floating-point matrix multiplication
by yooohu on Sep 18, 2017 |
yooohu
Posts: 1 Joined: Sep 17, 2017 Last seen: May 8, 2018 |
||
I want to write a floating-point matrix multiplication, but there are some difficults, can you help me?
|
RE: floating-point matrix multiplication
by dgisselq on Sep 20, 2017 |
dgisselq
Posts: 247 Joined: Feb 20, 2015 Last seen: Oct 24, 2024 |
||
Can I help you?
That depends upon what kind of help you want. There are many floating point cores within the OpenCores project list. These may be of some use for you. Are you building your algorithm from scratch? The first step, in my humble opinion, is to identify how your I/O will take place. Will you be passing numbers into your hardware one at a time? One bit at a time? Or will they all be available at once? The answer to this question will then inform your selection of both floating point algorithm--since it will let you know to what extent you can share your computational logic hardware between your operations. You may also find the pipeline strategies described on the zipcpu.com to be useful as well. If you dig a little farther, I think you will quickly discover that many of the individuals on this forum are available for hire on an hourly rate--at least I am. Hence, my question: what kind of help do you want? Dan |