How to Choose Your Parts, Part 2: The Video Card (GPU) — Micro Center

How to Choose Your Parts, Part 2: The Video Card (GPU)

TSTonyVTSTonyV admin
edited August 7 in Graphics Cards

If you haven't read part one yet, click here!

Greetings! Welcome to the Micro Center Community. This is the second in a series of posts I’m making to help you become better informed about the current market of parts for those of you wanting to build a PC. Whether you’re a first-time builder or an experienced one, it can be difficult to keep track of all the different information available so my hope is this will help you figure out what fits your budget and needs best. This post will be focused the Video Card, AKA GPU (Graphics Processing Unit).

First things first (again), we need to answer some basic questions:

  • What is your budget? The first question you should ask, as this determines what type of price and performance you can look at. The GPU is usually the most expensive part in the build, so keep that in mind when choosing which one you want.
  • What kind of games are you playing? Different games have different graphics requirements. Games like Fortnite and League of Legends are less GPU intensive, so you don’t need a super top-tier GPU to get good quality. Games like the Witcher 3 and Shadow of the Tomb Raider, on the other hand, will require a lot of GPU power if you want to run at higher settings.
  • What is your intended resolution/framerate? This goes together with the previous question. If your goal is to play games at 1080p/60 FPS, even the budget cards will be able to run most titles without a problem. You’ll need to lower your settings in the more GPU intensive games, but they’ll get the job done. If you want higher quality in more demanding titles you have to look at more powerful GPUs.

And once again, we have an all-important question:

NVIDIA or AMD?

AMD makes their return in another Great Debate, this time duking it out vs. Nvidia. There was a time when NVIDIA and AMD video cards (formerly ATI video cards) would trade spots with both value and performance, but eventually Nvidia solidified itself as the premier video card maker and until very recently AMD was only competitive in the budget category. AMD is still a strong budget contender, and now they’re starting to fight their way back in the mid and low-high end categories with some very compelling options.

With CPUs, there’s big differences in single threaded performance and the number of cores/threads available, which makes direct comparisons sometimes somewhat difficult depending on your use case. GPUs are a little easier to compare: what gives me better framerate at a given resolution? That’s what we’ll be focusing on

I also want to make one note about features: NVIDIA’s new RTX cards have one major feature that AMD cards do not have: Real-time Ray Tracing using RT Cores. The basic overview of Ray Tracing is that it’s realistic physics-based lighting calculated in real-time and allows for things like real shadows and reflections. Instead of a reflection being drawn in and modelled by an artist or a shadow being a pre-loaded texture attached to a model, Ray Tracing allows all those to be calculated realistically in real-time based on light sources present in the game. Before Ray Tracing was something typically reserved for things like movie CGI and rendering. It’s now starting to make its way into gaming, but it’s still new technology and most games don’t support it. For the ones that do, turning RTX on has a BIG performance hit (like, cut your FPS in half or worse). That said, if it’s a feature you want then NVIDIA is really the way to go.

For those of you who are workstation oriented and are doing graphics-accelerated work, NVIDIA and AMD both manufacture cards specifically for workstations, Micro Center just doesn’t really carry them (with one exception I’ll mention later). That said, for most cases you’ll find that each card’s performance in gaming will translate somewhat to production, relatively speaking. I.E. the more high-powered it is as a gaming card, the better it will usually be for workstation applications. That’s not always true depending on your specific needs, and even the top-tier gaming cards can perform very poorly compared to proper workstation cards in certain cases, but it’s a good guideline to go with. For a better breakdown of how video cards line up for production work, I would watch this video by Gamers Nexus: https://youtu.be/2JmWrO2bcsQ.

Finally, I do need to make a note about driver performance and optimization. The new AMD cards have finally started to put out some compelling performance numbers and are competing better with NVIDIA in non-budget categories, but AMD has unfortunately developed a bit of reputation for having poor driver stability on these newer cards. Most people won’t have issues, but at least for now, if you want the safe, more reliable bet for drivers, NVIDIA still has the edge. Along with that, games are all optimized differently. Some titles are very favorable for the AMD cards, some are favorable for NVIDIA cards. The relative performance of each card can be affected a lot by how well optimized the game is for each brand. In some cases cards from a lower tier in one brand can compete against cards from a higher tier in other, if the title is favorable. Keep this in mind when choosing a card as well. 

How do I decide between different models of the same card?

If you’ve looked up video cards on your own at all, you probably noticed how different manufacturers will have their own versions of each card, sometimes multiple ones. If you’re not sure how that works, or why they do it, it’s pretty simple:

AMD and NVIDIA design the GPUs, and manufacturers tweak them to come up with the different models you see on sale. The primary differences you’ll see will be the cooler design and clock speeds for the GPU itself. Some cards will include RGB features or may have different video ports on the back. All those things together will contribute to the price differences across the same type of card.

If you see a video card labelled “overclocked” all that means is that it’s clocked above the base specification AMD or NVIDIA provided to the manufacturers. Depending on how they overclock the card, it may feature a larger heatsink and more fans to increase cooling performance to match. And if you’re concerned that those cards could be unstable because you’ve heard overclocking can cause issues, don’t worry. Manufacturers do extensive testing to ensure the cards are stable at the settings they ship with.

Some of those price differences may be pretty large. A 2060 Super, for example, could be anywhere from about $360 all the way up to around $460 depending on the specific model you choose. A $100 swing is pretty big, so surely the performance changes to reflect that, right? Unfortunately, no. At worst, the least expensive version of a given card will only be a few percent slower than the most expensive. What you’re usually paying for on the more expensive models is better cooling, quieter operation, and/or aesthetics.

How to match a GPU to a Processor, and what it means to have a bottleneck:

Ideally, if you’re reading this, you’ve already gone through my previous post about choosing your processor. Now that we’re looking at the GPU, we need to start considering how our parts will match up with each other. 

There’s a concept known as “bottlenecking” where if you match a high-end part with a low-end part, you’re going to be held back by whatever the slowest part is.

Say, for example you have a Ryzen 3 3200G and you’re using it with an AMD RX 570 video card. You probably have okay to decent performance in most of the games you play, but you decide to make a huge upgrade and jump all the way up to an NVIDIA RTX 2080ti for your video card while keeping that 3200G. While you will certainly see a performance increase, if you stick with that 3200G you’ll never see the type of performance a 2080ti is capable of. It just doesn’t have the juice to take advantage of all that graphics power. 

Conversely, if you have an Intel i5-8600k with a GTX 1060, upgrading to a 9700k will certainly give you some performance gains, mostly in your FPS. But if you also decide to crank up your resolution from 1080p to 4k, you’ll find it won’t be playable because the GPU is just not designed for that kind of resolution.

It’s a delicate balancing act that involves a lot of factors. However, the important thing to note here is with the current crop of CPUs on the market currently offering more cores, threads, and better single-core performance than ever before, even mid-range CPUs like the Ryzen 5 3600 and i5-9600k will not be bottlenecks except for high end cards and if you’re pushing really high framerates.

Games these days tend to be more GPU bound than CPU bound, so when planning your build you need to keep this in mind. That’s the big reason why I had strong recommendations for all the AMD CPUs in the processor guide: what you save on CPU could let you jump up to the next tier of GPU. An i5-9600k + 2060 Super might be better in gaming than a Ryzen 5 3600 + 2060 Super. But if your budget allows the choice of a Ryzen 5 3600 + 2070 Super instead, that will always be the better choice because of the GPU difference.

Now that we’ve got all that stuff out of the way, let’s talk about the video cards themselves. 

Entry Level

These cards will be good for 1080p gaming at a low cost. You may have to lower the settings in certain games to get 60 FPS, but they’ll get the job done.  

NVIDIA GTX 1650 Super: The 1650 Super ranges from about $160-180 and is probably the best overall option in this price category. In terms of performance it competes directly with the AMD RX580 and RX 5500XT and will typically be within a few percent. Practically speaking, the performance is essentially the same. The regular 1650 has been completely invalidated by this card.

AMD RX 5500XT 4GB: AMD’s new budget card and effectively the replacement for the previous RX580/590. Ranging from $180-200 it’s a little more expensive than the 1650 Super. There’s a 5500XT model with 8GB of video RAM, but it usually has no effect on gaming performance and costs closer to $220 so it’s generally not worth it except specific scenarios you’d need more VRAM for.  

AMD RX580: The RX580 has been around for about three years now. It can usually be had for about $170-180, so it’s a little bit more compelling compared to the 5500XT. However, it’s a little harder to find.

AMD RX570: Previously the King of the Budget Card. It is weaker than the RX580 and 1650 Super, but if you can find them they’re usually under $150 or even $140 these days. If your budget is that tight, this can be a consideration, but since a 1650 Super can be had for as low as $160 I’d say save up a bit and go for that instead. Like the RX 580 this is getting harder to find.

Low Range

This is another category of card that will do well at 1080p, with the benefit of running more graphically intense titles a little better so you won’t have to lower settings as much. Lighter titles you can probably start running at 1440p.

NVIDIA GTX 1660 Super: The 1660 Super will run from $230-260, depending on which specific model you go for. Like the normal 1650, the normal 1660 has been invalidated by the Super version considering the nearly identical price points and performance boost.

NVIDIA GTX 1660ti: The 1660ti is around $280-300 and sees a small performance bump over the 1660 Super. However, now that you’re pushing up to that $300 price point, it’s starting to get into pricing territory with the next tier of GPUs so the value becomes less compelling.

Mid-Range

Now we’re starting to play with the big boys. The cards in this category will run pretty much any game at 1080p/60FPS at high settings with no issue; 1440p/60FPS becomes accessible with some compromise on settings depending on the card and title. 4k gaming may be possible on titles with light graphics requirements, but in general 4k will be very limited or require massive compromise on graphical quality settings to run at playable framerates. This is also where you’ll get into NVIDIA’s RTX cards if you want Ray Tracing.

AMD RX 5600XT: The 5600XT can be had from around $290-320 generally. It goes head-to-head with NVIDIA’S RTX 2060. Performance is typically the same or very close, trading the lead depending on the game.

NVIDIA RTX 2060: The RTX 2060 can range anywhere from $300-400 depending on the model you get. If you can get it at that near $300 price point, like with the EVGA RTX 2060 KO, then it’s worth the price and competes well with the 5600XT. Once you’re past that, you’re getting into RX 5700 and 2060 Super territory which are better performers.

AMD RX 5700: The RX5700 is found from about $340-380, with some models getting up to around $400. It competes closely with NVIDIA’s 2060 Super, generally being a small amount behind but usually offering better performance-per-dollar with the lower cost.

NVIDIA RTX 2060 Super: Another Super card, this is the updated version of the normal 2060, starting around $360 and going all the way up to around $460 depending on the model. It does generally pull ahead of the RX 5700 in performance but costs a little more. Like the regular 2060, if you look at the more expensive versions you’re getting into pricing territory for a more powerful 5700 XT so I’d stick with models closer to the lower end of its price range.  

AMD RX 5700XT: Ranging from $400-460, it sees a solid performance bump over the 5700 and is AMD’s most powerful gaming card available right now. This is the top of the “mid-range” cards and starts to break into high-tier performance. It will generally beat the 2060 Super and in the right titles will even compete closely with the 2070 Super. Given that the 2070 Super starts about $100 more, the 5700 XT has a very strong performance value and at the price point this card really can’t be beat. 

High-End

AMD unfortunately takes a backseat as they are not offering any new cards above their 5700XT in performance. We are now firmly in NVIDIA-only territory. At this level 4K gaming becomes accessible (with compromise) and you should be able to run almost any title at 1440p/60FPS+ at high/max settings. There’s also a big price jump for each card after this point.

NVIDIA RTX 2070 Super: These cards are generally $500-600 depending on the model. 4K gaming becomes accessible at playable framerates, but you will likely have to compromise on your settings. Games that support Ray Tracing may also start to become playable.

NVIDIA RTX 2080 Super: The 2080 Super makes a price big leap starting at about $700 and going up to about $800. You’ll see about a 15% performance increase over the 2070 super in general and 4k is doable with less compromise.  

Enthusiast Tier

“Enthusiast” tier is basically just a fancy way of saying “My video card costs more than your entire computer"... but I digress. There’s only two cards that occupy this tier. First:

NVIDIA RTX 2080ti

When it comes to gaming, this card is the cream of the crop. The best of best. It is also supremely expensive and sees an even bigger price jump than the difference between the 2070 Super and 2080 Super. The 2080ti starts at a whopping $1100, and stretches as high as $1500 or more. If you want to game at 4k with good settings and decent framerates, this is where it’s at. Just make sure you have enough leftover for the rest of your computer!

The second card that occupies this tier is…

NVIDIA RTX Titan

For a measly $2500, you can have this card.

Yes, you read that right. The RTX Titan is double the price of some 2080ti cards. But for gaming, the RTX titan is only marginally better than the 2080ti, small enough that you probably wouldn’t even notice a difference in a real setting. So why is it so much more expensive? The thing is, the RTX Titan is designed for a different purpose. This is that one exception I mentioned earlier about Micro Center carrying workstation cards, as this is a true workstation card, not a gaming card. While you can game on it, and game on it extremely well, what it’s really good at are workloads that require high amounts of VRAM such as machine learning and high-resolution video editing: the RTX Titan has 24GB of VRAM, more than double the 2080ti’s 11GB. Like with the super high-end CPUs I discussed back in part one, this is a card that you should only go for if you have a very specific purpose in mind.


Choosing a GPU can be overwhelming with how many different cards and partner models are on the market, so I hope this helps you narrow it down. Feel free to comment if you have any questions, and make sure you check out part 3!

Sign In or Register to comment.