-
WikiChip
WikiChip
-
Architectures
Popular x86
-
Intel
- Client
- Server
- Big Cores
- Small Cores
-
AMD
Popular ARM
-
ARM
- Server
- Big
- Little
-
Cavium
-
Samsung
-
-
Chips
Popular Families
-
Ampere
-
Apple
-
Cavium
-
HiSilicon
-
MediaTek
-
NXP
-
Qualcomm
-
Renesas
-
Samsung
-
Hardware acceleration refers to the use of specialized hardware circuitry in order to speed up certain computational tasks. Hardware acceleration is designed to speed-up certain functions more than would otherwise be possible using optimized software on a general-purpose microprocessor. The hardware circuitry that performs the acceleration is called an accelerator and the software utilizing those accelerators are called hardware-accelerated software.
Overview
As traditional single-core performance improvements slows down, a need to improve the performance of certain algorithms continues. Additionally, repetitive algorithms can consume a large amount of the processor's resources. Hardware acceleration is any use of specialized hardware circuitry used to speed up computational tasks. Typically, the speed-up is by as much as an order of magnitude vs. what's possible using ordinary software on a general-purpose microprocessor. Additionally, by off-loading those intense computations to be done by specialized hardware, the main processor's utilization goes down, freeing up resources for other workloads.
Typical accelerated workloads
- Artificial Intelligence
- Compression
- Cryptography
- Graphics
- Vector Manipulation