When, Where, And How To Overclock Your Graphics Card: A Full Guide

For gamers, graphics cards are the beating heart of the game rig.

In order to play games at a smooth frame rate and with most visuals enabled, the card must meet the minimum clock speed and RAM of the game. While you cannot do anything about the fixed RAM of the card, you wring more power out of its clockspeed. The process of modifying the clock speed is called overclocking.

When should you overclock your GPU? There’s no exact answer to this question. It will vary from system to system. This article will serve as your general guide to GPU overclocking.

Pre-overclocking: Points to Ponder

When you’re getting a new graphics card, or overclocking your existing one, the most important thing to consider is the power draw of your card and the power limits of your current system. If you’re getting a new card, you should consider how it’s going to fit into an already crowded PCIe slot. You’ll need to decide whether you want to run the card at the highest clock speed or whether you want to increase the number of cores by adding more cores.

If you’re overclocking your existing card, you’ll need to decide whether you want to increase the clock speed to achieve better overall performance, or whether you’re going to use all of the power available.

Does overclocking increase FPS? Overclocking your card is always going to increase the FPS at which you can play survival games. But it’s not always going to be the same as increasing the number of cores.

Which to Overclock: CPU or GPU?

CPU overclocks are nice too, but a good GPU overclock might have a stronger impact on the games you play. The best graphics cards still struggle at higher resolutions like 4K and 1440p on monitors pushing refresh rates past 60Hz.

Nvidia’s new Turing architecture is the foundation of today’s graphics cards, and it’s the biggest step forward in graphics technology in ages. At the core of each GPU is the basic building block. Nvidia calls it Streaming Multiprocessor (SM), and AMD calls it the Compute Unit (CU), but although specific implementations differ, every GPU usually has a number of clusters of SMs. The Turing Architectures SMs include scheduling engines, graphics cores, L1/L2 caches, texture units, and more.

Nvidia has drastically changed the Turing architecture SMs from previous Pascal and Maxwell architectures, so that is a lot to take in. Let’s start from the very top. First, there are now 64 CPU cores per SM, down from 128 in Pascal. Nvidia has jumped back and forth in the last decade between anywhere between 32 and 192 CUDA cores per SM, but with other architecture changes, 64 cores are now much more efficient, according to Nvidia.

We should note in its newer Turing cards, Nvidia has included more convenient auto-overclocking tools which allows users to do single-click overclocking for the highest performance possible in each card. It is called the Nvidia Scanner, and has been rolled out to every graphics card vendor selling RTX cards. Scanner is meant to replace the lengthy process of stepping up overclocks to see how much the card can push. Instead, the tool runs about 20 minutes worth of tests to see what every card is capable of with no crashes. Once testing is completed, it generates an overclocking profile that owners can turn on, if they wish.

An Nvidia scanner demo used on the GeForce RTX 2080 saw its baseline clockspeed increase from just over 1700MHz to around 2000MHz. And since this was done with NVIDIA’s own tools, this is a stable overclock that you can count on. Your mileage will vary, however, since every graphics card is going to be slightly different, and may only be stable at lower clock speeds.

When you increase the clock speed to achieve better performance, the number of cores is the same and you’re not going to get a performance increase. However, you will get an increase in the number of frames per second. And you can do this even if the card is already overclocked.

Your graphics card will also limit how much you can push when overclocking. It limits you to how much power can be pushed through. So, if you have a graphics card that’s designed for use in a gaming system, there’s a limit to how much power it can handle. The other thing you need to consider is whether you’re going to push the limits of your motherboard.

What Parts of a PC Can You Overclock?

Technically, any part of the PC that has a clock can be overclocked. However in practice, that doesn’t mean anything you can do to the motherboard will improve the performance of the graphics card. The motherboard controls the power to the components, but it doesn’t control the clock speed, so it doesn’t control how fast a component works or how much power it consumes. A motherboard’s power supply can, however, be adjusted to allow it to provide more power at a faster rate and thus allow components to run faster.

When you overclock a component you actually speed up its clock speed. The clock speed of a component measures how fast it produces electrons. A component that produces electrons more quickly (runs faster) consumes less power to do so, and is generally cheaper to produce.

A good way of overclocking your graphics card is to have two identical graphics cards in your system, and swap them in and out to test how they perform. You can also run benchmarks with multiple cards installed, if you happen to have a second GPU or your system still uses an SLI-linked card.

If you’re an Intel user, overclocking your graphics card isn’t something you have to do; the CPU is already overclocked, and Intel’s integrated graphics are quite capable. If you’ve got an AMD processor on the other hand, then you’ll need to overclock your graphics card. Windows 8 allows you to overclock your AMD graphics card using the AMD Overdrive software.

With your card cleaned up and ready for overclocking, you can typically max out the voltage and power limits on most graphics cards for even more PC performance. For Ampere cards, increasing the voltage doesn’t seem to have much of an impact. On the other hand, increasing the power limit has a noticeable increase in performance.

When it comes to the GeForce series, you can get a significant boost by overclocking the core. Unfortunately, the reference cooling solution doesn’t allow you to overclock the memory, so a result, you’ll need to remove the stock cooler and install a custom cooler. Now, just modify fan speed from here, and it should complete a decent overclocking.