Board index » cppbuilder » Re: XP Home also 64 Bit computing

Re: XP Home also 64 Bit computing


2005-08-06 04:38:17 AM
cppbuilder40
IIS is not available for download.
 
 

Re:Re: XP Home also 64 Bit computing

"tinyabs" < XXXX@XXXXX.COM >wrote in message
Quote
Get a FX57 with DDR600 just to compile code? I don't think so.
Why not ? <g>
Personally, I would pick a Pentium 4 Prescott 1MB L2 800FSB Hyper-threading
+ at least 512MB DDR400 + Seagate 80GB+ SATA, preferably 4 Seagate 80GB+
SATA RAID 5 coupled.
Jonathan
 

Re:Re: XP Home also 64 Bit computing

Quote
Why not ? <g>

Personally, I would pick a Pentium 4 Prescott 1MB L2 800FSB
Hyper-threading + at least 512MB DDR400 + Seagate 80GB+ SATA, preferably 4
Seagate 80GB+ SATA RAID 5 coupled.

I would rather spend money on a better compiler. And why would Sata raid 5
help in compilation? Even if I had 128 sata raid hdds, the speed is still
same. You need a very fast memory like DDR600 and a fast CPU to feel the
effect cos the compiler is looking up symbols most of the time. But I ain't
going to spend my money this way.
 

{smallsort}

Re:Re: XP Home also 64 Bit computing

"tinyabs" < XXXX@XXXXX.COM >wrote in message
Quote
I would rather spend money on a better compiler.
If one is using the VCL, there aren't exactly many compilers for that, is
there ? <g>
Quote
And why would Sata raid 5 help in compilation? Even if I had 128 sata
raid hdds, the speed is still same.
RAID 5 is 4 drives hooked up together in such a way that gives twice the
speed of one drive, with double backup. And because compiling is often
intensive HDD usage, then the fast drives should pay off.
Quote
You need a very fast memory like DDR600 and a fast CPU to feel the effect
cos the compiler is looking up symbols most of the time. But I ain't
going to spend my money this way.
I agree, but I think that a Prescott + DDR400 would do it.
Jonathan
 

Re:Re: XP Home also 64 Bit computing

As long as the drives are IDE drives then it should be over-complicated and
failure prone but otherwise I guess it would work.
Too hot, minimal memory, too many drives, processor selected for abilities
that compiling and linking do not use.
. Ed
Quote
Jonathan Benedicto wrote in message
news:42f3d090$ XXXX@XXXXX.COM ...

>Get a FX57 with DDR600 just to compile code? I don't think so.

Why not ? <g>

Personally, I would pick a Pentium 4 Prescott 1MB L2 800FSB
Hyper-threading + at least 512MB DDR400 + Seagate 80GB+ SATA, preferably 4
Seagate 80GB+ SATA RAID 5 coupled.
 

Re:Re: XP Home also 64 Bit computing

"Ed Mulroy" < XXXX@XXXXX.COM >wrote in message
Quote
As long as the drives are IDE drives then it should be over-complicated
and failure prone but otherwise I guess it would work.

Too hot, minimal memory, too many drives, processor selected for
abilities that compiling and linking do not use.
Thank you for your comments on my pick. ;-)
Though, if I built this machine, I would have two front 80mm fans, two rear
80mm fans, and a quality case. So heating shouldn't be a problem.
I'll agree 512MB is not that big, but is it really necessary to have a 1Gb
?
Why is this too many drives ? I believe that this is the hookup of drives
for the best performance / backup tradeoff.
I don't know what processor abilities compilers and linkers use, but you
did mention the Pentium M, and AFAIK this is just a slower, cooler variant
of the Prescott.
Jonathan
 

Re:Re: XP Home also 64 Bit computing

Quote
The ordinary Prescott is fast, you can see them because they have a meg of
L2 cache.
That's new news to me.
Quote
The Hyper-threading Prescott variant will speed up Windows, because it can
take 2 threads at once. It works on them one-at-a-time, doing little
chunks on one then the other etc.
When does it do this? I have not seen it. Some friends of mine complain
that it does not do this.
Quote
The Double-core, or Pentium D Prescott is the best yet. It actually is two
chips, and I would highly recommend this one for compiling, because even
BCB would take an advantage with it.
I never heard that it is two chips rather than multi-core. As long as the
code is both partitioned correctly for it and fed to the processor in a
manner consistent with it, then yes, it will kind of give maybe the
throughput of 1 1/4 or 1 /38 processors. However I do not know of any
applications that will take advantage of it.
Quote
>(statement about Pentium-M)
This is the benefit of the laptop, but it is said that they are not good
compilers because of the slower chip inside.
Yes, good compilers. For 80x86 code, faster than a Pentium III which in
turn is faster than a Pentium IV. If the Prescott is faster, that's
probably because of a larger cache since it's architecture, like the P4, is
not designed to be much better for 80x86 code than the Pentium II. What it
optimizes is multi-threading and to get benefit from that one needs to use
an operating system that is good at multi-threading, for example OS/2 but
not Linux (which does not have threads) and not Windows (which has finally
exceeded NT3.1's threading performance to where it can actually show more
than 5% gain from a 3rd processor).
. Ed
Quote
Jonathan Benedicto wrote in message
news:42f374bb$ XXXX@XXXXX.COM ...

>Prescott is the latest twist on the Pentium 4, P4, and I am a little

The ordinary Prescott is fast, you can see them because they have a meg of
L2 cache.

The Hyper-threading Prescott variant will speed up Windows, because it can
take 2 threads at once. It works on them one-at-a-time, doing little
chunks on one then the other etc.

The Double-core, or Pentium D Prescott is the best yet. It actually is two
chips, and I would highly recommend this one for compiling, because even
BCB would take an advantage with it.

>The PM seems to work at a good pace because of their internal
>architecture. I remember reading on ArsTechnica about the PM chips now
>being put out in desktop versions. I see it as the next 'sweet spot'
>after the Pentium III. The clock speed number itself doesn't really
>matter since how the clocks are used internally varies so widely between
>processors. What matters is how fast the machine runs. Compared to the
>non-Prescott P4, the PM's enhancements more address the instruction set
>that I normally use, that which compilers typically generate. I don't
>know if Prescott has made up for the difference.

I'm still working on a Celeron P3, but I know that the Prescott is the hot
and fast chip. By hot I mean literally, I think that it is the hottest
chip ever built. Uses more power than a Xeon I think.

>Note that although I use a laptop, the comment about the M is not related
>to speed throttling. That seems to not kick in in my normal usage which
>is with it plugged into the wall. The value of a laptop for me is that I
>can take it with me on the road and that it is oblivious to a power
>glitch. Other than during a power glitch or when I carry it over to
>someone's printer, I don't normally run it on battery.

This is the benefit of the laptop, but it is said that they are not good
compilers because of the slower chip inside.

Jonathan

 

Re:Re: XP Home also 64 Bit computing

"Ed Mulroy" < XXXX@XXXXX.COM >wrote in message
Quote
When does it do this? I have not seen it. Some friends of mine complain
that it does not do this.
AFAIK, it does it completely invisibly. Windows just thinks that it is two
chips, and feeds it two threads at a time.
Quote
I never heard that it is two chips rather than multi-core. As long as
the code is both partitioned correctly for it and fed to the processor in
a manner consistent with it, then yes, it will kind of give maybe the
throughput of 1 1/4 or 1 /38 processors. However I do not know of any
applications that will take advantage of it.
AFAIK, it is two chips on one die. Windows regards it as two chips, and
feeds it two threads at a time. However, unlike the HT, the D works on both
at one time.
Quote
Yes, good compilers. For 80x86 code, faster than a Pentium III which in
turn is faster than a Pentium IV. If the Prescott is faster, that's
probably because of a larger cache since it's architecture, like the P4,
is not designed to be much better for 80x86 code than the Pentium II.
What it optimizes is multi-threading and to get benefit from that one
needs to use an operating system that is good at multi-threading, for
example OS/2 but not Linux (which does not have threads) and not Windows
(which has finally exceeded NT3.1's threading performance to where it can
actually show more than 5% gain from a 3rd processor).
Why doesn't Linux have threads ?
Jonathan
 

Re:Re: XP Home also 64 Bit computing

Quote
Though, if I built this machine, I would have two front 80mm fans, two
rear 80mm fans, and a quality case. So heating shouldn't be a problem.
Why go to all that trouble? Save the money and take a laptop to an airport
near where the jets do their run up. <g>
Quote
I'll agree 512MB is not that big, but is it really necessary to have a 1Gb
?
I'm currently running 512 and am waiting for my 1 Gig to arrive. Windows
caches almost everything in sight and so does BCB. If you've a bit hunk of
RAM then things run a lot better. It is not uncommon to hear someone say
that doubling his RAM size gave better results than doubling his processor
speed.
Quote
Why is this too many drives ? I believe that this is the hookup of drives
for the best performance / backup tradeoff.
In my experience there are reliable mechanical devices and then there are
hard drives. I don't have copies of the studies anymore but in them the
statistics on failure rate versus number of hard drives showed failure rate
starting upward noticably at 3 and becoming steeply upward starting at 4.
SCSI drives were not more reliable than IDE except way back when IDE drives
were a new thing. Note that a RAID setup does not change this. If it is
properly configured, a subject in itself, then it masks the error and make
large drive arrays practical because you can continue on when there is a
failure. Also note that drives also generate a lot of heat and are
sensitive to the heat.
. Ed
Quote
Jonathan Benedicto wrote in message
news:42f3e80e$ XXXX@XXXXX.COM ...

>As long as the drives are IDE drives then it should be over-complicated
>and failure prone but otherwise I guess it would work.
>
>Too hot, minimal memory, too many drives, processor selected for
>abilities that compiling and linking do not use.

Thank you for your comments on my pick. ;-)

Though, if I built this machine, I would have two front 80mm fans, two
rear 80mm fans, and a quality case. So heating shouldn't be a problem.

I'll agree 512MB is not that big, but is it really necessary to have a 1Gb
?

Why is this too many drives ? I believe that this is the hookup of drives
for the best performance / backup tradeoff.

I don't know what processor abilities compilers and linkers use, but you
did mention the Pentium M, and AFAIK this is just a slower, cooler variant
of the Prescott.
 

Re:Re: XP Home also 64 Bit computing

Quote
Why doesn't Linux have threads ?
I don't know. They have a guru called Linus who came up with a lot of it.
Maybe someone knows his email address and if so, you can ask him.
Judging from what I've been told in these groups Linux programmers simulate
threads by spinning off processes. However processes are not handled in the
same way as threads and have higher overhead.
. Ed
Quote
Jonathan Benedicto wrote in message
news:42f3ec9c$ XXXX@XXXXX.COM ...
 

Re:Re: XP Home also 64 Bit computing

"Ed Mulroy" < XXXX@XXXXX.COM >wrote in message
Quote
I'm currently running 512 and am waiting for my 1 Gig to arrive. Windows
caches almost everything in sight and so does BCB. If you've a bit hunk
of RAM then things run a lot better. It is not uncommon to hear someone
say that doubling his RAM size gave better results than doubling his
processor speed.
Thank you for this info. I remember for when I build my next machine.
Quote
In my experience there are reliable mechanical devices and then there are
hard drives. I don't have copies of the studies anymore but in them the
statistics on failure rate versus number of hard drives showed failure
rate starting upward noticably at 3 and becoming steeply upward starting
at 4. SCSI drives were not more reliable than IDE except way back when
IDE drives were a new thing. Note that a RAID setup does not change
this. If it is properly configured, a subject in itself, then it masks
the error and make large drive arrays practical because you can continue
on when there is a failure. Also note that drives also generate a lot of
heat and are sensitive to the heat.
Thank you also for this info. I don't think I ever could afford 4 drives,
but in case I ever can, maybe reliability would have improved by then.
Jonathan
 

Re:Re: XP Home also 64 Bit computing

"Ed Mulroy" < XXXX@XXXXX.COM >wrote in message
Quote
I don't know. They have a guru called Linus who came up with a lot of
it. Maybe someone knows his email address and if so, you can ask him.

Judging from what I've been told in these groups Linux programmers
simulate threads by spinning off processes. However processes are not
handled in the same way as threads and have higher overhead.
That is interesting information. I thought that Unix much better than
Windows at running high thread counts.
Jonathan
 

Re:Re: XP Home also 64 Bit computing

Quote
Thank you also for this info. I don't think I ever could afford 4 drives,
but in case I ever can, maybe reliability would have improved by then.
Of course you can afford 4, $90 IDE drives. It's the fancy ones you can't
afford that many of.
Note that when you use RAID for drives then your storage capacity is much
smaller than the sum of the drive's capacities. In the simplest case, one
which doesn't give that much protection, 2 drives give you the capacity of
one. I worked with a drive farm at a railroad (think about trains hitting
each other at crossings and you'll immediately know their need for
reliability). Each RAID drive was actually 4 physical drives. The air
conditioning load from the drive array was massive!
Note that my information is a couple of years old. Hopefully someone who is
into how drives are behaving currently will jump in. (Andrue are you
reading this?)
. Ed
Quote
Jonathan Benedicto wrote in message
news:42f3eec1$ XXXX@XXXXX.COM ...

>I'm currently running 512 and am waiting for my 1 Gig to arrive. Windows
>caches almost everything in sight and so does BCB. If you've a bit hunk
>of RAM then things run a lot better. It is not uncommon to hear someone
>say that doubling his RAM size gave better results than doubling his
>processor speed.

Thank you for this info. I remember for when I build my next machine.

>In my experience there are reliable mechanical devices and then there are
>hard drives. I don't have copies of the studies anymore but in them the
>statistics on failure rate versus number of hard drives showed failure
>rate starting upward noticably at 3 and becoming steeply upward starting
>at 4. SCSI drives were not more reliable than IDE except way back when
>IDE drives were a new thing. Note that a RAID setup does not change
>this. If it is properly configured, a subject in itself, then it masks
>the error and make large drive arrays practical because you can continue
>on when there is a failure. Also note that drives also generate a lot of
>heat and are sensitive to the heat.

Thank you also for this info. I don't think I ever could afford 4 drives,
but in case I ever can, maybe reliability would have improved by then.
 

Re:Re: XP Home also 64 Bit computing

Quote
That is interesting information. I thought that Unix much better than
Windows at running high thread counts.
Note that I spoke of Linux. Unix and Linux are not the same thing. I do
not know about threads under any of the different operating systems that are
called Unix but I have not been into Unix in a very long time and when I was
it was with the Microport/UDL version and with AIX. BSD, FreeBSD, OpenBSD
and NetBSD are likely to be very different.
. Ed
Quote
Jonathan Benedicto wrote in message
news:42f3efc1$ XXXX@XXXXX.COM ...

>I don't know. They have a guru called Linus who came up with a lot of
>it. Maybe someone knows his email address and if so, you can ask him.
>
>Judging from what I've been told in these groups Linux programmers
>simulate threads by spinning off processes. However processes are not
>handled in the same way as threads and have higher overhead.

That is interesting information. I thought that Unix much better than
Windows at running high thread counts.
 

Re:Re: XP Home also 64 Bit computing

"Ed Mulroy" < XXXX@XXXXX.COM >wrote in message
Quote
Of course you can afford 4, $90 IDE drives. It's the fancy ones you
can't afford that many of.
I guess. I hadn't looked at the current prices, and I was going on a book
from 04 (I think). Yes, I can get Seagate 80GB SATA's for $80.
Quote
Note that when you use RAID for drives then your storage capacity is much
smaller than the sum of the drive's capacities. In the simplest case,
one which doesn't give that much protection, 2 drives give you the
capacity of one. I worked with a drive farm at a railroad (think about
trains hitting each other at crossings and you'll immediately know their
need for reliability). Each RAID drive was actually 4 physical drives.
The air conditioning load from the drive array was massive!
In compiling would the extra performance of RAID really help at all ?
Jonathan