| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9
Comments:
<0> your'e wasting EVERYONE's time
<1> corstan how about you just leave until you get a book?
<2> does anyone get paid to answer my questions?
<1> or if you need help, we could ban you until then
<0> no, everyone here is a volunteer
<0> but we don't have to put up with your crap
<2> it seems as though only you two who are pretty sick of me
<2> nobody else is uttering a word
<0> they're not as polite as we are
<1> most others aren't as active
<3> continuing this discussion would be unwise ;)
<0> we've been trying to ***ist you, but you're not taking any of it
<2> vawjr corstan look ****head....pay real ****ing close attention
<2> are you having a heartattack?
<0> not at all
<0> I just get tired of spending effort and watching it get wasted
<1> anyone have a link to a picture of that danish cartoon thats causing all the uproar?
<1> trying to find it... seems to be tough
<1> oh, found it heh... for anyone interested: http://www.brusselsjournal.com/node/698
<0> I've been wondering how they know it's an image of the prophet. I thought all such were "forbidden"
<0> so how the hell do they know what he looked like
<1> I think the point of the cartoon was to depict the prophet
<0> but the injunction is so that people aren't tempted to idolatry
<1> ah
<1> er... are they pissed simply because he was depicted, or more because of the way he's depicted?
<0> the CLAIM was that you can't depict him because of the idolatry "problem"
<0> either they're really ****ing stupid (who's going to pick a cartoon as an idol?)
<0> or they're really ****ing stupid because they can't say WHY they're upset
<4> Hi im currently converting some C Code to Vb.net and have got stuck with a line as to what its doing: dp is a unsigned char, dp[2] = (unitID >> 8) & 0x00ff; can someone explain what this is doing
<1> C to VB.net... neither of which touches C++
<1> >> is the right bitshift operator, and & is bitwise and
<1> if dp is unsigned char, then dp[2] is invalid
<0> no, dp is NOT an unsigned char
<0> I'm a little slow... rdragon beat me to it
<1> hehe
<5> I'm working with this object that has a method named gRef(). The object takes long/lat and converts it to image coordinates in pixels. But I don't know what "Longitude of Origin" is.
<5> Anyone here messed with maps?
<3> probably the prime meridian
<3> too bad he left
<6> hello
<3> yo
<7> i have this bitmap selected in a device context. i must get the bitmap width & height. how do i do that? (what gdi api)
<8> Deathmaster: I would try GetBitmapDimensionEx()
<9> bitmaps.. eww
<8> yeah, I had to open MSDN to find that out, double eww
<9> nothing wrong with dat
<8> i don't like MSDN very much, i find it increasingly hard to find stuff in it
<9> google
<8> actually, i find way too much stuff besides what i search for
<9> like what
<9> send in an email with your suggestions for improvements
<9> or better yet, fix it yourself
<8> like windows CE functions, knowledge base articles that have no connection, visual basic documentation, and now .NET stuff i'm not really interested in. I've tried to set subsets, but somehow crap creeps in the search results, or it resets spontaneously to "all"
<8> fix msdn myself? that's funny
<9> yeah.. get a job there and fix it
<7> hmm thanks your your support guys
<7> but getbitmapdimensionex doesn't work
<9> which URL did you go to to set the search subset
<8> getting the job would not be a problem, but what makes you think i'd willingly work on MSDN once at microsoft? :)
<7> it says there it returns the size from an internal structure
<7> but only if that structure is previously set
<8> what are you talking about? URL?
<9> Cd: someone's gotta do it.. you seem motivated
<8> EtoYa: i am not motivated to write or fix any documentation in any way whatsoever. i am a programer, it's like a professional disease, the aversion for writing documentation
<9> fix the search engine
<8> Deathmaster: sorry. no further ideas
<9> Death: what ya trying to do ?
<9> CD: what url were you searching at that wasn't properly limiting the results
<9> ?
<9> I'll take that URL and shove it up someone's fat and lazy ...
<8> EtoYa: you are talking about the online MSDN and i am talking about the one that installs locally. the web-based MSDN is not any better than the local one, and it's slower to boot.
<9> have you tried google? or msn search?
<9> it works great
<9> include "msdn" into the search criteria
<8> i don't have a need right now, and i am complaining about the local MSDN, not about google. google is okay
<9> send email to MS
<1> is it safe to remove an element from a System::Collections::Generic::List<> within the middle of a 'for each' loop?
<9> should be alright
<8> EtoYa: it's not worth the effort. i don't use MS stuff except very rarely, although i do use VC++ at work, i have to write portable code -- and i don't do UI at all, i work on a server application (lucky me)
<9> lucky u
<1> i write server stuff but have been writing a bunch of managed code lately
<8> how does it feel?
<9> Nitche said
<9> if it ain't gonna kill ya, might make you stronger
<8> the little i've seen i didn't much like -- but then again, there are things in plain c++ i can't say i particularly like
<9> Cd: what didn't you like
<1> eh, c++/cli and the .net framework has been a pleasure so far
<8> i don't remember
<1> it's neat, it works, and it implements a ton of low level stuff so I don't have to be bothered with it
<8> nice
<8> maybe i will take a look at it after all
<8> but not for now -- too busy at work, where .NET is out of the question (we're trying to squeeze as much performance as we can from the C++ code so that we don't have to buy more machines -- i don't think .net would be an improvement)
<9> it may be close
<8> actually, buying machines would be cheaper, but we can't subdivide the tasks finely enough yet
<9> .net's more productive environment for sure
<9> speedwise, it's pretty good, especially on the server side
<1> .net isn't slow
<1> aye
<1> and it makes it easy to do things like split tasks and work at a higher level overall
<9> once the stuff gets touched once, it's all JIT-ted
<1> though that doesn't mean it's good enough for whatever you're trying to do
<8> etoya: that is hardly the issue here. our problem is that we can't disseminate quotes fast enough for one trader using one machine, and the application can't be split across several machines for one given trader
<1> what's the bottleneck?
<8> parsing incoming option market data and computing the theoretical prices for options
<1> what format is the data in when it comes in?
<8> some proprietary stuff, it's pretty much text
<1> xml or more plain?
<8> plain
<1> nod
<8> xml would have been worse, i think
<1> how much data per second are you trying to parse?
<1> you can't split up the parsing?
<9> computational stuff might be best done in c++ native
<1> you should at least be able to split the parsing up into threads
<8> etoya: my point exactly
<1> i wouldn't make a blanket statement like that
<8> rdragon: the threads won't help much, unless we get them to run on other machines
<1> remember, MSIL *is* compiled to native code
<9> bottom line, noone knows where the bottleneck is, till it gets measured
<1> i just asked him where it was, and he said the parser
<8> parser and mathematical models are the main cpu hogs
<9> Cd: did you profile it in detail ?
<1> why won't threads help? single cpu machines?
<8> the parser we can alleviate by changing the protocol to something that requires much less work and doing the conversion on a separate box (hardware is cheap)
<1> oh, you have control over the protocol
<8> no, but we're already filling the two cpus in our machines
<1> then I guess that problem is solved... whats this 'math model' ?
<8> that's the main cpu hog, it's a "binomial model" that calculates option prices
<9> Cd: run kernrate on it
<1> is this by any chance software that gets advertised on infomercials? ;)
<8> under normal market conditions, a cache helps that calculation a lot, but when the things start to move a lot, the cache doesn't hit nearly often enough to make a difference, it even slows doewn things a bit
<8> rdragon: no :)
<1> k
<9> Cd: so when you profile the code, you may find out, for example, that a lot of time is spent re/allocation memory
<9> or locking
<9> or context switching
<9> or something else that can be improved
<1> well, it shouldn't be too much trouble to distribute the task at a high level
<8> so a cache reduces cpu usage from 10% to, say, 6%, but from 90% to 95% :)
<8> EtoYa: you are right, in principle, but we've been through that, and now it's down to the parser and the models, which actually do work that require cpu. pretty much everything else flows through effortlessly
<9> ok
<8> rdragon: that's what i'll be working on as soon as i'm done with an urgent feature request, splitting it up; there are some obvious points where it can be broken into components that can run separately (threads, processes on different machines)
<1> nod
<8> there will be some extra work in putting all data back together, but if i can do 10% extra work and use six CPUs instead of two, i'll be happy
<9> yeah, it seems like this type of calculations would be easy to parallel
<8> there are some details that are not that easy, and the initial design did not consider any type of parallelism, so while it is not difficult to see how to do it, it would take months to implement
<8> and i don't have months :)
<9> ya.. it's never easy
<9> just seems if you have, say, 10 stocks, it should be easy to subdivide the math, and process 5 here, and 5 on a different machine or something
<1> not if it was designed without that in mind from the start, heh
Return to
#c++
or
Go to some related
logs:
#chatzone
eviI side
#linux
#linux
#AllNiteCafe
#london
was fond of saying 'I will gladly pay you Tuesday for a hamburger today'
#MissKitten
#c++
#linuxhelp
|
|