Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Register Log in

Stochastic rounding for floating point

Status
Not open for further replies.
O

oAwad

Full Member level 2
Joined
Feb 15, 2017
Messages
136
Helped
0
Reputation
0
Reaction score
0
Trophy points
16
Activity points
1,312
If I want to design a stochastic rounding unit for a floating point adder in hardware (instead of a round to nearest even unit). What would be its architecture? I'm targetting to minimize the area.

This is the equation of the stochastic rounding scheme:
stochastic rounding.png
 

For each rounding, generate a random number z the range 0 - 0.999999999...

Then

Round(x) = | x + z |

Make sure that the x+z operation is done with truncation, not rounding.
 

Can't you just have an LFSR or counter or similar generate a value, and then compare the bits that would be truncated to this value?
 

std_match said:
For each rounding, generate a random number z the range 0 - 0.999999999...

Then

Round(x) = | x + z |

Make sure that the x+z operation is done with truncation, not rounding.
Click to expand...

With | x + z | i meant the integer part of x+z, because I thought you had used the same notation.
I now see that you used another symbol than |.
I adjust my answer to Round(x) = Integer_part_of(x + z)
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top