Talk:Carry-save adder
Add topic | This article is rated C-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | ||||||||||||||||||||
| |||||||||||||||||||||
Untitled
[edit]it is good that application to public key cryptography is mentioned, but the concept of carry save adders had not been explained clearly. I would expect Wiki to have a more intuitive explanation for the carry save adder.
Please see the following pdf. CSA has been nicely explained. http://www.ece.tamu.edu/~sshakkot/courses/ecen248/csa-notes.pdf
I think there is an error. There is a column where the result should be 2 instead of 1. If no one disagrees, I will edit the page. Vegasprof (talk) 20:50, 20 April 2020 (UTC)
The section about "the basic concept" is hard to understand. Towards the end, how do you obtain those two binary numbers as the result of passing through the carry save adder? That has not been sufficiently explained. Mr.Mog (talk) 17:11, 26 February 2021 (UTC)
I fixed that now Mr.Mog (talk) 09:25, 28 February 2021 (UTC)
This won't work
[edit]In section Technical detais there is some underestimations. If there is 3 binary numbers to add then it will work and carry propogation is needed from those two number addition (XOR sum of bits and OR sum of ANDs shifted one possition to left). But if there is need to add 4 binary numbers then absolutly unclear how to do it. It seems like there are generated with acumulator-safers 2 binary digits numbers from each 4 bits addition. If 8 binary numbers added together, then are generated 4 digits binary numbers, because 1000 in binary is 8. If 64 binary digits added then generated 6 binary digits number of each added 64 digits column (because 100000 is 64). Thus in the end there is 64 (if there was 64 binary numbers added) additions of 6 digits binary numbers shifted by one position to left more than previous (6 digits number). Thus arriving to the same problem from which begin, only now need to add 6 digits number 70 times (for multiplication say) instead to add 64 times 70 digits numbers like was in begining for getting 64 bits precision multiplication result of two 64 bits integers (result is most significand digits, like need in floating point multiplication for Extended precision). Thus needed additional tricks and schemes with milions transistors (for instant multiplication). ~2026-34791-94 (talk) 21:33, 13 June 2026 (UTC)