Re: Version after Version
- From: erewhon@xxxxxxxxxx (J French)
- Date: Sat, 29 Oct 2005 13:32:43 +0000 (UTC)
On Sat, 29 Oct 2005 14:36:27 +0200, "Frank de Groot"
<franciad@xxxxxxxxx> wrote:
>"Maarten Wiltink" <maarten@xxxxxxxxxxxxxxxxxx> wrote in message
>news:436360ab$0$11076$e4fe514c@xxxxxxxxxxxxxxxxx
>
>>
>> I tend to do my operations one method call at a time, not MxN bits at a
>> time.
>Are you saing that when you need to, say, add 1000 numbers in an array, thay
>you do this:
>for i := 0 to High(anArray) do
> AddNextNumber(anArray, i);
I doubt that Maarten would do that
- but if they are 4 byte integers there would be a /loss/ using a 64
bit chip
- and 1000 8 byte numbers is a bit silly if they don't need to be
Int64s
>That would be insane.
>I am sure you are not *that* stupid.
>What you say is nonsensical. Every time you do more than on 32-bit operation
>on an array for example, you can make things much faster by loading 64 bits
>at a time from memory (as my dual-cpu, dual-core 275 machine does),
>operating on those 64 bits and write them back in one operation, to 64-bit
>wide memory.
That is only the case if you have everything 64 bit aligned
- which it ain't
- and you'll probably find that it will seldom be, as that is just
gobbling (doubling) memory usage.
>When you can't imagine that applications do operations on data structures in
>memory, when your own applications don't do such a thing, merely do a few
>"method calls" here & there and for the rest waiting for user input, I feel
>for you.
If you had written your own filing systems then you would be well
aware of the value of 'Packed'
>But please don't think that all desktop software is like yours. Most isn't.
>Some of the software I recently bought is ISObuster (or something). It
>extracts ISO images (a kind of decompressing). This benefits greatly from
>64-bits. Other software I bought is
>ASPack, it is an EXE compressor I use. This would benefit greatly from
>64-bits. Other software I bought is Ghost install, to make installs of my
>1.5 GB install package. The compression engine would benefit greatly from 64
>bits.
Don't count on it
- compression and decompression works at the Bit level
>Other software I bought is Absolute Database. I use ZIP compression (LZW) of
>some fields in my 2 GB of databases. This would benefit greatly from
>64-bits.
LZW is the until recently proprietory algorithm of Uinisys
The ZIP format uses Huffman compression
The 'Z' in LZW is, I believe, nothing to do with the Z in ZIP
>Etcetera, etcetera. You truly haven't got the slightest clue what you're
>talking about. EVERYTHING benefits GREATLY from 64 bits.
No it does not
- some things benefit, huge memory addressing and copying large chunks
of memory
- but most things don't benefit
>I want to go as far as prediction that a software company has no future if
>they do not start the transition to 64 bits NOW.
Well I know a very successful software company that had not migrated
from 16 bit to 32 bit 18 months ago, when they floated a minority
chunk of their equity to make a number of employees GBP millionaires.
( I was at the launch party - a right good shindig, nice seeing old
friends even though I had left 11 years before )
Of course the 'real' shareholders were already well in the Sunday
Times UK 'Rich List' - but I'll not go into details in public.
Incidentally they specialize in fast data interrogation and
presentation - using their own proprietory filing systems that are
designed to minimize disk reads and memory movement.
Personally I would have re-written the systems in 32 bit a long time
ago, but not to obtain speed since any disk based system is choked by
the File Transfer rate
- simply because I would have liked to get them back out of 'C' and
into a sensible development language.
Think about processor bus width like the frequency range on an audio
system
- at a certain stage the higher the frequency range the bigger the
yawn.
Huge addressing ranges might be Ok for servers, but that was around
for a long time on the VAX, and guess what, they had to back it up
with SRAM - it was just one big read cache.
- and the huge address range was generally down to it being needed by
incompetent programmers who did not know how to organize their data.
.
- Follow-Ups:
- Re: Version after Version
- From: Frank de Groot
- Re: Version after Version
- References:
- Version after Version
- From: ProfitMaxTrading.com
- Re: Version after Version
- From: Frank de Groot
- Re: Version after Version
- From: J French
- Re: Version after Version
- From: Frank de Groot
- Re: Version after Version
- From: Marc Rohloff
- Re: Version after Version
- From: Frank de Groot
- Re: Version after Version
- From: Marc Rohloff
- Re: Version after Version
- From: Frank de Groot
- Re: Version after Version
- From: Maarten Wiltink
- Re: Version after Version
- From: Frank de Groot
- Version after Version
- Prev by Date: Re: Version after Version
- Next by Date: Re: Delphi IDE Code Editor
- Previous by thread: Re: Version after Version
- Next by thread: Re: Version after Version
- Index(es):
Relevant Pages
|