Re: 64-bit Windows for AMD 64 is here...
From: Brian Moelk (bmoelk_at_NObrainendeavorSPAM.FORcomME)
Date: 03/15/04
- Next message: TObject: "Re: Delphi 8 common quality"
- Previous message: Rudy Velthuis (TeamB): "Re: 64-bit Windows for AMD 64 is here..."
- In reply to: aLBi: "Re: 64-bit Windows for AMD 64 is here..."
- Next in thread: aLBi: "Re: 64-bit Windows for AMD 64 is here..."
- Reply: aLBi: "Re: 64-bit Windows for AMD 64 is here..."
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Mon, 15 Mar 2004 13:00:55 -0500
> Brian, first do NOT call Win64 a new technology!
Is "newer" ok? <g> I believe my points are still valid.
> What is new in Win64 for
> a developer to adapt to? Just the size of pointer and word, which would be
> insane to call them problematic to the porting of a software.
I don't know what the specific issues are since I've never done a Win32 ->
Win64 port. In theory, yes it should be fairly straightforward, however the
devil is in the details.
> Please, note that my strongest point in requiring a 64bit compiler is not
> that we can calculate faster with numbers larger than 2^32. It is that as
> the world will move to 64bit, 32bit application will be isolated. They
> will not have access to the latest technology. And you know, someone who
> needs to build server apps cannot afford stand alone.
Noted.
> Servers are made to
> communicate. Delphi became so widespread in corporations because it
> offered easy ways to take advantage of new technologies immediately after
> their release. I must admit again that if Delphi for Win64 would be
> released in 2 years the situation would be a lot different. Delphi 6 is
> still good for todays standarts (maybe with some work-arounds). But
> telling me that Delphi 6 will be good in 10 years is a little hard to
> believe.
I don't recall mentioning 10 years? I don't think Borland can wait that
long to support 64-bit. However they may only support 64-bit through
dotNET, which would be dissapointing to me. But this all depends on the
market. Borland is there to make money, if they can make money with a
64-bit native compiled product why wouldn't they do it?
> And starting a new project is not like maintaining one.
Agreed.
> We are still
> maintaning a complex visual application for the university here. Delphi
> with Graphics32 was simply the perfect solution for our programming needs.
> I can't imagine how much we would have had to do if we went with VC++.
> However, now that they upgraded the servers to 4-way Opterons they
> requested us to upgrade the software to 64bit by the end of the year with
> an additional payment. We don't know what to tell them! We're still
> looking for possibilities.
Sounds like an unfortunate situation. If your customer and your application
can indeed benefit from a native compiled 64-bit version you're out of luck
with Delphi (at this time from what we know).
But, this makes me wonder considering your previous statements. I believe
you stated your main reasons for 64-bit was not for performance, but to keep
pace with the 64-bit world. This is what I gather from the paragraph where
you say: "Please, note that my strongest point in requiring a 64bit
compiler is...". It strikes me as odd that you would want to move to
64-bit at all considering that it performs the task adequately now and
Delphi has demonstrated a significant business advantage.
The only reason that I can think of is to make your customer happy about
spending the money on a 4-way Opteron system. But that money has already
been spent hasn't it? Did you ask them to upgrade? Did they want to get
better performance out of the application?
> And do you think it would be correct to build a scientific application in
> .NET, where "integer" is class and you have to handle to the garbage
> collector every memory deallocation task (which is pretty slow)?
Agreed. IMO, dotNET isn't for scientific applications. That's not been the
point of this discussion so far, FMPOV.
> algorithms. We wanted, we needed Delphi (<Brian wrote: That may be *your*
> situation, not everyone elses.!> - You are right). But a fully optimized
> 64bit processor offers a much better performance in imaging because it can
> process almost twice as much data (it is hard, but it can be achieved!).
> So, how would our software stand against Adobe, Jasc, Ulead etc. in 2007
> (the release year we planned)? It would be an ugly, incompatible, slow,
> overpriced failure. Sorry to admit that. Our dreams vanished :(
That's an unfortunate situation. But this indicates to me that you are
hitting performance limitations with 32-bit technology; which conflicts with
your previous statements.
Regardless, if you need to go 64-bit, I'm not sure what to do, but since the
only viable 64-bit solution I know of today is C++, I'd start there and see
how much effort it would take.
> btw, we are not working with PowerBASIC at all. We are currently maintaing
> our Delphi apps and mastering Win32 development on VC++2003.
> Some friends and I bought a copy of the PowerBASIC compiler (we copied it
> ilegally, but just for testing:) and messed with it to see what
> performance would BASIC give if compiled. But I am sure that they will
> soon (next year) release a 64bit compiler, because that compiler is all
> what the company does.
What makes you so sure? There are no published plans. Did you talk to
their developers?
> If a compiler is well-designed, porting it to 64bit
> is an overnight task, since the eight additional registers have no new
> functionality, just to store more memory on the CPU. Although for Delphi
> it is much harder, since it has a huge library to be checked for
> incompatibilities and is also assembly optimized.
I thought the porting issues for Win32->Win64 were "Just the size of pointer
and word, which would be insane to call them problematic to the porting of a
software."? If that is the case, then the core issue is the compiler is it
not?
I find it strange that you have a good deal of faith in PowerBASIC, but yet
very little in Borland.
-- Brian Moelk bmoelk@NObrainendeavorSPAM.FORcomME http://www.brainendeavor.com
- Next message: TObject: "Re: Delphi 8 common quality"
- Previous message: Rudy Velthuis (TeamB): "Re: 64-bit Windows for AMD 64 is here..."
- In reply to: aLBi: "Re: 64-bit Windows for AMD 64 is here..."
- Next in thread: aLBi: "Re: 64-bit Windows for AMD 64 is here..."
- Reply: aLBi: "Re: 64-bit Windows for AMD 64 is here..."
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
