Agner`s CPU blog

Software optimization resources | E-mail subscription to this blog | www.agner.org

Stop the instruction set war
Author: bitRAKE Date: 2010-01-12 11:49
Couldn't the instruction cache store an efficient post-decode encoding for instructions? IIRC, Intel already has a patent for doing this. Another possiblity would be to completely remap the instructions to favor parallel decoding. This would support backward capatiblity thru a processor external translator. Getting the cooperation (from both producers and consumers) might require a demonstraition of what can be gained.

Why couldn't Transmeta succeed? Does hyper-threading make the decoder a greater target? Has Intel used instruction set changes to negatively impact competitors (i.e. publishing the second best while secretly working on the target design)? Can a fair market exist with Intel's obvious advantage (both capital and market share)? Should something so one-sided be called a war?

 
thread Stop the instruction set war new - Agner Fog - 2009-12-05
replythread Stop the instruction set war new - Agner Fog - 2009-12-06
last reply The instruction set war's effect on virtualization new - Yuhong Bao - 2009-12-28
reply Stop the instruction set war new - Agner Fog - 2009-12-15
replythread Stop the instruction set war new - Norman Yarvin - 2010-01-09
last replythread Stop the instruction set war new - Agner Fog - 2010-01-10
last replythread Stop the instruction set war - bitRAKE - 2010-01-12
last replythread Stop the instruction set war new - Agner Fog - 2010-01-13
last reply Pentium Appendix H new - Yuhong Bao - 2010-02-10
replythread Stop the instruction set war new - Agner Fog - 2010-09-25
last reply Stop the instruction set war new - Agner - 2011-08-28
last replythread Stop the instruction set war new - Ruslan - 2016-04-17
last reply Stop the instruction set war new - Agner - 2016-04-17