From 76a6f268194d68fc2d64e37e4189b3e9860f33a6 Mon Sep 17 00:00:00 2001 From: lkcl Date: Sun, 30 Apr 2023 02:55:24 +0100 Subject: [PATCH] --- openpower/sv/rfc/ls012.mdwn | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/openpower/sv/rfc/ls012.mdwn b/openpower/sv/rfc/ls012.mdwn index 693e2dcec..2dbde3444 100644 --- a/openpower/sv/rfc/ls012.mdwn +++ b/openpower/sv/rfc/ls012.mdwn @@ -577,6 +577,13 @@ assistance on, to make the evaluation easier. an FP64 Multiply typically takes between 12 to 15,000. Not counting the cost in hardware terms is just asking for trouble. +If the number of gates gets too large it has an unintended side-effect: +power consumption goes up but so does the distance between functions +on-chip. A good illustration here is the CDC6600 and Cray Supercomputers +where speed was limited by the size of the *room*. In other words larger +functions cause communication delays, and communication delays reduce +top speed. + **How long will it take to complete?** In the case of divide or Transcendentals the algorithms needed are so -- 2.30.2