# RF Wireless World

## Home of RF and Wireless Vendors and Resources

One Stop For Your RF and Wireless Need

## CPU vs GPU-difference between CPU and GPU

This page compares CPU vs GPU and describes difference between CPU and GPU. The other difference between terms are also provided here. CPU stands for Central Processing Unit and GPU stands for Graphics Processing Unit. As we know, 4004 chip was the first 4 bit CPU developed by Intel. Later on x86 architecture has been designed which has become more popular. Later ARM has come up with 32 bit microprocessor which was originally created by Acorn computers.

Prior to the development of first GPU by NVIDIA in 1999 with the introduction of "GeForce 256" chip, graphics functionality in the desktop computers was handled by VGA(video graphics array) controller. The VGA job was to render image data in the proper order to the monitor for display.

Initially GPU was fixed function device designed to take care of fixed set of tasks. Later on GPUs have become programmable like CPUs. Small GPU programs can run for each vertex, triangle as well as pixel processing. This made it easy to generate the visual effects quickly and efficiently. We will see comparison between CPU vs GPU which describes the difference between CPU and GPU.

### Basic difference between CPU and GPU

• Architecture wise, CPU consists of few cores with good amount of cache memory while GPU consists of hundreds of cores.

• CPU can handle few software thread at one time while GPU can take care of thousands of threads at a time.

• CPUs are much better at performing a much larger array of tasks than GPUs.

• CPUs can reduce reduce latency more effectively compare to GPUs for individual tasks.

• GPU is more power efficient and cost efficient compare to CPU.

• The host code runs on CPU, CUDA code runs on GPU.

• The following is the comparison of array increment operation program running on GPU and on CPU.
****Code running on GPU*****
block = 1:4 by 1:4
if y[i][j]= within block
y[i][j] = y[i][j] + 1

****Code running on CPU*****
for i = 1 to 4
for j = 1 to 4
y[i][j] = y[i][j] +1

Inspite of difference between CPU and GPU, the companies are taking advantages of both of them and developing SoC(System on Chip) which houses both of these components/modules. Figure below shows NVIDIA's Tegra chip which is designed for smartphones.

Unlike CPU component which was manufactured by few companies such as Intel and AMD there are many GPU vendors. The companies such as ARM license their CPU core architecture as well as they have fixed set of core designs to be sold. Companies can choose readily available ARM designs and package them along with GPU and other components or make their own custom CPUs along with other devices. Due to availability of many GPU vendors companies have options to choose right GPU along with CPU for the SoC design as per their requirements.

Popular vendors for GPUs are ARM, Qualcomm, PowerVR and NVIDIA. Following table mentions GPUs from these companies used in a smartphones released by companies such as Samsung, Apple, LG, Micromax and so on.

### CPUs and GPUs used in smartphones

Following table mentions CPUs and GPUs for smartphones and tablets from popular companies such as ARM, Qualcomm, PowerVR, NVIDIA, Apple etc.

Smartphone type Chipset CPU GPU
Apple iPhone 5S Apple A7 Dual core cyclone Power VR G6430
Samsung Galaxy Alpha Exynos 5 Octa 5430 Quad-core Cortex A15,
Mali-T628 MP6
Micromax Canvas Knight Mediatek 6592T Octa-core Cortex A7 Mali-450MP4