We’ll use Nvidia GeForce graphics cards as an example today. AMD and their manufacturers follow a similar structure, but there’s no sense confusing everyone. Here’s a graphic to explain model names:
So Nvidia makes the GPU. That’s the big central part of the graphics card that does the majority of the calculations — nearly all of it. The brands you know of (EVGA, Gigabyte, Asus, MSI, etc.) buy that GPU and slap it on a big card with their own design and specifications.
The component quality of each card differs from company to company. I’ll go over some favorite and top-pick companies that you can reliably expect to have available, quality products.
Aside from the obvious aesthetic differences of cards, each manufacturer uses a different cooler, and will have different sub-versions of the card. So for example, they might have 5 different version of an “Nvidia GeForce GTX 1060”
Each version will have some difference in either memory configuration, power management (typically # of power phases), coolers, clockrates, etc. Some will have different display connections (HDMI, VGA, DVI, DisplayPort).
For example, MSI has their well-known Twin FROZR design, with large sinks and large fans. They also have, for most of their cards, a standard version, a “G OC”, “GT OC”, “OCV1”, “Gaming 3G”, “Gaming X”, and ARMOR variants (white color) for each. Each has either a different fan design or different degree of overclocking.
Meanwhile someone like EVGA has their more minimalist, rectangular look and just as many power steppings. They have their ACX 2.0 cooler, the old style, the new ACX 3.0, ACX Single (one fan, half length card), FTW, FTW DT, SC, and Classified series. There’s also Black Editions which have fewer LEDs and a much darker color scheme.
These are two very high end card makers, with (usually) very high end components and a meticulous focus on advanced and quiet cooling. They pride themselves in having multiple steppings with which enthusiasts can purchase a pre-overclocked card, or cards with tested, known additional overhead for manual overclocking.
There’s also reference design cards (aka blower style) which have a singular fan on one end, a nearly air-tight design that forces the air to blow across the whole card, and then a rear vent that exhausts the air out of the case.
Nvidia makes the “Founders Edition” card which releases early and sports their high end blower cooler.
There’s different reasons to get different cards, but for the most part you should stick to the middle-ground in pricing for the cards. I know that’s super generalized, horrible advice for purchasing, but I rarely see cards that sit in the middle of the price setting that are either designed for massive manual overclocking or are total pieces of crap.
Personally, I’d recommend avoiding PNY except for budget builds. Although their products are usually OK, I’ve had a couple issues with them while having nearly no issues with anyone else. Innovision and Powercolor are even lower end brands than PNY. I’ve never even dared touch one of their cards.
A respectable mid-range card maker is Gigabyte, but be prepared for ridiculous coil whine. If you have a nice case or the computer isn’t on the desk next to you, this isn’t an issue.
My preferred high-end maker is EVGA. Their FTW DT cards are nice for anyone who doesn’t intend to overclock. You usually pay a hair less than a superclocked card, but get a nice factory boost. Do your research if you intend to overclock. Stick with the highest end cards offered by a particular maker if you want to OC.