Blizzard might have written it, but that doesn't make me buy their games:
- Their quality is of course nr 1. - The broad support and usage is two, and bnetd
servers for intranets/lanparties, or places with
low internet connectivity is also important.
Blizzard screws up the last one: Their games just went down in value, and it should be quite cool before I buy one again.
Yes, but the entry is much less. An own main processor is a much bigger barrier to entry than a customized mobo on commodity chips.
One can easier step into that market than invest the 1E9 dollars.
It is not that you can buy the Itanium machine in the shop on the corner, but there might come more companies that can supply them, which will result in a price drop.
Nautilus compiles and runs afaik, but is a bit shaky.
I ran make install in and Gnumric and Nautilus ports trees just to see it, and it compiled instantly and
ran. Both of them, with no Gnome installed before.
(about a month ago btw)
However IMAP functionality of Nautilus crashed, GNumeric ran fine.
The frequent inbetween releases for linux kernels
are to be able to have some more broad testing, since
keeping a "make world" environment working is somewhat more complicated with that much
distributions, embedded use etc.
But BSD isn't doing that bad. Yesterday installed FreeBSD 4.5 on my laptop, and to my surprise it detected all devices of both laptop and docking station.
It's an older laptop, but still....
- D6 supports OO afaik.
- limited MI via interfaces. (which is all you need anyway)
- By using the same interfaces, one can also release difficult datastructures automatically.
Free Pascal can be used sometimes to compile for other x86 platforms too.
That is a bit light on reasons. Why is it inflexible?
Or did you never really look at it?
I really like the language, and like the calm, clear IDE. A lot of IDEs are getting to baroque, and most
of those baroque features are only funny to show to
friends or use in magazine reviews. Not to work with.
.NET and Java are not an option for me, performance wise, and because of deployment.
VC++ is an option, and the VS studio IDE is nice,
but I don't like the language and the libraries.
For some purposes it does matter (likewise with scarce floorspace where you want as much FPU power in running as cool possible in a box), for some it doesn't. A lot of clusters are run by students.
That your group somehow happened to have no unix experience, and a ton of PPC hardware laying around
is fine, but that doesn't justify the fact that
all discussion about price is left out. Since Unix
is well entrenched in most universities, and I will
have to search very well to find a simgle mac overhere.
Without naming price in this kind of articles, the picture is incomplete, and it is essentially misinformation
If you have software that runs on both (like Linux),
you pick the one with the most performance per $,
not PPC because it is has more performance per tick.
I can see one or two advantages:
- no static. Only the research papers are searched.
- Could make a strong negotiating position to get access (and retrieve via the portal) from archives not connected to the internet
However the article doesn't actually name things like that
It is (afaik) also not clear what the Lisp code competes against on a software level.
A large part of the article is about the cluster mainframe difference, the rest just says things like Lisp code "WITH BETTER ALGORITMS" vs old mainframe assembler.
It seems to get the job done, but doesn't reveal if that is because someone sat down, and did his math
homework properly, or because LISP is great.
A lot of problems in engineering still use a lot of
computational time, contrary to other domains where
speed isn't an issue anymore.
OOP programming is often slower than purely procedural, though a lot depends on the programmer (but will eat into ease of use)
I'm not 100% sure about the intrisic speed of functional programming, maybe you could make some comments on that?
How fast are current implementations, and are there reasons why proper functional programming would be systematically slower than proper functional programming?
The article is nice in its description of autism (
and not Rainman'ing too much), but detoriates in
the engineering thing.
It seems that the author is confusing symptom and cause.
Symptom: unsocial behaviour by average standards
Cause 1 : Neural disorder with failure to learn
social behaviour "autistic"
Cause 2 : Long time social contacts mostly with peers. Result: Social standards of group geeks
deviates from mass. Actual problem: none. Purely cultural.
But there is more than just rabbits.
IIRC Opuntia (some cactus used for fences),
a form of dessert grass (arrived in Aussie land as
packaging material for fences)
wild donkeys, horses and camels.
I was talking about a replacement for TP.
That doesn't mean that FPC can't be a good substitute
for Delphi for some tasks. (and standard RAD isn't one of them)
Most FPC programmers are also Delphi programmers that use FPC for more specialistic tasks.
Slashdot Mirror
Blizzard might have written it, but that doesn't make me buy their games:
- Their quality is of course nr 1.
- The broad support and usage is two, and bnetd
servers for intranets/lanparties, or places with
low internet connectivity is also important.
Blizzard screws up the last one: Their games just went down in value, and it should be quite cool
before I buy one again.
Yes, but the entry is much less. An own main processor is a much bigger barrier to entry than a customized mobo on commodity chips.
One can easier step into that market than invest
the 1E9 dollars.
It is not that you can buy the Itanium machine in the shop on the corner, but there might come more
companies that can supply them, which will result
in a price drop.
Nautilus compiles and runs afaik, but is a bit shaky.
I ran make install in and Gnumric and Nautilus ports trees just to see it, and it compiled instantly and
ran. Both of them, with no Gnome installed before.
(about a month ago btw)
However IMAP functionality of Nautilus crashed, GNumeric ran fine.
I'm BSD'er, but this is nonsense.
The frequent inbetween releases for linux kernels
are to be able to have some more broad testing, since
keeping a "make world" environment working is somewhat more complicated with that much
distributions, embedded use etc.
But BSD isn't doing that bad. Yesterday installed FreeBSD 4.5 on my laptop, and to my surprise it detected all devices of both laptop and docking station.
It's an older laptop, but still....
From the enhanced performance link:
:-)
* Access much larger memory spaces
Hihi, probably needed to keep SWING performing
- D6 supports OO afaik.
- limited MI via interfaces. (which is all you need anyway)
- By using the same interfaces, one can also release difficult datastructures automatically.
Free Pascal can be used sometimes to compile for other x86 platforms too.
That is a bit light on reasons. Why is it inflexible?
Or did you never really look at it?
I really like the language, and like the calm, clear IDE. A lot of IDEs are getting to baroque, and most
of those baroque features are only funny to show to
friends or use in magazine reviews. Not to work with.
.NET and Java are not an option for me, performance wise, and because of deployment.
VC++ is an option, and the VS studio IDE is nice,
but I don't like the language and the libraries.
"MY" cost and my "PRICE" don't matter.
For some purposes it does matter (likewise with scarce floorspace where you want as much FPU power in running as cool possible in a box), for some it doesn't. A lot of clusters are run by students.
That your group somehow happened to have no unix experience, and a ton of PPC hardware laying around
is fine, but that doesn't justify the fact that
all discussion about price is left out. Since Unix
is well entrenched in most universities, and I will
have to search very well to find a simgle mac overhere.
Without naming price in this kind of articles, the picture is incomplete, and it is essentially misinformation
Of course it isn't a troll.
If you have software that runs on both (like Linux),
you pick the one with the most performance per $,
not PPC because it is has more performance per tick.
For clusters this is a serious calculation
Typical mac article. All about "performance", and
performance/clock, but nothing about performance/price.
The simple fact that they don't mention it, says enough.
The whole idea of commodity hardware clusters revolves around the lower cost per gflop.
Using PPC that is 2 times as fast and 4 times as
expensive makes it uninteresting for a large group
of researches.
Some will choose PPC because less nodes is less spaces, but most will go for costreduction, and stay with intel (or better: Amd)
The explanation on how to achieve this reads a bit funny.
It seems to assume that if one mixes two transparent
components (e.g. glass grid, and some transparant matrix), the result is also transparent.
This is not true, as every high school boy that studied optics can tell you. Refraction index, surface properties etc.
It will probably be pretty hard to make a transparant material from two components, let alone keep the other properties of concrete.
Seem to target research/academic papers mainly.
I can see one or two advantages:
- no static. Only the research papers are searched.
- Could make a strong negotiating position to get access (and retrieve via the portal) from archives not connected to the internet
However the article doesn't actually name things like that
Sorry, replied to the wrong msg.
Should have gone to one level higher.
Point is:
Any sane person can choose those language for
performance too, not just C.
There are of course more languages, the point is that
C is not holy in that respect.
It is (afaik) also not clear what the Lisp code competes against on a software level.
A large part of the article is about the cluster mainframe difference, the rest just says things like Lisp code "WITH BETTER ALGORITMS" vs old mainframe assembler.
It seems to get the job done, but doesn't reveal if that is because someone sat down, and did his math
homework properly, or because LISP is great.
(And I think it is the math thing
If it lags behind C, it lags behind Pascal and Fortran too.
A lot of problems in engineering still use a lot of
computational time, contrary to other domains where
speed isn't an issue anymore.
OOP programming is often slower than purely procedural, though a lot depends on the programmer (but will eat into ease of use)
I'm not 100% sure about the intrisic speed of functional programming, maybe you could make some comments on that?
How fast are current implementations, and are there reasons why proper functional programming would be systematically slower than proper functional programming?
The article is nice in its description of autism (
and not Rainman'ing too much), but detoriates in
the engineering thing.
It seems that the author is confusing symptom and cause.
Symptom: unsocial behaviour by average standards
Cause 1 : Neural disorder with failure to learn
social behaviour "autistic"
Cause 2 : Long time social contacts mostly with peers. Result: Social standards of group geeks
deviates from mass. Actual problem: none. Purely cultural.
Cause 1 Cause 2
They are pretty much the same species :-)
The devastating part is the culture, not the species.
Though maybe even the evolutionary aspect of the
arrival of the Aboriginals is still not done yet.
Nature is slow...
I wouldn't take Jurassic Park too seriously
But there is more than just rabbits.
IIRC Opuntia (some cactus used for fences),
a form of dessert grass (arrived in Aussie land as
packaging material for fences)
wild donkeys, horses and camels.
Note that I would increase that Windows memory requirement for Win2000 and XP.
(these OSes use more memory by default, less mem for apps)
Some current students will get a bag full of money,
but all next generations will have to sign something
that waives their rights.
The only result will be a twisted working relationship (Is this yours or mine?).
Simply attribute all rights to the University would
be a better choice IMHO.
See your own subject :-)
I was talking about a replacement for TP.
That doesn't mean that FPC can't be a good substitute
for Delphi for some tasks. (and standard RAD isn't one of them)
Most FPC programmers are also Delphi programmers that use FPC for more specialistic tasks.
Note that I didn't mean to imply that C/C++ is better
than Pascal.
But there are some rare C/C++ features that don't have a direct eq in Object Pascal. If they are really
useful is a matter of taste and implementation
Most pascal bashers can't even tell why C/C++ is more
superior than an advanced dialect like Delphi.
One of the major things is templates, which I'd love to
see in Delphi
Most people don't have numeric computational software that need the last penny.
Other computer subsystems (network, memory/cache,DISK) are the bottleneck.
There is only a very small market for ultra optimizing compilers, and you have to know them well
to get a real performance boost.