Building a gaming PC is quite a daunting task, not least because of the huge variety of graphics cards available in the market today. Companies spend so much money on marketing their products that it often becomes next to impossible to determine whether the card you are buying is suitable for your needs. In this article, we will try to remove (at least some of) the fog that is between you and your dream PC and will give you a complete GPU hierarchy to buy your next graphics card.
We will cover everything from entry-level GPUs to enthusiast and professional GPUs, illuminating upon their main characteristics, use cases and examples. We will also make a holistic comparison of their important parameters like performance and power consumption. We will build upon these parameters further with other factors such as pricing that you need to consider when buying your GPU.
Finally, we will talk about some of the ongoing trends and advancements going on within the graphics card market. Topics such as upcoming architectures and the role of artificial intelligence will also be discussed. There’s a lot to cover so let’s get started with the very basics.
What is a Graphics Card?
In simple terms, a GPU is a device largely responsible for rendering graphics in your computer. While some simple graphics workloads can be handled by the CPU alone, a GPU becomes crucial when you intend to do any demanding task like gaming or video editing.
Often, the terms GPU and graphics card are used interchangeably. However, there is a slight distinction between the two in that the graphics card is the chip housed inside the GPU.
GPUs come in many shapes and sizes, both in terms of performance and use case, both of which determine each other. The two main categories are gaming graphics cards and workstation cards. In the former, the number of cores is the main determining factor of performance while in the latter, it is the memory size.
This is because in gaming, rendering new environments is more important than storing existing ones, whereas, in workstation cards, huge drawings such as protein structures need to be generated and stored for quick access for quite a long time. This is the reason why even relatively older workstation cards had dozens of gigabytes of VRAM despite modest processors.
GPU Hierarchy Chart
GPU hierarchy is important for multiple reasons. The most prominent reason is that there is no objectively best graphics card. There are always trade-offs involved in terms of price, performance, stability, and so on.
For those for whom price is an important constraint, the graphics card with the highest performance-price ratio, such as 60FPS at 1080p settings, might be the best while for someone whose budget is practically unlimited, getting the absolute maximum performance such as ridiculously short video editing times for 8K videos may be the most important factor. In such a situation, even the fastest available graphics card might fall short of their expectations.
Therefore, a wide array of customers need to be taken into consideration to provide meaningful comparisons between graphics cards. The ultimate outcome of this exercise manifests itself in the form of a GPU hierarchy.
Entry-level GPU Hierarchy
Now that you know a little bit about the important determining factors of a GPU hierarchy, let’s shift our attention towards entry-level graphics cards, their main characteristics, suitable use and examples.
Description and characteristics
If you want to do slightly more than what the bare minimum integrated graphics cards have to offer, then this is the category of cards you should focus on. These graphics cards are quite cheap, priced between $25 and $100, have a very limited number of cores and VRAM, and as a result, often have the highest performance-price ratios.
On the bright side, however, this also means that not only will you save money on the graphics card’s price, but you will also save it on your electricity bill as these cards consume barely more than a few dozen watts of power. For example, the Nvidia GT 710 comes at a price of only $50 and consumes a maximum power of 20W.
All this means, that alongside the pricing, your expectations from this card too should be modest as we look into their use cases.
Typical use cases
As mentioned earlier, entry-level graphics cards have very limited overall performance. You can use them for basic graphics work like photo editing, retro gaming, rendering CAD drawings, and so on. If you want to do anything more than that, such as video editing or playing modern AAA titles, then you are going to be severely disappointed with these cards, unless of course, you have an infinite amount of time to kill, anyways, or enjoy gaming at 360p resolutions at low settings.
Examples of Entry-Level GPUs
Sarcasm aside, here are a few entry-level graphics cards that are quite popular in the entry-level graphics card market and might be worth considering. Keep in mind that the most important purpose of these cards is to ensure your PC is functional as some modern processors sometimes don’t come with integrated graphics. If your processor comes with integrated graphics, spend these fifty bucks somewhere else, such as RAM or a better processor.
PowerColor AMD Radeon 240: Among entry-level cards this is the one that has one of the highest absolute and relative performance in terms of price. It comes in at around $35, features 4GB of GDDR5 VRAM, and should work just fine if you intend to watch high-definition movies, play gaming titles older than 2010 and do very basic graphics work.
MSI GAMING GeForce GT 710 1GB: Next comes the MSI Gaming GeForce GT 710 1GB. Don’t buy this card unless the Radeon 240 is out of stock. It is more expensive at $50 and offers poorer specs such as fewer cores, lower VRAM, and only slightly higher clock speeds. The use case of both cards is similar.
MSI Geforce 210: Lastly, if you can’t use either of the previous ones in your build, then you should consider going with the GeForce 210. It comes in at $45, features 1GB of VRAM and should be adequate for basic computing tasks. Keep in mind that this card is not compatible with Windows 11. So, don’t buy it if you plan to upgrade your OS any time soon.
Mid-Range GPU Hierarchy
Next comes the mid-range category, the one that has something for just about everyone. With the extreme surge in graphics card pricing in recent times, it has become a bit difficult to determine the upper and lower limits of what should be considered midrange. However, generally speaking, graphics cards priced between $100 and $300 are considered mid-range graphics.
Description and characteristics
As their name can imply fairly well, these graphics cards seek to provide a balanced mix of performance, pricing, power consumption and so on. Pricing of these cards is quite competitive as this is the market segment within which the majority of the consumers fall. These graphics cards often feature decent core count and clock speeds, and 4-8GB of VRAM which makes them suitable for a wide variety of tasks such as rendering CAD, video editing, gaming, and so on.
The power consumption of these cards also stays within reasonable limits. These cards generally consume between 120-200W at their max performance. This also means savings on your PSU component as PSUs for these graphics cards also tend to be reasonably priced.
Typical use cases
These cards are suitable for quite a diverse array of use cases, the most prominent one being gaming. You can expect to play most modern AAA titles at 1080p or 1440p resolution at medium to high settings at decent framerates in most graphics cards within this category. Of course, you would be asking for it if you want to play horribly badly optimized games like CyberPunk 2077 at high settings on these cards.
If you are a content creator, then to these graphics cards can save you time when editing high-definition videos without breaking the bank. If your videos are recorded at 1080p, then these cards are more than adequate for your performance needs. However, for 1440p or 4K videos, you might need to slightly adjust your performance settings and expectations.
Similarly, these cards can also be more than sufficient if you are an engineer or work in a lab where you need to render large design files, floor plans of a building, links between different components in a car, or electrical schematics.
Examples of mid-range GPUs
Considering the huge size of the mid-range GPU market, it is a bit difficult to shortlist the best budget graphics cards as the competition is quite literally neck-in-neck. That being said, here are a few mid-rangers that we recommend checking out for your next PC build.
Radeon RX 6600: The Radeon RX 6600 can be had for $220 as of this writing and is one of the best budget graphics cards that you can buy today for 1080p gaming at ultra settings. It will struggle with 1440p and 4K unless you lower some of the settings. At 135W, power consumption too is quite reasonable.
Overall, this is a great all-purpose graphics card that can easily meet most of your video editing needs too if you are a content creator. What’s more, the card’s 8GB of GDDR6 VRAM means that it is going to handle your modelling workloads without breaking any sweat either. Highly recommended for budget builders!
Intel Arc A750: Next comes a card from a new entrant into the graphics card market. At $240, the Intel Arc A750 is only slightly more expensive than the RX 6600 but offers noticeable performance improvements over the RX 6600. Overall, the performance of the two is similar in most use cases.
However, one of the drawbacks of the A750 is its higher power consumption. At 185W, it is noticeably higher than the power consumed by the RX 6600. So, unless you want to give Intel cards a shot, you might be better off going with the RX 6600.
RTX 3060: Last, but not least, we have the RTX 3060. At $289, this is the most expensive mid-range graphics card on our list. Despite the higher performance, gaming benchmarks are similar to those of the other two and even slightly poorer than the RX 6600 at 1080p settings. Power consumption is also a bit on the higher side.
However, the distinguishing aspect of this card is its raytracing capability. This is an area where Nvidia leaves the competition in the dust. If you really are serious about ray tracing, then this single-handedly is one of the best budget graphics. The card can handle 1080 resolution with RT enabled. However, there is a serious framerate drop if you choose to play at higher settings. So, you might want to keep that in mind.
High-End GPU Hierarchy
Now let’s jump into the category of cards which offer the highest absolute performance, catered to meet the needs of the most demanding workloads, gaming and professional. Generally speaking, these cards are targeted towards enthusiasts who merely want performance for the sake of it, rather than out of necessity. Let’s dive into them a bit more deeply.
Description and characteristics
The most defining characteristic of these cards is almost everything about these cards is maxed be it specifications, be it power consumption, be it price. The core count and clock speeds of these cards are much higher than those available in midrange cards. Memory size too often ranges between 10-24GB, compared to 4-8GB in mid-range graphics cards. The result: extremely high frame rates at ultra settings and almost instantaneous loading of games.
Since these cards are targeted towards enthusiasts, the cooling solutions employed in these cards are also quite heavy-duty, so to speak, to meet the overclocking needs of such users. Overclocking, however, often is risky and doesn’t provide significant performance improvements. So, it should be avoided, if possible.
Typical use cases
High-end graphics cards often find themselves employed in a diverse array of situations. The most commonly known use is in of course gaming and content creation. With these cards, you can easily play the latest AAA titles at 4K max settings at 60FPS+. Video editing times are also significantly lower especially if you record at 4K or higher.
Examples of high-end GPUs
Below we list some of the best high-end gaming GPUs in the market available. We have considered both absolute performance and value for money in coming up with these recommendations.
RTX 4090: The RTX 4090 is the most powerful graphics card available in the consumer market today. Unsurprisingly, it is also the most expensive card at an MSRP of $1600. Add in shortages due to supply chain issues, and you should find yourself spending another $500 bucks on this card. However, once you have managed to get your hands on this card, the possibilities are only limited by your imagination. This card should be able to chew away whatever you throw at it, be it the latest games like CyberPunk 2077 at 4K ultra-high settings, or large 4K raw files that you need to edit before uploading them to YouTube.
RX 7900 XTX: Next in line is the flagship card of Nvidia’s rival in the GPU market since time immemorial: AMD. The 7900 XTX comes at a significantly lower MSRP of $1000. On its own, the card is quite powerful and offers a similar performance-price ratio as the RTX 4090. However, in relative terms, there is a noticeable difference between frame rates no matter what setting you choose. For example, at 1080p, the RX 7900 XTX maintains an average framerate of around 140 FPS compared to 180 FPS by the RTX 4090 in AAA titles. Ray Tracing is an area where AMD cards lag behind Nvidia, generally. So, if raytracing is important for you, then go with an RTX series card.
RTX 4070: The RTX 4070 falls on the more reasonable side of things when it comes to high-end graphics cards. It comes at a price of around $600. The card matches in performance with the RTX 3080 and comes at a slightly lower MSRP compared to the $700 at which the RTX 3080 came.
We could have mentioned the AMD 6950 XT here too, but both cards offer similar performance but the 4070 consumes noticeably lower power. Plus, its ray tracing capabilities are ahead of its rival card. Overall, this card is going to meet your needs just fine for most games if you don’t push too far beyond 1440p resolution.
Professional GPU Hierarchy
Moving beyond gaming and content creation, we have professional GPUs. These are also known as workstation cards. While both consumer-grade and professional GPUs can be used to do the same task, both are optimized for specific tasks. Generally, these cards are reserved for enterprise applications, are produced in fewer quantities and have much stricter quality control measures. As a result, they are much more expensive than consumer-grade GPUs. Read on to get a more in-depth understanding of professional GPUs, their natural uses and examples.
Description and characteristics
In these cards, certain features are more strongly optimized than others. This means that while these cards may have fewer compute units than consumer-grade cards, they often have much higher memory. Even in relatively older workstation cards, a memory size of 8GB or higher was not uncommon. This expansive memory size is important to ensure rapid access to different portions of already rendered design files which may take quite a long to be rendered again.
While parallel processing is largely dead in the consumer market, professional GPUs still make extensive use of parallel processing to reduce computation times in tasks like deep learning and simulations.
Typical use cases
Professional GPUs are generally reserved for scientific and engineering tasks. These tasks are accomplished on top of software manufactured by companies like AutoCAD, Adobe, MathWorks etc. These companies work closely with GPU makers, mainly AMD and Nvidia, to optimize their cards for the applications produced by these companies.
The areas where these cards find applications are almost endless. These range from construction work, computation, physics simulations, biology, and so on. For example, for a construction company, almost all designs are rendered beforehand inside the CAD software to get an overall picture of the expected end result to ensure there is as little deviation between the practical and the theoretical.
In mathematics, these cards are often used to test whether certain huge numbers, spanning thousands of digits, are prime or not. This has important implications for our internet security as prime numbers are the bread and butter of cryptography.
Similarly, weapons manufacturers may use these cards for software that is used to test the impact of a projectile, the deviations in its pathway caused by air. As the requirements for exactness get tighter, so do the computational loads. This is where these cards come to the rescue and save time
Examples of professional GPUs
Now, let’s have a look at some of the examples of professional GPUs that are quite popular in enterprise applications. While AMD has been making strides in the workstation GPU market, Nvidia still holds the majority of the market in this area. Therefore, we are only going to be including Nvidia cards in this list, starting from the most basic to the most powerful.
NVIDIA Quadro P4000: The Nvidia Quadro P4000 is a highly popular graphics card in the entry-level workstation GPU market. It comes at a price of ~$350 and features 1792 CUDA cores and 8GB of VRAM. It is highly suitable for small to medium-sized modelling workloads and features support for multiple 5K displays.
So, if your tasks are divided up into multiple windows, and there is not enough real estate on your screen, you can add multiple displays, thanks to the capabilities that the P4000 offers.
Nvidia Quadro RTX 4000: Moving up a bit, we have the RTX 4000 which costs almost exactly twice as much as the P4000 at $700. While it features the same 8GB of RAM as the Quadro P4000, one of the distinguishing features of this card that Nvidia highly touted is that it is the world’s first workstation GPU that comes with support for ray tracing applications.
Traditionally known for enabling surrealistic gaming environments, ray tracing also has multiple scientific applications such as realistic depictions of fluid dynamics simulations.
Nvidia Quadro RTX A6000: Coming in at a price of around $2400, the RTX 6000 is the most expensive and powerful workstation GPU. It is newer than previous models. However, we included them because they are still quite popular in the market. The A6000 features a monstrous 48GB of GDDR6 memory.
The card features 4 DisplayPorts and supports a maximum resolution of 8K which means you hook up multiple 4K (or lower resolution) displays. Keep in mind that this card is not suitable for gaming even though it might play them reasonably well. Enterprise users are the main market that this card is geared towards such as companies that specialize in AI applications, training AI models on large datasets.
GPU Hierarchy Comparison
Now let’s have a look at the top three GPUs in different categories like absolute performance, value for money, and power consumption to determine the most suitable GPU for your needs. In terms of pure performance, Nvidia easily takes up all the top slots, both in gaming and content creation as we will see below.
Performance metrics and benchmarks
In terms of absolute performance, the RTX 4090 tops the list of GPUs. In Blender, the RTX 4090 achieves a mean score of 3870 which is the highest of any card. Similarly, in the gaming arena, the RTX 4090 can maintain a stable 120FPS at 4K settings in CyberPunk 2077.
Next in line is Nvidia RTX 4080 which has a mean score of 2950 on Blender, significantly lower than its immediate superior. In gaming too, the framerate drops to 100 FPS when similar settings are used in CyberPunk 2077. This is still more than playable, of course.
The third position is, of course, the RTX 4070 Ti. In Blender, the 4070 Ti has a benchmark score of 2300, which again is a significant drop from the RTX 4080. In terms of gaming performance, the RTX 4070 Ti again takes a drop in framerate falling to around 85 FPS.
Price-to-performance ratio
In the realm of value for money, AMD is often ahead of Nvidia, especially in the upper midrange category.
If you plan on gaming at 1080p or 1440p, then you need to go no further than the AMD RX 6600. At $200, it is the cheapest card that you can buy to comfortably play games at 1080p high settings. The card can run even the latest AAA titles like CyberPunk 2077 while maintaining an average of 78 FPS. At 1440p high settings, the frame rate will drop to around 51 FPS, which though not quite as smooth, is still playable. Only when you push to the 4K arena does the framerate notice a noticeable drop unless you lower the other graphics settings, which kinda defeats the purpose of playing at 4K in the first place?
However, if you are looking for the cheapest graphics card that can allow you to organically play at 4K, high settings with ray tracing enabled, then your best bet is the RTX 4070. At ~$600, it is the cheapest card that you can buy that will enable you to play your favourite titles at ultra-high settings. In CyberPunk 2077, at these presets, the card can manage to pull ~40 FPS or higher consistently. If you are used to playing at 60FPS or higher, then 40 FPS may sound like a downgrade, but it doesn’t significantly affect the overall gaming experience.
Power consumption and TDP
While it is desirable to have a card that consumes as little power as possible, at some point we have to become realistic with our expectations and make some compromises depending upon our gaming preferences. With that in mind, here are some of the most power-efficient cards available.
For 1080p gaming, both RX 6600 XT and the RTX 3060 consume the lowest power at 160W and 170W respectively while offering an excellent gaming experience. If you upgrade to the RTX 4070, even then, the increase in power draw is not significantly higher as this card too consumes around 200W at max.
It is only when you climb further and further in the hierarchy that power consumption becomes a significant factor. For example, RTX 4090, the fastest card available today in the market has an average power draw of a whopping 450W. This is the baseline power draw. Sometimes, it might draw even more power. This means that investing in a robust, heavy-duty PSU is crucial if you choose to go with the RTX 4090.
At such high power draws, thermals also become a significant factor as you need to continuously pump heat out of the tower and draw in cooler air, which means you will need a spacious CPU case and multiple system fans to keep things manageable.
GPU Hierarchy Ranking Factors
While there can be infinitely many factors that can come into play when creating a hierarchy of GPUs, the most prominent ones have to do with the specifications of the cards themselves. These include core count, clock speed, memory size, and power efficiency. The model and architecture of the graphics card largely determine these factors.
For instance, graphics cards built on newer architectures often are more power efficient than their predecessors, resulting in greater overall performance. For example, RTX 3080 is faster than RTX 2080 merely due to its power efficiency. Within a given architecture, it is the model that determines overall performance. For example, the RTX 4090 is faster than RTX 4060 largely due to higher core count and clock speeds.
Sometimes, power efficiency improvements become so significant after a few architectures that a midrange card of a newer generation may perform faster than the best card of an older generation. For example, RTX 4060 vs GTX 1080.
