How to choose a Good Graphics Card for Gaming

Do you want to know how to choose a good graphics card for gaming? You are on the right spot to know the answer of this question.

Comparing graphic cards is not as easy as it seems. In fact, many compare graphics cards in the wrong way, and all they get is data that is not really correct.

In this article you will learn more about how to correctly compare these elements to choose the right one.

Mistakes to Avoid While Choosing Graphics Cards

The most common mistakes when comparing different graphics cards are:

1. Makes Comparison

Do not compare different generations of graphics cards, as they are not equivalent. Ideally, to see which is better or worse is to compare the same GPU architecture and see which performs better.

See Also: How to Underclock Graphics Card (AMD or NVIDIA)

This guarantees that both have the same functions and technologies. For example, a low-end of a new generation could have similar performance to a high-end of a previous generation.

You may be tempted by the price of the old generation, but the truth is that it may not have some technologies, such as, Ray Tracing. Therefore, you should not compare between pears and apples.

2. Comparing between different AMD and NVIDIA

It is also not logical to directly compare characteristics of one brand with another. This is what many usually do, but there are things that have no direct equivalent in one or architecture, so it is not recommended to look at that either.

For example, when AMD graphics cards included HBM-type memory, it couldn’t be directly compared to NVIDIA’s GDDR.

They are two memories working with different ways, each one with its advantages and disadvantages.

3. Compare by Processing Units

And closely related to the above, the processing units are not the same on both architectures, so you can’t compare AMD cores to NVIDIA cores.

If you do, you will see that the numbers of them does not always imply a higher performance.

See Also: Types of Blower Heatsink

It is a very big clumsiness, since each type of processor works in a different way and achieves performance in a different way.

4. Compare by VRAM

As I mentioned before, VRAM should also not be a benchmark for comparing graphics cards. Larger capacity does not always imply higher performance or work well at certain higher resolutions. Therefore, do not think that a graphics card with 6 GB is better than another with 4 GB for the mere fact of having an extra 2 GB.

5. Shop by FLOPS

Last but not least, many also look at the FLOPS that a graphics card can deliver. Another big mistake you can see how AMD with its architecture achieves more FLOPS or floating point operations per second.

NVIDIA in these types of calculations is a bit behind. Does that mean that AMD has better performance? No. In fact, NVIDIA is a tiny step ahead in performance.

This alerts us to look for an alternative unit other than FLOPS to compare the performance of the graphics.

How to Choose a Good Graphics Card

After reading the previous section, some myths will have been dismantled and you will think that how can one graphics card be compared with another. Well, here are the keys to choosing the best GPU:

1. Pixel Rate

This is a more reliable unit than FLOPS to compare graphics cards. This unit determines the number of pixels per second that a GPU can process. It is generally measured in Giga-pixels per second (GP/s).

See Also: Best Low Profile Graphics Card

So when it comes to pure graphics performance, comparing this figure over the power of single-precision and double-precision floating point is much better.

2. Texture Rate

Similar to the previous one, but in this case it is measured in GTexel/s, that is, to define the number of textures that the graphics card can generate in just one second.

This unit is also usually much more reliable than FLOPS for the same reason I mentioned in the previous point.

3. GDDR Generation

It is important to know the type and generation of VRAM. For example, GDDR5 is worse than GDDR6, and GDDR6 is worse than GDDR6X.

For example, the X is much better, as it uses PAM4 encoding (instead of NRZ) and doubles the bandwidth of the conventional 6.

In addition, it implies other improvements such as less temperature generated, less consumption, lower operating voltage, and greater margin for overclocking.

4. Bandwidth

Not only is it important that the latency to memory is low to access faster. The bandwidth of the channel that connects the VRAM with the GPU is also important, since performance also depends on it.

See Also: NVENC vs. NVDEC

The idea is to have high bandwidths. For example, the GeForce RTX 3050 has a GDDR6 with a bus width of 128-bit and a bandwidth of 224 GB/s.

5. Clock Frequency

Clock frequency is also a point to consider for performance. The higher it is, the more cycles per second it will perform, which means higher performance.

But again, don’t compare AMD’s GHZ to NVIDIA’s. You can only compare between models of the same brand to see the differences.

In fact, if you notice, just like with FLOPS, AMD has more, but it doesn’t have performance gains to match the clock increase it has.

6. Architecture (Gen)

It is vital that you look at the architecture of the GPU, its generation, since what it can and cannot do and its performance will largely depend on it.

See Also: How to choose Case Fans for your Pc

It’s always best to go for the latest generation. To get the most out of it, even if you end up going for a mid-range or low-end graphics card, but make it the latest Gen.

7. Technologies

This is related to the generation and the type of processors included in the graphics. But, it’s always good to see technologies like Ray Tracing, Tensor units to speed up AI, etc.

Now you know what things that you should not do when choosing a graphics card and those that you can use as tools for choosing. Don’t forget to comment with your doubts.