1. Home
  2. Field-Programmable Gate Arrays
  3. Double buffering

Double buffering

I've read much about Double Buffering, especially that it is good practice (on Xilinx devices) to double buffer data signals (such as ADC inputs), and place the double buffer in the IOB associated with the pin.

Can someone explain to me, what double buffering is and why you would use it? Any links to reference information would be appreciated.

Many thanks.

Reply to
dh2006
Loading thread data ...

dh2006 schrieb:

Double buffering is one way of syncronising signals which are being passed between clock domains, primarily in order to prevent problems with metastability.Its not only xilinx devices in which this is used its used in any situation where signals which are asyncronous to some logic need to be syncronised.There are other methods which can be used to achieve the same thing.A google search on metestabilty will give you further information.

Reply to
jez-smith

... and as an extra precision, "not only Xilinx devices" does not only mean Altera's, Cypress and other programmable logic companies. Double-registered async inputs are common fare in clocked digital circuits of any sort, ASICs are not somehow magically exempted from this.

Reply to
Daniel S.

"Double buffering" is a very vague term and means different things in different contexts. Here are a few things it might mean in FPGA terms:

  • Double-registering - as a previous poster pointed out, it is very common to use two ranks of synchronizing flip-flops when moving from one clock domain to another. This is a favourite topic in these parts so you'll have no difficulty in finding an abundance of information about that. :-)
  • Double data rate - many FPGAs nowadays have a special configuration flip-flops to cater for DDR SDRAM and the like. These allow data to be clock in and out to the external device on both edges of the reference clock, while still permitting a single-clock design inside the FPGA itself. Refer to your device datasheet to find out how these can be used.
  • Double-buffering - say this to a designer of DSP or graphics processing hardware and they will immediately think of a pair of RAM blocks being used as a buffer between two processing elements in a so-called "ping pong" configuration. This is a method for increasing the throughput of a system, by allowing the data source to write a block of data to memory 'A' while the data processing unit is reading the previous block from memory 'B'. When both operations are done, A and B are swapped and the processing continues.

Judging from your use of the phrase "good practice", I think you are probably refering to double-registering for synchronization. In which case, do listen to the many experts who inhabit this group and you will likely save yourself much pain and suffering!

Cheers,

-Ben-

Reply to
Ben Jones

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.