Skip to: Site menu | Main content

Intro to FPGA Based PCI Hardware Acceleration Cards

As it is harder to make processors go faster another solution is being used. FPGAs (Field Programmable Gate Arrays) can be used with, or replacing microprocessors for many applications. By placing FPGAs on PCI cards, you can achieve far better throughput than a current generation CPU. This article discusses PCI acceleration cards that work with a computer to allow it to solve an important task more efficiently.

What are FPGAs?

FPGAs are devices containing logic building blocks that you may configure. Similar to a CPU, FPGAs may be configured to perform multiple tasks. Most of the logic blocks inside FPGAs are very simple. A few one bit wide inputs, and one main output typically. Some FPGAs have bigger logic units, such as PowerPC processors, Ethernet MACs and RAM blocks. These logic units are connected using a large network of programmable connections and wires. For PCI acceleration cards, an FPGA is connected with other devices, such as memory and IO, to create a deployable system. You can compare PCI acceleration cards to the mother board of a computer, and the FPGA as a replacement for the microprocessor. Also required are software tools, and infrastructure building blocks to simplify the creation of the users designs.

 How do FPGAs compare against processors?

A CPU and an FPGA are able to implement many of the same task. You can in fact implement processors with in an FPGA. Some FPGAs even have processors within them. A microprocessor included in the FPGA may use the FPGA to add instructions that allow programs to run much quicker than a unaided microprocessor. Some key distinctions between microprocessors and FPGAs are in how they implement their function, how they are programmed, and which task they are best at. A given microprocessor has a set hardware configuration consisting of different types of functional units... more>>>