The first 3D graphics cards appeared 25 years ago and since then their power and complexity have grown at a scale greater than any other microchip found in a PC. Back then, these processors packed around 1 million transistors, were smaller than 100 mm2 in size, and consumed just a handful of watts of electrical power.
Fast forward today, and a typical graphics card might have 14 billion transistors, in a die 500 mm2 in size, and consume over 200 W of power. The capabilities of these behemoths will be immeasurably greater than their ancient predecessors, but have they got any better at being efficient with all this tiny switches and energy?
A Tale of Two Numbers
In this article, we’ll take a look at how well GPU designers have utilized the increase in die size and power consumption to give us ever more processing power. Before we dive in, you first might to brush up on the components of a graphics card or take a walk through the history of the modern graphics processor. With this information, you’ll have a great foundation with which to follow this feature.
To understand how the efficiency of a GPU design has changed, if at all, over the years, we’ve used TechPowerUp’s excellent database, taking a sample of processors from the last 14 years. We’ve picked this timeframe because it marks the start of when GPUs had a unified shader structure.
Rather than having separate circuits within the chip for handling triangles and pixels, unified shaders are arithmetic logic units that are designed to process all the math required to do any calculation involved in 3D graphics. This allows us to use a relative performance measurement, consistently across the different GPUs: floating point operations per second (FLOPS, for short).
Hardware vendors are often keen to state FLOPS figures as a measure of the peak processing capability of the GPU and while it’s absolutely not the only aspect behind…