Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

A prior art reference #1

Open
leobru opened this issue Mar 7, 2016 · 2 comments
Open

A prior art reference #1

leobru opened this issue Mar 7, 2016 · 2 comments

Comments

@leobru
Copy link

leobru commented Mar 7, 2016

It might be worth mentioning in the article that a similar functionality existed in legacy architectures.
Cf. the "young bits" register in BESM-6: http://mailcom.com/besm6/instset.shtml

@Maratyszcza
Copy link
Owner

Thank you for the link. We'll mention it in later revisions of the paper, but the operation is not quite the same as FPADDRE. I don't fully understand the "young bits" register, but the YTA pseudo-code sets the sign bit to 0, while a true FPADDRE instruction can produce negative values (e.g. for a and b with non-overlapping mantissas and |a| > |b| we have FPADDRE(a, b) == b, which can be negative)

@leobru
Copy link
Author

leobru commented Mar 8, 2016

To perform the double precision calculations, rounding of the single precision result was suppressed (bit 2 of register R was set), therefore the "young bits" were always increasing the absolute value of the mantissa, i.e. their sign was guaranteed to be the same as the sign of the result. The decision not to replicate the sign in the representation likely had some rationale, but I won't be able to reverse-engineer it on the spot.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants