How to overclock an AMD graphics card
Many graphics accelerators from AMD are initially overclocked better than Nvidia, so they do not have such significant overclocking potential. Especially extreme models with factory overclocking. In terms of frequency, Radeon memory “chases” perfectly, the peak value is often limited by the physical capabilities of the chips themselves.
Overclocking RX 6000 series
New generation architecture based on the Navi 2 video chip divided into clusters. Each consists of 64 compute units and 4 texture units. 7nm process technology with increased energy efficiency (up to 50%) raised frequencies and overclocking capabilities. Boost frequencies exceed 2 GHz. Ray tracing calculation blocks improve the realism of scene lighting.
An additional third-level cache will reduce the frequency of external memory accesses. The bandwidth of Infinity Cache on a 256-bit bus will reduce electricity consumption, heat, and will delight fans of overclocking. Variable Rate Shading will change the rendering quality of scene segments (relevant for racing).
Parameter Map Model | 6900 XT | 6800 XT | 6800 | RX 6800 |
Core Voltage, В | 1.15 | 1.025 | 1.15 | 1.15 |
Power Limit | 115% | 115% | 115% | |
Core Clock, MHz | 2750 | 2294 | 2306 | 2600 |
Memory Clock, MHz | 2150 | 2150 | – | 4232 |
Overclocking RX 5000 Series
The RDNA architecture is the result of 8 years of work by the computer giant, aimed at working with three-dimensional applications. This is not an ordinary architecture, but a synthesis of a number of technological solutions of the company for playing three-dimensional content – games. Three frequencies are indicated: nominal, game, Boost. They work through the new high-speed PCI Express 4.0 interface. Optimization managed to reduce the thermal pack by ~23% and increase the performance up to 14%. Cost-effectiveness is achieved due to the almost halved area of the crystal – to get a noticeable increase in performance during overclocking, you will have to change the cooling.
Parameter Map Model | RX 5600 XT | RX 5700 | RX 5500 XT |
Core Voltage, В | 0.993 | – | 1.145 |
Power Limit | 130% | 120% | 120% |
Core Clock, MHz | 1820 | 1750 | 1900 |
Memory Clock, MHz | 1800 | 1810 | 1860 |
Overclocking RX Vega Series
In addition to processing real and integer numbers, Vega performs integer operations with different accuracy (relevant for machine learning, mining). The NCU chip enhances throughput by combining single-bit operations. Redesigned tile rendering technology reduces the number of external memory accesses. It is activated at the driver level, noticeably accelerates gaming performance and reduces bus load by up to 33%.
Parameter Map Model | RX Vega 56 | RX Vega 64 | RX Vega 64 Liquid |
Core Voltage, В | 1.175 | – | 1.218 |
Core Clock, MHz | 1490 | 1645 | 1700 |
Memory Clock, MHz | 1790 | 2100 | 2150 |
Overclocking RX 500 Series
Devices based on Polaris chips with RX 400 microarchitecture with increased clock speeds and updated 12nm process technology. Changes: improved VR support, faster geometry processing, prioritization of asynchronous calculations. The ruler has an increased cache of all levels. They are characterized by increased power consumption, they accelerate well.
Parameter Map Model | RX 550 | RX 560 | RX 570 | RX 580 | RX 590 |
Core Voltage, В | – | 1.1 | +0.2 | 1.2 | – |
Power Limit | – | Maximum | – | 150% | – |
Core Clock, MHz | 1405 | 1370 | 1444 | 1500 | 1635 |
Memory Clock, MHz | 2000 | 2010 | 2000 | 2175 | 2300 |