Talk:Cycle stealing

Copyright Violation
There is copyright information on the "freedictionary.com" page as follows: "Computer Desktop Encyclopedia copyright ©1981-2005 by The Computer Language Company Inc. All Right reserved. THIS COPYRIGHTED DEFINITION IS FOR PERSONAL USE ONLY. All other reproduction is strictly prohibited without permission from the publisher"

Therefore, I believe I interpret it correctly when I say we shouldn't host material from freedictionary on Wikipedia. Kareeser|Talk! 00:28, 27 January 2006 (UTC)


 * Aaah! Will try rewrite the article now... thanks :) Foant 09:43, 27 January 2006 (UTC)

Definition of cycle stealing as not interfering with the CPU may not be meaningful any more?
I'm not sure the first definition is that meaningful any more, in these days of loosely coupled memory controllers and caches; the original use is more appropriate to the days when the CPU. memory bus, and I/O bus all ran synchronously; in modern systems, CPU stalls now happen all the time when cache misses occur, and DMA can occur without stalling the CPU providing the relevant data is in-cache.

As far as I know, the distributed computing meaning is now the most common. -- The Anome 09:57, 27 January 2006 (UTC)


 * Yea but for historical reasons I think we should describe what was meant back then, it is important to consider for example Programmed IO even if that is obsolete today. Foant 12:12, 27 January 2006 (UTC)

Cycle Stealing vs. DMA
I don't agree with the strange distinction between DMA and cycle stealing. In my mind, DMA describes a peripheral or other non-CPU logic block that can make a direct memory access. Cycle stealing describes a strategy for a memory controller to arbitrate between CPU and DMA requests, where the memory controller delays a CPU request ("steals a cycle") to service a DMA request instead. That is, DMA is a capability associated with a peripheral. Cycle stealing is a strategy associated with a memory controller. Am I off base here? --Mr z (talk) 10:23, 20 January 2019 (UTC)