[Trial]Contest 1 - VB (.NET & Legacy) - Prime Numbers

Printable View

Show 60 post(s) from this thread on one page

Sep 14th, 2004, 02:12 PM
Pino

Re: Re: Assumptions

Quote:

Originally posted by ntg
This is not a solution to a generalized problem, so I don't think that you would break any rules if you use 7919. After all, you can use prime theory to find that out without actually calculating all primes up to 7919. The scope of the contest is very limited so I wouldn't think that anybody would mind your using the number, unless of course Electroman says otherwise. If the contest was about calculating the first n primes (n entered by user and always larger than 1,000,000), that would be a different story, if you needed the last prime you would actually have to calculate it on the spot and it would affect your timing. But then again, in such a case I'm willing to bet that most of our algorithms would break or would be slower than 50,000,000 primes per second.

BTW, last accomplished time is 0,025 ms. This was entertaining, looking forward to see Merri's code and for the next contest.

Cheers,
NTG

no its fine to inlclude that number, you could hard code ALL of the prime numbers and just display them, but your code wouldnt get many marks for sexynesss :) i think most people will be using one or two hard coded primes :)
Sep 14th, 2004, 02:47 PM
wossname

:stern tone of voice:

Now listen here electroman (sir)!

We, being expert professional coders and supergeeks, need to be reassured that the judges doing the timings are going to do a proper job. Not just some boy scout rubbing two sticks together. I too recommend the Mentalis.Stopwatch class.

Please inform us as to what is going to be done. [angry customer]If you don't know then please find out.[/angry customer] :D

Don't worry I still love you :lol:
Sep 14th, 2004, 08:55 PM
mendhak

AFAIK, just two people doing it in VB.NET: Me and wossname.

I'm really not sure why. I thought most of us would be doing it in VB.NET instead of VB6. I guess I guessed wrong. :p

Oh, and the Mentalis stopwatch class! Use it!
Sep 14th, 2004, 09:02 PM
Merri

That is because VB6 is so much better than VB.NET

And I guess VB6 still has the weight of being ranked as a slow language, so people are more likely to try to get to the edge of it :D
Sep 14th, 2004, 09:27 PM
dsheller

I wittled it down to 0.3ms (compiled) for the generation of the primes, how far that place me in the lineup heh =)
Sep 14th, 2004, 11:57 PM
mendhak

Quote:

Originally posted by Merri
That is because VB6 is so much better than VB.NET

Let's save our boxing match for later, shall we ;)
Sep 15th, 2004, 12:35 AM
Ex-FB

In fairness, .Net does have a number of great features. But when it comes down to this sort of performance, JIT does suck the hind teet a bit compared to VB6. It's not such an issue in most jobs nowadays, but for a problem like this, it would seem prudent to go the VB6 root if you were comparing like for like.

Of course, they're not, so it doesn't really matter here... :p

* Waits for the fastest .Net solution to be faster than the fastest VB6 one... that learn me :p
Sep 15th, 2004, 02:17 AM
BodwadUK

So what timing method is going to be used in VB6? :confused:
Sep 15th, 2004, 02:53 AM
ntg

Quote:

Originally posted by Ex-FB
In fairness, .Net does have a number of great features. But when it comes down to this sort of performance, JIT does suck the hind teet a bit compared to VB6. It's not such an issue in most jobs nowadays, but for a problem like this, it would seem prudent to go the VB6 root if you were comparing like for like.

Of course, they're not, so it doesn't really matter here... :p

* Waits for the fastest .Net solution to be faster than the fastest VB6 one... that learn me :p

To tell you the truth, I feel that if we're looking to such a limited problem and want to squeeze every last drop of performance, comparing VB.Net to VB6 doesn't make much sense. If it's that critical, C and ASM are the only viable options.

Regarding this little contest, I was curious and coded my algorithm in both VB.Net and VB6. VB6 actually came up 5% faster in the end and that's pretty marginal to me. Of course, I'm a bit biased towards business applications where this kind of speed improvements are easily counterbalanced by OOP features and a full-blown and coherent class library.

Don't get me wrong, I still love VB6 and I've used it for several projects in the past. But consider this: if the contest was to find the first 1,000,000 primes on a computer with hyperthreading, in VB.Net I'd spawn a couple of threads in three lines of code and split the calculation in half, something that I could never do in VB6 (or at least not without the use of a pricey external library or extensive arcane hacks).

Cheers,
NTG
Sep 15th, 2004, 03:02 AM
Wokawidget

You could do it in VB6 using the multithreading link in my sig :)

I was gonna enter, but I have been told there is much confusion regarding the marking, and couldn't be bothered getting dragged into a pointless argument :D

0.3ms Hmmmmmmmmmmmm
I am intregued on how you did that since adding 1000 0's to a listbox is WAY slower than that, and that's using API and hiding the listbox.

Wopof
Sep 15th, 2004, 03:09 AM
NoteMe

Quote:

Originally posted by Wokawidget
You could do it in VB6 using the multithreading link in my sig :)

I was gonna enter, but I have been told there is much confusion regarding the marking, and couldn't be bothered getting dragged into a pointless argument :D

0.3ms Hmmmmmmmmmmmm
I am intregued on how you did that since adding 1000 0's to a listbox is WAY slower than that, and that's using API and hiding the listbox.

Wopof

Hmmm...he didn't say it was adding the nubers to the list box....;)
Sep 15th, 2004, 03:09 AM
NoteMe

Quote:

Originally posted by BodwadUK
So what timing method is going to be used in VB6? :confused:

The sTimer class I posted....
Sep 15th, 2004, 03:15 AM
Merri

It makes no sense to time with listbox, because the time it takes to add items to listbox varies a lot. I even got bigger results most of the time for my faster algorithm with listbox, as weird as that sounds. I only optimized the algorithm, I didn't optimize initialization (because I wanted flexibility and the algorithm at top speed) and viewing of data (because I didn't want to code my own listbox... ;)).

The only problem with the algorithm I submitted is memory usage: it eats up a lot of memory. I couldn't get a memory saving version to time below 0,49 ms.

It takes almost no time at all to add to a listbox 1000 times :confused:
VB Code:
Dim t As Single
Dim I As Long
 
    t = Timer
    
    For I = 0 To 999
        List1.AddItem I
    Next I
    
    Me.Caption = Timer - t
seems to come up with minus numbers most the time :eek:

Sep 15th, 2004, 03:59 AM
NoteMe

Not sure what that function is wrapped around, it doesn't say so in my old MSDN. But compiled with all the code on a form, with all the advanced optimization on that code you posted takes 14-15ms on my 1900+.

But I have also found out that the various implementations varies much on diffrent machines. Like the ones with a lot of branching P4 machines are getting realy big numbers, while they are doing great on older machines.

Like my code now does 3.7ms on my machine, 2.1ms on a 2.4Ghz, then 13MS on Electromans SUPER fast PC....so thats it for me....no more optimization for me....as long as I have no way to test it on the computer it will be tested on in the end, there is realy no way of telling when it is fast, and when it is not fast.....

I hate the 20 pipes on a P4....*grrrrrrr*
Sep 15th, 2004, 04:03 AM
Wokawidget

Damn...I was adding 10,000 numbers to my listbox :D

Bad Woka *SLAP*
Sep 15th, 2004, 04:05 AM
BodwadUK

:confused:

Oh well

*WHACK SLAP KICK*

Bad Woka get back in your set and dont come out until I say so. No dinner for you :mad:
Sep 15th, 2004, 04:09 AM
NoteMe

Quote:

Originally posted by Wokawidget
Damn...I was adding 10,000 numbers to my listbox :D

Bad Woka *SLAP*

I told you that last time...don't you remember that? On MSN.....sick dog...sick dog...
Sep 15th, 2004, 04:54 AM
riis

I can't get the prime calculation below 15 ms on my 1.4 GHz laptop... :sigh: :(
The time including display them in the listbox is50 ms.
Sep 15th, 2004, 05:24 AM
mendhak

Quote:

Originally posted by riis
I can't get the prime calculation below 15 ms on my 1.4 GHz laptop... :sigh: :(
The time including display them in the listbox is50 ms.

Change your algo.
Sep 15th, 2004, 05:49 AM
NoteMe

And set all the optimization you can....forget about the array boundary checks and so on....
Sep 15th, 2004, 06:15 AM
wossname

GET BACK TO WORK BODWAD!!

Noteme: what version of VB are you using?
Sep 15th, 2004, 06:43 AM
NoteMe

6.0....why?
Sep 15th, 2004, 07:07 AM
ntg

Quote:

Originally posted by Wokawidget
You could do it in VB6 using the multithreading link in my sig :)
Wopof

Yeah, that's what I meant by "extensive arcane hacks". It's a great job you've done there, but it took you 8 months. It took me half that time to design, build, test and deploy a complete .Net business solution for a bank. That's what I'm talking about, all that time and energy could be spend more productively. Plus, I really don't know what would the results be if you were to actually measure performance of threading code powered by the .Net framework and threading code powered by your library. That's *not* a criticism towards your code, what I'm saying is that a lot of dev effort has gone into the framework and it's only natural to expect that it would be fine-tuned to the core.

Cheers,
NTG
Sep 15th, 2004, 07:18 AM
Wokawidget

Quote:

Originally posted by ntg
Yeah, that's what I meant by "extensive arcane hacks". It's a great job you've done there, but it took you 8 months. It took me half that time to design, build, test and deploy a complete .Net business solution for a bank. That's what I'm talking about, all that time and energy could be spend more productively. Plus, I really don't know what would the results be if you were to actually measure performance of threading code powered by the .Net framework and threading code powered by your library. That's *not* a criticism towards your code, what I'm saying is that a lot of dev effort has gone into the framework and it's only natural to expect that it would be fine-tuned to the core.

Cheers,
NTG

I 100% agree with you :D
Sep 15th, 2004, 07:52 AM
Merri

NTG: But you forget that now we do have Woka's code here and we can more or less easily add it to our projects without the eight month delay if we need multithreading :D

When I happen to do bigger projects, I can use old code and other stuff I have done before, which basically gives me all the functionality and ease of coding VB.NET could give. Plus stuff .NET framework can't give me - of course it would be possible to recode the stuff for .NET, but that is what takes the time and effort.

But yeah, this goes a lot to personal preference. I like to use "good old stuff" until it is near useless. For example, I used C64 regularly until the end of 1997, even though I had had PC for two years or so. Why? I got VB4 and that finally took away the last reason to use C64 :)

Btw, I'm still looking for a faster algorithm :D
Sep 15th, 2004, 08:25 AM
ntg

Quote:

Originally posted by Merri
NTG: But you forget that now we do have Woka's code here and we can more or less easily add it to our projects without the eight month delay if we need multithreading :D

When I happen to do bigger projects, I can use old code and other stuff I have done before, which basically gives me all the functionality and ease of coding VB.NET could give. Plus stuff .NET framework can't give me - of course it would be possible to recode the stuff for .NET, but that is what takes the time and effort.

But yeah, this goes a lot to personal preference. I like to use "good old stuff" until it is near useless. For example, I used C64 regularly until the end of 1997, even though I had had PC for two years or so. Why? I got VB4 and that finally took away the last reason to use C64 :)

Btw, I'm still looking for a faster algorithm :D

The last thing I want to be a part of is VB.Net vs VB6 forum war. Although I agree about reusability, my point about "arcane hacks" still applies. Yeah, you can use it but to do that you have to understand it and you have to code for using it. Take a long, hard look at what you need to do in order to comprehend and use that library (again, this is not a criticism of what I think is a nice piece of code). You end up putting a lot of effort to both understand and use it. The equivalent .Net approach is more simplistic, easier to learn and use. It's done away with the WinAPI and COM and that kind of stuff, that generally suck and/or are hard to master. Two months ago I hired a guy that had a solid programming background but extremely limited experience in .Net. Today, the same guy can enhance, modify and support 150k lines of code of both VB.Net and C#. With VB6, there was no way this could happen. And about functionality, we had to code for file I/O, DB operations, TCP/IP communications, remoting, web services, GUI interfaces, web interfaces, threading, sorting, merging, searching and plug-in DLL loading without once resorting to the WinAPI or an external component. That kind of power goes a long way. And, like it or not, support for VB6 and the old version of Studio is ultimately going to fade away for good - it would have already be entirely gone if it wasn't for the huge user base of the products. It's ok if you're a hobbyist (great actually, not just "ok"), but at some point in time it's going to be impossible to commercially develop in VB6 - it's already a construct that doesn't have strong enough foundations to support the plethora of additional features that seem to be attached to it using duct tape (see SOAP toolkit). At some point in time, saying that you develop commercially in VB6 is going to be today's equivalent of saying you're commercially developing in VB4. Existent, but tiny and shrinking market.

What you say about personal preference is true though. It was fun using the BBC and the C64 although it was a b*tch to wait for the tape ;) (boy, r we giving away our age). Myself, I used to be a Turbo Pascal fanatic until I gave it away when Windows killed DOS for good. I loved VB4/5/6 for several reasons, but for me it's time to move on. VB.Net is my current choice, other people choose something else whether it's C# or Java or Delphi or whatever, but it's definitely time to move.
Sep 15th, 2004, 09:06 AM
riis

Quote:

Originally posted by NoteMe
And set all the optimization you can....forget about the array boundary checks and so on....

That only made a very slight difference.
Changing the algorithm might work, but this was just my first attempt, hacked in a moment while I was fed up with work.
Sep 15th, 2004, 09:26 AM
NoteMe

Quote:

Originally posted by Merri
....Blah blah blah....

Quote:

Originally posted by ntg
....Blah blah blah....

Very nice said...blah blah blah...but as I say to everyone that is arguing VB6 vs. VB.NET or VB vs C++ or something like that back and forth...."Finding the right tool for the job is the answer....not trying to fit a tool for a certain job"...

VB6 doesn't have answers to all our promlebs, and VB.NET doesn't have all the answers to our problems....end of discussion...:)
Sep 15th, 2004, 09:36 AM
ntg

Quote:

Originally posted by NoteMe
Very nice said...blah blah blah...but as I say to everyone that is arguing VB6 vs. VB.NET or VB vs C++ or something like that back and forth...."Finding the right tool for the job is the answer....not trying to fit a tool for a certain job"...

VB6 doesn't have answers to all our promlebs, and VB.NET doesn't have all the answers to our problems....end of discussion...:)

Well said.:thumb:
Sep 15th, 2004, 09:38 AM
BodwadUK

Vb 6 solves more though :D
Sep 15th, 2004, 09:53 AM
NoteMe

Quote:

Originally posted by ntg
Well said.:thumb:

Thanks..small kids do also have brains...:)...well I have the last 7 years twisted that quote around to fit what I belive in. It was the authour of my first VB book that gave me the quotes.

"Finding the right tool for the job"

and

"Don't reinvent the wheel"

Well guess he didn't make up those either....at least not the last one...:)
Sep 15th, 2004, 09:58 AM
Pino

http://www.vbforums.com/showthread.p...78#post1788178

Would be a better place to discuss these things :)

as for the contest, Your running out of time people get them entries in :)

and incase anyone is wondering we are working on making the next contest a bit more 'interesting'
Sep 15th, 2004, 10:04 AM
NoteMe

Can you give us any hints on the next one?
Sep 15th, 2004, 10:39 AM
Pino

heheh concentrate on this one first! the next one is top secret infomation
Sep 15th, 2004, 12:59 PM
Tec-Nico

Oh no! Electroman, Pino... :(

I was about to post mine... The deadline is today, isn't it?
Sep 15th, 2004, 01:06 PM
Ex-FB

The deadline is 10 days away... (25th September).
Sep 15th, 2004, 01:09 PM
Pino

Quote:

Originally posted by Tec-Nico
Oh no! Electroman, Pino... :(

I was about to post mine... The deadline is today, isn't it?

Plenty of time yet :) 10 days left to perfect your code :)
Sep 15th, 2004, 01:09 PM
Merri

Ten days is soo long time to wait...

Btw, I noticed a small "bug" in my code, the one I posted: by removing two lines and adding their calculation into another line (two characters), the algorithm goes below 0.02 ms on both of my machines. Oh well.
Sep 15th, 2004, 02:10 PM
Tec-Nico

Thanks, Pino... :)

I hope I will get the best out of me for that day.

Hmmm.. Just a bit off topic (And out of curiosity) which was the avatar VisualAd used to have that made Brad say that? (What NoteMe is quoting)

Show 60 post(s) from this thread on one page