puzzle
in [General Programming]
|
Prev: Algorithm book recommendation?
Next: TaskScheduler.dll
From: Randy Howard on 20 Jun 2005 17:14 In article <42B6CDFB.236CC8C8(a)yahoo.com>, cbfalconer(a)yahoo.com says... > spinoza1111(a)yahoo.com wrote: > > This is what optimizing compilers do, mate. And in software > > maintenance you need to know how to make low-risk code > > improvements to "bad", Commie, terrorist, child molestor > > algorithms. > > Clearly showing the quality of the Nilgean denial. Rhymes with > Augean, I wonder if that is a coincidence? Either requires much > shoveling. > > However this business closely resembles the training of > puppy-dogs. Rubbing their noses in it doesn't usually improve > their behaviour - they just don't understand. > > So I pose the question to the readership: Are there any such > redeeming qualities that attempts should be made to train the Eddie > entity? Or should we enjoy the relief that comes from ceasing the > banging of foreheads on brick walls. You better be careful, he'll sick his lawyer on you if you speak the truth too often. :-) -- Randy Howard (2reply remove FOOBAR) "I don't really care about being right you know, I just care about success." --Steve Jobs
From: Randy Howard on 20 Jun 2005 17:18 In article <d96qrf$emd$2(a)grapevine.lcs.mit.edu>, cbarber(a)curl.com says... > I find his posts amusing, although presumably not in the way they > were intended. I know what you mean, I laugh out loud reading them frequently, although I can rarely stand to wade them in their entirety. You know what I mean, the diatribe portions that people always snip out of his posts. > However, I don't think that he is "trainable" and I suspect that > anyone that expects him to change is bound to be disappointed. Correct. > However, we cannot simply let him make false claims without at least > some refutation lest someone new to this group actually believe them. Exactly the point, one I and many others have made over the last year or more. -- Randy Howard (2reply remove FOOBAR) "I don't really care about being right you know, I just care about success." --Steve Jobs
From: Randy Howard on 20 Jun 2005 17:33 In article <5IKdnTTgnMqZEivfRVn-vg(a)comcast.com>, gershwin(a)splintermac.com says... > Randy Howard wrote: > [...] > > > > > > I think it's the same compiler that mystically hoists strlen() > > calls out of the top of for loops. It's never been demonstrated, > > but he swears that "any decent compiler" will do it. Apparently, > > there are no decent compilers in nilgewater land. > > Not to rise to the defense of Nilge here, but gcc 4.0 at least will > hoist strlen() like that. I am curious why you said that, because I just went back and used the exact same source to measure it on a PowerMac G5 (dual 2.0GHz) and didn't get results that seem to confirm what you say. FYI, the code was posted by me to comp.programming on 9/8/2004. The search string "Here's the test code" on that date should extract it from google if you're interested. Here are the results I just got using gcc 4.0.0 on OS X Tiger 10.4.1. Manually Hoisted version No optimizer 13.6s -O3 6.6s Unhoisted Version No optimizer 48.9s -O3 36.7s You get less of a performance speedup, but far, with the strlen in the loop and the optimizer on than you do with it done by the programmer, and the overall execution time is 6X longer. Based upon this, I don't see how you can conclude that gcc will optimize such poorly thought out code. -- Randy Howard (2reply remove FOOBAR) "I don't really care about being right you know, I just care about success." --Steve Jobs
From: Michael Wojcik on 20 Jun 2005 17:31 In article <MPG.1d1d43ddc56cf78698aada(a)news.verizon.net>, Randy Howard <randyhoward(a)FOOverizonBAR.net> writes: > In article <d8urr80a5r(a)news3.newsguy.com>, mwojcik(a)newsguy.com says... > > > > Indeed. Out of curiousity, Tisdale, how would endianness, or any > > ordering, affect any Boolean operator? > > I think you have ERT and nilgewater confused. Much as they both exhibit > some similar traits, albeit in different newsgroups, I am fairly well > convinced that not even ERT is this badly confused about programming > concepts. Argh. Thanks for catching that. Inexcusable on my part, really; normally I pay close attention to my attributions. And my apologies to Mr Tisdale, of course. -- Michael Wojcik michael.wojcik(a)microfocus.com How can I sing with love in my bosom? Unclean, immature and unseasonable salmon. -- Basil Bunting
From: Randy Howard on 20 Jun 2005 18:03
In article <d97cio02l4k(a)news4.newsguy.com>, mwojcik(a)newsguy.com says... > > I think you have ERT and nilgewater confused. Much as they both exhibit > > some similar traits, albeit in different newsgroups, I am fairly well > > convinced that not even ERT is this badly confused about programming > > concepts. > > Argh. Thanks for catching that. Inexcusable on my part, really; > normally I pay close attention to my attributions. > > And my apologies to Mr Tisdale, of course. :-) First time in a while anyone has felt sorry for Tisdale, but in this case it seems appropriate. -- Randy Howard (2reply remove FOOBAR) "I don't really care about being right you know, I just care about success." --Steve Jobs |