This is the mail archive of the cgen@sources.redhat.com mailing list for the CGEN project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [patch][rfa]: Decoding (not-so) ambiguous insns in sid/sim


Frank Ch. Eigler wrote:

>brolley wrote:
>
>>[...]
>>      (list->vector
>>       (map (lambda (p0 p1)
>>  	    (logit 4 p0 "/" p1 " ")
>>! 	    ; (sqrt (+ p0 p1 (* p0 p1))) ; funny function - nice curve
>>! 	    (sqrt (* p0 p1))) ; geometric mean
>>  	  (vector->list 0-population) (vector->list 1-population))))
>>
>
>By the way, have you tried the "funny function" above?  It would have
>the property of separating the 0/0 and 0/N cases like your new
>ordering function does, and is monotonic in p0 & p1, so it should not
>change the relative order of existing decoders.
>
Yes, I considered the "funny function" but rejected it because it 
assigns a non-zero value in the case where a given bit must always be 1 
or always 0, e.g. p0==num_insns,p1==0. Such a bit is useless for 
decoding purposes and is correctly assigned a value of zero both by the 
old function and by my new function.

Can I consider your response to be approval to commit?

What about "filter-harmlessly-ambiguous-insns"? Should I remove it from 
insn.scm?

Dave


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]