E7-4809 has a v2, v3, and v4. And they don't have any similar specification (one is 6-core, the others are 8-core).

 

Generally speaking, the higher the number, the higher performance part . However, Intel on occasion has released parts that are of higher number but not of higher performance. (For those who want to be snarky about how Kaby Lake is an example, Intel has released parts that are of higher number, but lower performance in the same generation, such as the i7-4700HQ and the i7-4702HQ). If the model name contains a prefix letter, such as in the Pentium G3258, the letter doesn't mean anything.

 

Suffix represents special features or modifiers about the processor. Intel has been mostly consistent about this across generations, so here's a list of the ones they've used:

  • [No Letter] - A desktop class part with no special features.
  • F - No integrated GPU
  • H - Mobile part containing a "high performance" GPU.
    • Note: If used with the Core i7, it's usually a quad core part even if the entire name does not contain the Q suffix.
  • K - Multiplier is unlocked, allowing for easier overclocking.
  • M - Mobile part.
  • Q - Quad core mobile part.
    • This has fallen out of use over time
    • Note: This can trip you up if you aren’t careful. If you think a Core i7 is always a quad core part, you’ll be in for a surprise if you get a laptop with a Core i7-6500U as it’s a dual-core part.
  • R - Contains the Iris Pro GPU.
    • Note: Models can have an Iris Pro GPU but not have an R suffix, such as the Core i7-7560U.
  • S - Low power desktop part, usually 65W TDP.
    • This letter seems to be fallen out of use since Skylake, due to the no letter desktop part having a power rating of either 65W or 51W TDP.
  • T - Lower power desktop part, usually 35W TDP.
  • U - Low power mobile part, usually for Ultrabooks or 2-in-1's.
  • X - Extreme part, the highest-end on the consumer spectrum.
    • Note: If used in a mobile Core i7 model, it's usually a quad core part even if the entire name does not contain the Q suffix.
  • Y - Ultra low power mobile part, usually for tablets

If you want a generalized cheat sheet of all of this:

  • Model Names: Remember, Atom < Core M < Celeron < Pentium < Core/Xeon
    • Atoms up to 4 cores, 4 threads.
    • Celerons and Pentiums are usually 2 cores, 2 threads.
    • Core i3 is usually 2 cores, 4 threads. Lower end models may only be 2 cores, 2 threads.
    • Core i5 is usually 4 cores, 4 threads. Mobile parts may reduce this down to by doing 2 cores, 4 threads.
    • Core i7 is usually 4 cores, 8 threads. Mobile parts may reduce this down to 2 cores, 4 threads.
  • Model Number: Higher is better. Or rather, pay attention to only the two leftmost digits. The two rightmost digits do what they want.
  • Suffix: K if you care about overclocking. Pay closer attention if you're getting a Y or U processor so you know what you're really getting. You can generally ignore the others.

 

Processors from AMD

Spoiler

This will cover three families of processors, based on the AMD's last three major sockets: AM3+, FM1/FM2/FM2+/AM1, and AM4

 

Socket AM3+

Or really, the entire AMD FX lineup. AMD was really good at the naming convention because it's just this:

FX-[Core Count][Three digit model number][Lower power flag]

Example: FX-8320

 

Core Count is the number of integer cores the processor has, except for the FX 9000 series, which has 8 cores.

Model Number is the model number of that part. Generally the higher this is the better performing the part is.

Lower power flag is tacked on if the part has a lower power dissipation. It's denoted by the letter E, such as the FX-8320E.

 

Socket FM1/FM2/FM2+/AM1

This is where AMD took that really good naming convention and did bugger all with it. It generally follows this format:

[Model Name] [Model Subname] [Model Number] [Suffix]

 

Model name is the name of the part. For this socket series, it was Athlon, Sempron, and the A4/A6/A8/A10/A12 series.

Model subname applies to the Athlon and Sempron parts. This is easy though, it's the number of cores. X2 means a dual core part, X4 means a quad core part. However, AMD did not do this for the Kabini-based chips.

Model number has different meanings depending on the model name. For Athlon parts, it's usually a three digit model number. For Sempron and A-series parts, a four-digit model number that follows the GPXX format. Generally speaking, a higher number is better.

Suffix are special modifiers to the processor, if any. AMD uses two letters: K for an unlocked multiplier for easier overclocking and B for a second revision part.

 

If you're looking to get an A-series APU though, note that A4/A6's are dual core parts, while A8/A10/A12's are quad core parts. Otherwise, look for a higher part number and if you want overclocking, a K part.

 

Socket AM4

AMD went back to something sensible! Now the processing naming schema is:

Ryzen [Tier Number] [Model Number] [Suffix]

Example: Ryzen 7 1700X, Ryzen 3 1200

 

Tier Number is the performance tier the processor is in. As of this writing there's only 3, 5, and 7.

Model Number is the model number of the part. It looks like AMD is taking the GPXX approach.

Suffix is a modifier that shows any special features are on the processor:

  • G: Has integrated graphics.
  • X: Has the Extended Frequency Range feature which can boost the processor past the maximum turbo speed if cooling allows.
  • W: Meant primarily for workstations and other high-performance work.

 

At first glance, Ryzen 7 and 3 are easy to figure out their main features. The Ryzen 7 is a 8 core, 16 thread part while the Ryzen 3 is a 4 core, 4 thread part. Ryzen 5 is the one that has trouble, as it can be a 6 core, 12 thread part (1500 or 1600) or a 4 core, 8 thread part (1300 or 1400).

 

Video cards and GPUs

GPUs from Intel

Spoiler

For the most part, Intel has a straight forward naming schema of:

[HD or Iris Pro] Graphics [Model Number]

Example: HD Graphics 4200, HD Graphics 530, Iris Pro Graphics 540

 

Where the model number was a four-digit number that followed the GPXX format. Since Skylake, it got reduced to a three-digit GPX format. If the GPU has no model number, it's the lowest tier.

 

Summary: Look at the first three digits of the model number. The higher it is, the better. If it has no model number, it's one of the worst versions.

GPUs from AMD

Spoiler

For the purposes of this topic, that is only discussing names since about 2010, AMD had two naming schema. The first one was:

HD [Model Number] [Modifier]

Examples: HD 4870 X2, HD 6850, HD 7970 GHz Edition

 

This naming schema follows the GPXX format. It's also relatively easy since a higher number means better, every time. Unless it has a modifier like the "GHz edition", then it's usually better than the base model.

 

After the HD 8000 series, AMD went to a a little more complicated naming scheme.

R[Number] [Model Number/Other] [Modifier]

Examples: R7 290, R9 380X, R9 Fury

 

R[Number], later just RX was supposed to be an indicator to give you a quick glance at what it's performance tier is supposed to be. The trouble is, you can figure that out from the model number and that AMD has broken this pattern a few times (There's an R9 370 and an R7 370, but the R7 370 is the better performer). Since this probably was confusing and/or redundant, AMD seems to be sticking with just RX.

 

Model Number follows the GPX format. Special mention though to the "Fury" model, which is the highest tier. For now.

 

Modifier is whatever AMD thinks is appropriate to add here. Most model numbers in AMD's GPUs have a version with an "X" at the end. Some have one-off words like "Pro Duo" and "Nano". The "X" suffix though means it's a better version of the base model card.

 

Summary: Look at the model number, if it's higher, it's generally better. If it has an X at the end, it's better than the one without an X, but not a model number higher. It's better to think of an X as an "in between" card.

 

GPUs from NVIDIA

Spoiler

NVIDIA has been following this naming schema, more or less:

[Performance Prefix] [Model Number/Other] [Modifier]

Examples: GTS 450, GTX 560 Ti, GTX 1080, GTX Titan

 

Performance Prefix was supposed to give a "quick glance at what performance tier this part is in" much like what the other companies what to do with their parts. Except after the GeForce 500 series, every model you could buy was a GTX card. As of the GeForce 10 series, only one "GT" card exists in the lineup, and it's for the lowest end.

 

Model Number followed the GPX format, then the GGPX format start with the GeForce 10 series. The highest tier cards however are simply Titan (e.g., GeForce GTX Titan Black).

 

Modifier is whatever NVIDIA thinks is appropriate to add here. NVIDIA seems to not like model numbers not divisible by 10, so if they have a GPU that's between two models, they usually stick a "Ti" on at the end.

 

Summary: Look at the model number. If it's higher, it's generally better. If it has a "Ti" at the end, it's better than the one without a "Ti".

 

As a note, NVIDIA's cards following the GPX number format were called the "GeForce [G]00 series", where [G] was the generation number (so GeForce 200, GeForce 400, etc.) After the GeForce 900, NVIDIA started using the "GeForce [GG] series", so GeForce 10 for all the 1000 series cards.

 

What Each "Series" Manufacturers Use Mean

Spoiler

Now onto video card manufacturers, who add to the confusion of model naming by doing their own thing as well. A few things of note before I begin:

  • Most manufacturers will include the memory size in the model name, like an MSI Gaming X RX 480 8G vs. an Gaming X RX 480 4G.
  • The comparison at the end describes the card's potential. Most manufacturers have a series with the following aspects:
    • Stock/reference design. May have a version with a factory overclock.
    • Reference board design with a custom cooler.
    • Reference board design with a custom cooler and a factory overclock
    • Enhanced board design with a custom cooler and factory overclocked

If you do not plan on tinkering/tweaking the card, do not buy a card from a series with an enhanced board design. It doesn't really buy you any more performance for the cost.

 

EVGA:

  • Note that EVGA seems to have tacked on "Gaming" to the end of every series as of late.
  • EVGA does not call the reference design cards anything special.
  • The "ACX" series is using EVGA's custom heatsink design that uses two radial fans. This was later replaced by the "iCX" design.
  • The "SC" series means the card is factory overclocked with a reference board design. It can be combined with the ACX/iCX series.
  • The "FTW" series has a more robust power delivery design. The "FTW DT" series are not factory overclocked.
  • The "Classified" series has even more options for power delivery and is meant for serious overclockers.
  • The "Hybrid" and "Hydro Copper" series are geared for water cooling. The "Hybrid" series using an AIO, while the "Hydro Copper" includes a pre-installed water block.
  • In short: Stock < ACX/iCX = SC < FTW DT = ACX/iCX SC < FTW < Classified. The Hybrid and Hydro Cooper series are in its own category.

MSI:

  • The "Aero" series is the reference board version. It has an "OC" version that includes factory overclocking.
  • The "Armor" series provides a custom cooling solution with two radial fans. They have an "OC" version that also includes factory overclocking.
  • The "Gaming" series provides a beefier cooling solution than the Armor series, and also includes other premium features such as a cooling plate, higher quality power delivery components for overclocking, and usually come with a factory overclock. The "Gaming X" version usually has a better overclock.
  • The "Duke" series provides an even beefier cooling solution with three radial fans. It also appears to be built with structural integrity in mind. They're also factory overclocked by quite a bit.
  • The "Sea Hawk" series are geared towards water cooling, either by providing an AIO or a pre-installed water block.
  • In short: Stock < Aero < Armor < Gaming < Duke. Sea Hawk is in its own category.

Gigabyte:

  • Gigabyte's cooling system seems to revolve around the two-fan or three-fan "Windforce" design.
  • The "Turbo" series is based on the reference board design.
  • The "G1" series is the lesser premium series, featuring the two-fan Windforce cooler. They're usually factory overclocked.
  • The "Xtreme" series is the greater premium series, featuring the three-fan Windofrce cooler. Also usually factory overclocked.
  • They also have a water cooling series called "Waterforce", with the regular model containing an AIO water cooler and the "Waterforce WB" containing a pre-installed water block.
  • In short: Stock < Turbo < G1 < Xtreme. Waterforce is in its own category.

ASUS:

  • The "Turbo" series is based on the reference board design.
  • The "Phoenix" series appears to be reference board design with a better fan.
  • The "Dual Series" uses a custom cooling solution with two radial fans.
  • The "STRIX" series uses a beefier cooling solution than the Dual series and with three fans. It also has a more robust power delivery for overclocking.
  • The "ROG" series is the same as the STRIX series, just with more flare and possibly using premium power delivery components for better overclocking.
  • In short: Stock < Turbo = Phoenix < Dual < STRIX < ROG

Motherboard Chipsets

Chipsets from Intel

Spoiler

Intel has a less complicated naming system for their chipsets than their processors, at least from the ones the consumer can buy. It follows the following convention:

[Letter][Model Number]

Example: H81, Z270, B150, X99

 

Letters indicate the intended feature set the chipset has.

  • The H and Z letters are the consumer lineup. The Z series being more for enthusiasts and gamers to include features like overclocking and SLI/Crossfire support while the H series is more for everyone else and may lack those features.
  • The B and Q letters are for businesses, supporting some features like Intel's SIPP and RST. There's no real difference otherwise, and the B/Q series are pretty much the same as the H series.
  • The X letter is for the "prosumer," offering a lot of options for high performance expansion.

Model number is just a way to indicate what peripheral options it has. It follows the GP or GP0 format. While the number of features changes per generation, the following usually applies:

  • Tier 1 (e.g., H110, H71): Contains the fewest amount of peripheral options, excludes high-end options like M.2 NVME connective as well.
  • Tier 5 (e.g., B150, Q95): This tier appears on the Q and B series. It mostly affects which Intel business feature is available. It also doesn't have any or few high-end options.
  • Tier 7 (e.g., Q97, H170): This is the top tier for business/consumer end.
  • Tier 9 (e.g., X99): This is only for the X series chipset.

For most builders, tier 7 is what's often chosen. Tier 5 and Tier 1 are chosen for budge builds.

Chipsets from AMD

Spoiler

For the FX series processors, your choices of chipset (or Northbridge, rather) is pretty small:

  • 890 FX
  • 970
  • 990X
  • 990 FX

The feature set differences are minimal, aside from how they support multi-card configurations. The 990X and 990 FX support SLI.

 

For socket FM1/FM2/FM2+/AM1, AMD opted to use an A followed by a two digit number, then a suffix letter, if they felt the need for it. Generally speaking for these sockets, the higher number means better features offered.

 

For socket AM4, AMD has opted to use the [Letter][Model number] scheme similar to Intel. Since there's not a lot of variation on the model numbers for each letter (yet), this will only cover the letters:

  • X: The "enthusiast" tier chipset, with the most expansion options and features. Notably it offers SLI support.
  • B: The "mainstream" tier. Fewer options like not supporting SLI, but still supporting Crossfire.
  • A: The "entry level" tier. Does not support multi-card configurations and overclocking.

Memory (RAM)

Spoiler

Memory models are given by the module's speed. There are two ways to represent this:

  • The clock speed, given by the format DDR[Generation number]-[Number]. The generation number is excluded for the first generation of DDR memory. Examples include DDR-400, DDR2-800, DDR3-1600, and DDR4-3000. This is more of a specific type of memory as DDR refers to DDR SDRAM, as opposed to RDRAM or EDORAM.
  • The data rate, given by the format PC[Generation number]-[Number]. The generation number is excluded for the first generation. Examples include PC-3200, PC2-6400, and PC4-24000. This value is more generic and was used regardless of RAM type.

The values mean the same thing performance wise and can be converted back and forth. So you will see something like DDR4-3000 / PC4-24000. What you will not see is something like DDR4-3000 / PC4-30000. That is to say, DDR4-3000 = PC4-24000 as far as specs are concerned. If someone tries to label DDR4-3000 as PC4-30000, they have no idea what they're talking about.

 

You may see latency as well advertised with the memory kit. This is usually referring one type of memory latency called CAS latency. Generally speaking, the lower this number the better. However, this is given in clock cycles. So while a DDR4-2133 module may have a CAS latency of 15 and a DDR-400 module with a CAS latency of 4, the DDR4 module is faster because 15/2133 < 4/400.

 

Other Model Name Trickery

Spoiler

Storage devices will often have its capacity in the model name itself.

 

Power supplies will often have their maximum rated wattage in the model name.

 

Wi-Fi adapters from some companies also list the maximum speed and Wi-Fi network type they can interface to.

 

Monitors appear to have a fortunate de facto standard in model names: the size. Most monitor model names will include the diagonal size in the model name itself. For example, the Dell monitor P2517Q is a 25" monitor. ASUS's PB287Q is a 28" monitor. LG's 25UM56-P is a 25" monitor. Usually by the letters, you can also infer what features it has, but it only works within the manufacturer.

 

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×