Lolling at programmers, how many ways are there to create a bitmask ? ;) :)
|
Prev: PC = Personal Copier :)
Next: x86 instruction set usage-difference between windows 95 andwindows xp ?
From: Keith Thompson on 27 May 2010 11:46 Seebs <usenet-nospam(a)seebs.net> writes: > On 2010-05-27, Skybuck Flying <IntoTheFuture(a)hotmail.com> wrote: >> The lookup table right now seems to be the best solution for 32 bit >> support. > > I think at this point I'm plonking you. You're extremely verbose without > saying much of anything, you overuse exclamation marks, you're overly > exciteable, and apparently you don't have any clue how to frame a problem, > think about what you need, and develop a plan for approaching it. Maybe > you should consider decaf. A quick look at Skybuck Flying's posting history in other newsgroups might be illuminating. -- Keith Thompson (The_Other_Keith) kst-u(a)mib.org <http://www.ghoti.net/~kst> Nokia "We must do something. This is something. Therefore, we must do this." -- Antony Jay and Jonathan Lynn, "Yes Minister"
From: Skybuck Flying on 27 May 2010 12:06 "Seebs" <usenet-nospam(a)seebs.net> wrote in message news:slrnhvt42m.72l.usenet-nospam(a)guild.seebs.net... > On 2010-05-27, Skybuck Flying <IntoTheFuture(a)hotmail.com> wrote: >> So "shr 32" and "shl 32" could result in garbarge ?! > > On many systems. > >> That is pretty shitty ! > > No, it isn't. Don't do something incoherent. > >> Suppose one wants to write a longword to some bit stream then bitcount >> would >> always be 32 ! > > I have no idea what you think you are talking about. There is no reason > you > need to use shifts to write to a bitstream. Perhaps more importantly, > what > on earth do you think you're shifting? If you have a 32-bit value, and > you > shift it by 32, you have shifted ALL of the data out of it. Why bother? > If you're not going to be using any remaining bits at all, why are you > performing an operation? I think to get rid of a branch (branches slow down cpu's) and thereby speed up the code. Would you rather write: // 1. Z := X shl Y; // or // 2. if Y < 32 then begin Z := X shl Y; end else begin Z := X; end; ? Bye, Skybuck.
From: Seebs on 27 May 2010 13:10 On 2010-05-27, Keith Thompson <kst-u(a)mib.org> wrote: > A quick look at Skybuck Flying's posting history in other newsgroups > might be illuminating. I have enough information to convince me that he's a waste of valuable electrons. I think he's probably purely trolling, but it's the kind of trolling where genuine stupidity shines through. -s -- Copyright 2010, all wrongs reversed. Peter Seebach / usenet-nospam(a)seebs.net http://www.seebs.net/log/ <-- lawsuits, religion, and funny pictures http://en.wikipedia.org/wiki/Fair_Game_(Scientology) <-- get educated!
From: James Harris on 27 May 2010 14:58 On 27 May, 18:10, Seebs <usenet-nos...(a)seebs.net> wrote: > On 2010-05-27, Keith Thompson <ks...(a)mib.org> wrote: > > > A quick look at Skybuck Flying's posting history in other newsgroups > > might be illuminating. > > I have enough information to convince me that he's a waste of valuable > electrons. I think he's probably purely trolling, but it's the kind > of trolling where genuine stupidity shines through. I'm not going to try and defend him but having seen his posts for some time I don't think he's trolling. His interests are or have been video processing. He puts a lot of effort into getting the best x86 instruction sequences from his Delphi compiler. The rest is mainly communication style. James
From: James Harris on 27 May 2010 15:06
On 27 May, 17:06, "Skybuck Flying" <IntoTheFut...(a)hotmail.com> wrote: > "Seebs" <usenet-nos...(a)seebs.net> wrote in message > > news:slrnhvt42m.72l.usenet-nospam(a)guild.seebs.net... > > > > > On 2010-05-27, Skybuck Flying <IntoTheFut...(a)hotmail.com> wrote: > >> So "shr 32" and "shl 32" could result in garbarge ?! > > > On many systems. > > >> That is pretty shitty ! > > > No, it isn't. Don't do something incoherent. > > >> Suppose one wants to write a longword to some bit stream then bitcount > >> would > >> always be 32 ! > > > I have no idea what you think you are talking about. There is no reason > > you > > need to use shifts to write to a bitstream. Perhaps more importantly, > > what > > on earth do you think you're shifting? If you have a 32-bit value, and > > you > > shift it by 32, you have shifted ALL of the data out of it. Why bother? > > If you're not going to be using any remaining bits at all, why are you > > performing an operation? > > I think to get rid of a branch (branches slow down cpu's) and thereby speed > up the code. > > Would you rather write: > > // 1. > Z := X shl Y; > > // or > > // 2. > if Y < 32 then > begin > Z := X shl Y; > end else > begin > Z := X; > end; If you are talking about x86 don't be afraid of branches. Instead, be afraid of unpredictable branches. Further, from examples I've seen in the past the processors can make a surprisingly good job of predicting sequences we might consider random. It does help if you can present it stable sequences: longish runs of Y < 32, longish runs of Y >= 32 in your example. James |