Best CPU-GPU Combos for Gaming (1080p, 1440p, 4k) in 2020

| | ,

The most important components of a gaming PC are the graphics card and processor. But which combination is the best? We will show you the optimal configuration for your new Gaming PC below.

For the PC to develop its full performance, the graphics card and processor must be optimally matched. Anything else causes unnecessary costs: If the CPU and CPU do not harmonize with each other, this slows down the respective faster component. Especially graphic cards have been increasing in price since the beginning of the year (because of mining), that’s why a bad combination of Graphics cards and CPU is even more annoying. The ongoing popularity of virtual currencies will probably not make GPUs cheaper for the next years to come.

What are the best CPU/GPU combinations? To find out, we have brought a wide range of current CPUs and GPUs into the test lab. The processors were used in four test systems, on which we tested six current graphics cards for each GPU. This resulted in our Top 3 Ranking of the best CPU and Graphics card combination.

The comparison table shows you the results of the cross test at a glance. If you want to completely rebuild your computer, you can find out there which CPU-GPU combinations perform very well. The motto “a lot helps a lot”, but it can become an expensive mistake. The cross test also shows that the fastest CPU and the fastest GPU in the computer does not necessarily give the best gaming performance, at least when it comes to efficiency. That’s why we created a Top 3 Ranking with the best performing combinations of CPU and GPU, which you can check out below.

Ranking First: MSI GAMING GeForce RTX 2070 + AMD Ryzen 9 3900X

Pros

  • Best Gaming Performance
  • Great for 4K Gaming
  • High endurance

Cons

  • Just the price

Nvidia’s Geforce RTX 2070 is a great graphics card if you value a smooth gaming experience at high resolutions – and don’t want to spend dramatically high amounts of money. But whoever has the choice is spoilt for choice: Of course, board partners like Zotac, Gigabyte or MSI offer their own designs with and without overclocking at very different prices. We have tested eight models and tell you who should buy which card.

In the high-end graphics card works one of the current Turing chips: The TU106 holds 2,304 shaders, the main processing units of every graphics card. The predecessor, the Geforce GTX 1070, still had 1,920 of this unit, also called “CUDA core” by Nvidia. Turing also adds integer units, RT and Tensor cores. The integer units are supposed to intercept the integer operations, which are increasingly found in games, better.

RT and tensor cores, on the other hand, are responsible for the implementation of Nvidia’s flagships Ray-Tracing and Deep Learning Super Sampling. These are methods to calculate light effects spectacularly in real time on the one hand, and to ensure efficient edge smoothing with the help of AI. You can read more about this topic in our review of the Nvidia RTX 2080 Ti. In the standard version, the clock frequencies range from 1.410 to 1.620 MHz.

The bottom line is that every RTX 2070 is thus a good graphics card for smooth gaming in 1440p resolution – or to exploit the high refresh rate of a gaming monitor in Full HD resolution. But some are better in comparison, others worse, thanks to factory overclocking by the board partners.

AMD Ryzen 9 3900X

The Ryzen 9 3900X is the new top dog among AMD processors away from the enthusiastic CPUs and brings top performance in games and applications.

At first glance, the price of the AMD Ryzen 9 3900X seems high compared to other AMD CPUs, but: In the application index, which summarizes the performance of six applications, AMD’s new twelve-core processor achieves 95.9 percent of the performance of an Intel Core i9-9980XE – but the Intel CPU costs almost four times as much.

With twelve cores and 24 threads, the Ryzen 9 3900X is also a good investment for the future. Especially when it comes to games on the PC, the processor can definitely be seen with a brilliant performance of 93 percent. The new 3000 generation is not only compatible with the new X570 chipset, but also with X470, B450, X370, B350 and A320 motherboards. With the older mainboards you should pay attention that the BIOS gets an update before.

Ranking Second ( Best Price): AMD Radeon RED Dragon RX 580 + AMD Ryzen 5 3600

Pros

  • Great Gaming Performance
  • Awesome Price
  • Good for 1080p Gaming

Cons

  • Graphics card a bit noisey under heavy workload

AMD Radeon RED Dragon RX 580

With the Radeon RX 580 Red Devil and the Radeon RX 580 Red Devil Golden Sample in the test, Powercolor is launching two models based on the revised Polaris 20 graphics chip. As usual with the series, the manufacturer relies on a high chip clock and equips the graphics cards with a massive cooler.

Powercolor equips the standard RX 580 Red Devil with the Polaris-20-XTX chip which clocks with 1.257 MHz / 1.380 MHz in boost. The Golden Sample uses the more clocked XTR version of the Polaris chip, Powercolor has a boost clock of 1.425 MHz, which is significantly higher than the 1.340 MHz specified by AMD. Only the Sapphire Nitro+ Limited Edition tested by us clocks even higher with 1.450 MHz.

The 8.0 GByte GDDR5 memory remains unaffected with 8.0 GHz clock, both Red Devils come with the 256 bit wide memory interface thus to a memory bandwidth of 256 GByte per second. Despite the higher chip clock, the TDP remains at 185 watts, the graphics cards have a 6-pin and 8-pin power connector each.

The biggest innovation concerns the cooling: Powercolor says goodbye to the two-slot design with three 80 millimetre fans of the Radeon RX 480 Red Devil and relies on a triple-slot design with two 100 millimetre axial fans in the new top models. The cooling block extending over the PCB with horizontally aligned fins varies in size depending on the model. The normal Red Devil measures 46 millimetres in height, and the cooling system with four heatpipes is correspondingly more compact. The RX 580 Red Devil Golden Sample is considerably more massive with a height of 57 millimetres and has five heatpipes.

The familiar switch on the top of the card to switch from OC to silent BIOS is supplemented by another one that takes care of the red illumination of the logo on the side. By the way, this has been rotated in relation to that of the RX480’s predecessor so that it is no longer upside down once the card is inserted in the case. Visually, the cooler cover, which is kept in black and red, remains true to the series, and the backplate is also still decorated with a pentagram.

An Intel Core i7 7700K, which we let run with the maximum standard clock rates, works in the updated test system. The processor only becomes a potential bottleneck late in the game thanks to the high clock rate of its four cores and the ability to process eight threads simultaneously. It’s cooled by a Noctua NH-D15.

The ASUS ROG Maximus IX Hero with a Z270 chipset is equipped with 16.0 Gbyte DDR4 memory from G.Skill, which we run with Intel’s designated clock rate of 2,400 MHz. Power is supplied by a be quiet! power zone with 1,000 watts. On the software side we rely on Windows 10 (64 bit), the applications and games are installed on a 480 GByte Crucial BX200 SSD.

For the gaming measurements, the graphics card has to prove itself in four DX11, one DX12 and one volcano game (Battlefield 1, Doom, Rise of the Tomb Raider, Total War: Warhammer, The Witcher 3: Wild Hunt, Watch Dogs 2). We test all of the above titles with the highest preset including edge smoothing and sixteen times anisotropic filtering.

The measured power consumption refers to the entire system; we determined the temperatures with the MSI Afterburner under typical gaming load. We carry out our volume measurements in a sound-insulated room with a Voltcraft SL200. The measuring device is placed at a distance of 50 cm, at an angle to the graphic card and the open test system.

Conclusion

The Powercolor Radeon RX 580 Red Devil and the Radeon RX 580 Red Devil Golden Sample score in the test with high chip clock and corresponding performance in our gaming benchmarks. Atypical here: The somewhat more expensive and higher clocked model delivers the better overall package.

Due to the more massive cooler, the Golden Sample can keep its boost clock of 1.425 MHz constant in games and is audible with 41.2 decibels, but not very noticeable or even disturbingly loud. The standard RX 580 Red Devil can’t keep its lower set boost clock constant, the chip clock settles between 1.330 and 1.380 MHz. We measured a clearly audible 43.1 decibels in games due to the somewhat undersized cooling. Particularly disturbing are the fans’ outliers, which repeatedly rotate with over 2,000 revolutions per minute under load before they level off.

As already with the Radeon RX 570 Red Devil, the silent BIOS also helps the RX 580 Red Devil to a significantly lower noise level with a painful performance loss. In the Silent BIOS, the fans of both RX 580 variants operate in the same way, only the Golden Sample clocks 100 MHz higher on average.

Both RX 580 Red Devil models are well manufactured and make a high-quality impression. Due to the new cooling design, both test candidates are three slots high, but only 25.5 centimeters long and should therefore fit into any standard case. New is the red illumination of the Red Devil logo on the side, which is activated by a switch on the board.

Overall, we recommend and rate the somewhat more expensive RX 580 Red Devil Golden Sample, while we rate the standard RX 580 Red Devil with one point less due to its higher volume.

AMD Ryzen 5 3600

The AMD Ryzen 5 3600 is the most affordable entry into the new generation of Zen2 processors.

With this processor you get AMD’s latest technology at the lowest price. The Ryzen 5 3600 with 6 cores/12 threads achieves 89.3 percent in our performance index for PC games – so you can be sure that you’ll be at the top of the list even with brand new games. In the applications performance index, the Ryzen 5 3600 achieves 59.1 percent and is thus in the “decent performance” range. However, it is not future-proof in terms of applications with 6c/12t – gamers can sit back and relax even longer.

Ranking Third: Sapphire Radeon Nitro+ RX 5500 XT + Intel Core i5-9400F

Pros

  • Fast work speed
  • Fair price
  • Good for 1080p Gaming

Cons

  • 60 fps not always available

When it comes to equipment, the Radeon RX 5500 XT Pulse does not reinvent the wheel. Although it offers three DisplayPorts in addition to an HDMI output, it does without the classic DVI output. In terms of width, it occupies two PCIe slots and is 23.3 centimeters long. Inside, the card has an 8 GByte GDDR6 video memory and promises a powerful graphics performance in combination with a boost clock of 1,845 MHz.

For this, the RX 5500 XT Pulse also requires a recommended power supply of at least 450 watts. However, the system only consumes 232 watts under full load in the test (idle 59 watts). If the graphics card is not used to great effect, cooling is passive. With the fans turned on, the GPU always stays under 72 degrees in 3D mode. Fortunately, the volume is kept within limits with a pleasant 1.4 sone and can hardly be heard even under full load.

To make it short: Full HD gaming is definitely possible with the Radeon RX 5500 XT. However, the details have to be reduced considerably if you want to achieve 60 fps or more in demanding games. In tests with games like Ashes of the Singularity, Far Cry 5 or Total War Warhammer II, we otherwise only achieve values between 30 and 45 fps – GTA V is the exception with over 90 fps.

The RX 5500 XT is thus primarily aimed at Full-HD gamers with low demands on graphic details or refresh rates – or if you want to mix smoothly in the most popular and usually graphically highly optimized multiplayer games. As an entry-level model the performance is therefore okay. However, a Geforce GTX 1660 offers something similar at a lower price. We have summarized all measured values for you once again in the following.

Conclusion

The Sapphire Radeon RX 5500 XT Pulse from AMD concentrates especially on Full-HD gaming and also delivers quite good results in the test. Although some details have to be scaled back for frame rates above 60 fps, the card manages above the 30 fps mark in full graphic splendor even in current titles. Merely gaming in UHD falls flat here; the refresh rates are simply too low for this. In return, the low volume and the low power consumption are pleasing. The price is luckily a bit below the direct competition.

Intel Core i5-9400F

The Intel Core i5-9400F is clearly the CPU for those who want the best gaming experience on the PC for a cheap price. In our tests, the CPU achieves 80 percentage points in terms of gaming. In the performance index for applications, the i5-9400F achieves 71.7 percent – a good performance. The fact that the i5-9400F overtakes the competitor AMD Ryzen in terms of performance for games is due to the massive boost clock of 5 GHz. Accordingly, power consumption and heat development are higher, so you should pay attention to good cooling. With 6 cores, the Intel CPU is still prepared for newer games in the long run, but the number of cores could slow down a bit in the long run for computationally intensive applications.

Important Tips for finding a good CPU and GPU Combination

The close coupling in the interaction of CPU and GPU has direct consequences for the combination or coordination of both components, i.e. of processor and graphics card.

What you should absolutely avoid is that the CPU cannot “tell” the GPU fast enough what to render next.

If this is the case, the GPU regularly goes into an idle state, i.e. it does not calculate a new frame although it has free capacity and could already calculate the next frame. If this situation occurs that the CPU cannot tell the GPU fast enough what to calculate next, it is called a CPU bottleneck.

The (insufficient) performance of the CPU thus throttles the actual performance of the GPU and fewer frames per time unit can be calculated.

The frame rate per second (frames per second) often quoted in games decreases and the gaming experience literally falters and the game jerks.

So make sure that you have an adequately powerful CPU at your side, otherwise you will simply waste money on a graphics card whose full potential you cannot use.

What CPU power do i need for the graphics card?

Which CPU is appropriate in terms of performance naturally depends on your target graphics card.

Basically, the stronger the graphics card, the more powerful the CPU should be to avoid the risk of the graphics card being limited by the bottleneck problem described above.

On the other hand, however, it doesn’t make sense to pair an overly strong CPU (e.g. a Core i7 8700K) with a relatively(!) weak GTX1050, since you are spending a lot of money on a processor that is never exhausted, at least in games, because the graphics card can’t keep up with the generation of the single frames (at least at higher resolutions and high graphics quality) in this case.

An interesting tool to find out how CPU and GPU fit together is the “Bottlenecker”, which can be found at http://thebottlenecker.com/

With the help of the tool, a so-called bottleneck percentage can be determined for a combination of CPU, GPU and RAM, which gives a rough indication of how well the system or its components are matched to each other. A high bottleneck percentage thus indicates that a component slows down the performance of the system above average (and thus unnecessarily).

It’s worth experimenting a little bit to get a feeling for the performance classes of the different CPUs and GPUs.

As a guideline for reasonable CPU/graphics card combinations that are performance-wise reasonable or not so reasonable, you can use this table, which was created with the bottleneck calculator as an example. Of course, such results should always be considered with caution, but the tendency of the tool’s results is surprisingly good and often matches actual experiences.

CPUGPUBottleneck-PercentageNote / classification
Intel Core i5 8400GTX 1070 TI6%Well coordinated
Intel Core i5 8400GTX 105055%CPU significantly oversized (waste of money)
Intel Core i5 8400GTX 1080 TI9%Well coordinated
Intel Core i7 8700GTX 1080 TI1%Very well coordinated
Intel Core i3 8100GTX 1080 TI18%CPU undersized
Ryzen 7 1800XGTX 1080 TI1%Very well coordinated
Ryzen 7 1800XGTX 97025%CPU significantly oversized (waste of money)

There is a tendency to underestimate the performance of processors and therefore often a model is chosen which is actually a bit too performant (or too expensive).

Basically, however, a modern Core i3 8100 with 3.6 Ghz and 4 cores already has enough power to drive most games without problems, especially when gaming in higher resolutions and the graphics card gets a bit busy.

Display Resolution particularly relevant for power requirements in the CPU/GPU combination

Surely it is obvious that the CPU limits the graphics card only as long as the graphics card is in principle faster in calculating an image (a frame) than the CPU needs for preparing the next render instruction.

With graphically demanding games and high resolutions (e.g. 3840 × 2160 pixels) with high details, the most powerful graphics chips will of course at some point reach their limits and have to “plough” properly to calculate the respective frame, so that with increasing resolution the probability of a CPU bottleneck becomes less and less.

The connection is obvious when you think about how an image is actually composed: from thousands of colored pixels, which together make up the actual image.

The higher the resolution, the more pixels and their color values the graphics card has to calculate to display the image, so the generation of the output becomes more complex.

Basically, the higher the resolution in which you want to play your games, the more powerful your graphics card has to be, because more pixels have to be calculated.

The following table illustrates the relationship between resolution and pixel count for some popular resolutions:

ResolutionWidth (in pixels)Height (in pixels)Total number of pixels (pixels)
Standard-HD1280720921,600
Full-HD192010802,073,600
WQ-HD256014403,686,400
Ultra HD (4k)384021608,294,400

What determines the performance of graphics cards?

The performance of a graphics card is essentially determined by the chip architecture, the number of computing cores, their clock speed and the size and speed of the GDDR RAM (graphics memory).

Similar to multi-core CPUs (which, however, only have a few computing cores), modern graphics cards consist of many hundreds or thousands of individual computing cores with which different tasks can be worked on in parallel. Basically, more computing cores can perform more operations per time unit.

The following table shows the relationship between the performance class of a graphics chip (GTX1080 TI as the most powerful graphics chip in the list, GTX 1050 as the weakest) and the technical specification or equipment.

As can be easily seen, the more powerful graphics chips are equipped with more computing cores or higher clock rates of the cores, more and faster GDDR memory.

Graphics chipComputing cores (CUDA cores)GPU base clock in MhzGDDR memory speedGDDR size
GTX 1080 TI3584148011 GBIT / second11 GB
GTX 10802560160710 GBIT / second8 GB
GTX 1070 TI243216078 GBIT / second8 GB
GTX 1070192015068 GBIT / second8 GB
GTX 1060128015068 GBIT / second6 GB
GTX 1050 TI76812907 GBIT / second4 GB
GTX 105064013547 GBIT / second2 GB

The graphics chip series from AMD follow a similar logic, whereby the computing cores at AMD are called “stream processors”.

Difference between graphics chip and graphics card

As you may have noticed, up to now people have mostly spoken of “graphics chip” instead of “graphics card”. This is because the graphics chip manufacturers (nVidia and AMD in the gaming sector) sell the graphics chips to graphics card manufacturers, who then install the chips in graphics cards. The graphics chip thus forms the heart of the graphics card, but as such is initially only a component of the actual graphics card.

Among the most well-known graphics card manufacturers are Zotac, Palit, EVGA, Asus and GIGABYTE.

The interesting thing is that each manufacturer now pursues its own ideas and concepts, starting with the actual graphics chip (e.g. from the GTX 1070 Ti series), e.g. within the framework of special overclocking versions or particularly quiet or well cooled graphics cards. The chip manufacturers supply so to speak only the “heart” and the graphics card manufacturers have leeway to modify this heart and its performance.

To do this, the graphics card manufacturers use e.g. leeway regarding the clock rates and often they raise the base (or reference) clock to get a little more performance out of the graphics chip. Sometimes different cooling concepts are used, so that one manufacturer gets higher clock rates and thus more power from a graphics chip than others.

It is therefore often worthwhile to analyze the performance of different graphics card models (but with the same chip) from different manufacturers. Sometimes you can save a lot of money if you are willing to give up 1 or 2% performance and buy the graphics card from manufacturer A instead of manufacturer B.

Similar to gaming mainboards, there is often a lot of marketing tam-tam in gaming graphics cards, whose critical scrutiny can save a few Dollars of the usually precious budget.

Actual performance is also always a question of software optimization and architecture

In practice, however, the performance of a graphics chip cannot be reliably judged from the pure number of computing cores, their clock or other attributes.

Due to various factors, it is impossible to predict the actual performance; at best, it is possible to derive trends with regard to the expected performance.

In addition to the architecture of the graphics chip (a highly complex issue), the question of the degree of software optimization is also crucial. There are quite a few game development studios that optimize their games specifically for nVidia graphics chips and thus the performance of an nVidia-based graphics card in this game is significantly better than the performance of an AMD-based graphics card that is similarly specified.

In practice, however, the performance of a graphics chip cannot be reliably judged from the pure number of computing cores, their clock or other attributes.

Due to various factors, it is impossible to predict the actual performance; at best, it is possible to derive trends with regard to the expected performance.

In addition to the architecture of the graphics chip (a highly complex issue), the question of the degree of software optimization is also crucial. There are quite a few game development studios that optimize their games specifically for nVidia graphics chips and thus the performance of an nVidia-based graphics card in this game is significantly better than the performance of an AMD-based graphics card that is similarly specified.

Is it Necessary to look at benchmark tests when choosing a graphics card?

Especially if you want to enjoy a concrete, already released game with your new computer, it makes sense to look for specific reviews of graphics cards on the net that have been tested with the respective game title.

It can also happen that different graphics chips perform better or worse with different titles, because the titles were optimized better or worse for a specific platform.

By the way, it is interesting to know that many popular online games like CS:Go, Fortnite, PUBG etc. usually have lower hardware requirements than the big AAA titles like Battlefield 1, CoD etc.

How much Performance in a Graphics card is enough?

Another thing you should always keep in mind when choosing a graphics card is the question of a reasonable compromise between the desired performance of the graphics card and your actual needs.

For example, if you only play on a 24″ monitor with a native resolution of 1920 * 1080 pixels without downsampling CS:GO, you don’t need a graphics card monster from the GTX 1080 class, here a model with a GTX 1050 chip or an AMD counterpart will do.

Of course, more performance is always nice, but whether you play at 150 FPS or 300 FPS makes practically no difference in the example sketched.

If you stay realistic and down-to-earth here, you can save money, which you might better invest in other components (e.g. a faster or larger SSD).

To answer this question in a meaningful way, you first need to know what you are actually looking for when gambling.

Do you want to play games (no matter what genre) because you enjoy the great graphics and you like to be drawn into the deceptively real gaming world?

Or do you gamble online because you like competition and like to prove your skills and like to compete with other gamers?

Are graphics not so important to you, and do you play games because you like the story or because you enjoy strategy?

So it depends on your personal requirements and your goals to find a graphics card that both meets your needs and fits your budget as well as possible.

Two Graphics cards in one system

Whoever takes a closer look at the topic of gaming graphics cards will sooner or later stumble upon so-called graphics card associations.

Here, two graphics cards are operated in parallel and thus ensure increased performance. This technology is called “SLI” for NVidia and “Crossfire” for ATI/AMD.

Even if it may sound tempting to have double graphics power, you should keep the following in mind (if your budget allows such experiments at all ;))

  • Operating two graphics cards does not automatically mean double power in the form of double frames per second (FPS). Not all games scale performance 1:1, i.e. they do not use the full power of a graphics network. In general, the performance increase of graphics networks is even significantly less than factor 2 (more like 1.5).
  • Graphics card networks logically require significantly more power – so your power supply must be able to cope with this additional demand. This means that a larger power supply will cost more and the heat generated by your system will increase significantly.
  • Especially with smaller cases (with few active fans), the installation of 2 graphics cards can contribute to a noticeably higher heat development besides general space problems. Without a suitable cooling concept, e.g. a large tower with additional active fans and sufficient space between the components, you expose your system to a certain risk of overheating

What PC Parts are important for the graphics card?

  • Mainboard – PCI-Express-Slot (ideally Version 3.0) should be connected with at least 8 lanes to not limit the potential of the graphics card
  • Power supply unit – Power supply unit must be able to provide sufficient power (watts); especially important in the higher power range. With our free power supply calculator you can calculate the required power of your power supply
  • Case – In case of graphics card networks, make sure that the case is large enough to ensure sufficient air circulation; mainboard must match the design of the tower (e.g. ATX design)
  • CPU – Graphics card and CPU should play in a performance league, a very powerful gaming graphics card is useless if the processor is too weak and vice versa, a too strong processor is a waste of money with a weak graphics card


Are you interestet in more hot tech deals? Subscribe to our Email-list and we let you know when the best deals come up.

Leave a Comment