- Re: Merge Sorting
Hi.
The nearest I could come was code from "Software Tools", Kernighan and
Plauger, around page 117. The code is in ratfor, but should be easily
translatable into (say) f77. It will require some associated support
routines. I seem to recall doing the translation as I typed (from
ratfor to Minnesota mnf / m77). I also recall from those days that
- nike clothing,nike shoes,online store,http://www.maidi2008.net air Jordan Shoes,Nike Air Jordans, Air Force Ones,Retro Air Jordan, Bape Hoodies,Bape shoes-Tencent Traveler|Jordan Shoes,Nike Air Jordans, Air Force Ones,Retro Air Jordan, Bape Hood
nike clothing,nike shoes,online store,[link] air
Jordan Shoes,Nike Air Jordans, Air Force Ones,Retro Air Jordan, Bape
Hoodies,Bape shoes-Tencent Traveler|Jordan Shoes,Nike Air Jordans,
Air Force Ones,Retro Air Jordan, Bape Hoodies,Bape shoes-Tencent
Traveler|Jordan Shoes,Nike Air Jordans, Air Force Ones,Retro Air
- Re: Error message from random number generation
Arguments are mismatched. RTM.
see [link]
The mvn_a0a1b_list array is one dimensional in the calling program,
but the documentation says it is two dimensional. I can not tell from
the docs which dimension comes first! Should it be (10,3) or (3,10)? I
- Re: Error message from random number generation
In article <1j2a6ty.150a0io1y7bsg4N%nos.. .@see.signature>,
I would put it more strongly. It is extremely unlikely to have
anything to do with random number generation. From its name, I would
expect it to have to do with I/O.
I agree completely.
I wouldn't. I would start looking for array bound errors, argument
- Re: Error message from random number generation
I fail to see what random number generation has to do with the error at
all. As noted, the error message refers to vsldeletestream. Not knowing
anything about that routine, I can't be of much help other than to say
that...
No, this doesn't sound like a step in the right direction. I'd say that
your problem likely has little or nothing to do with sampling from
- Re: Error message from random number generation
Maybe it's your source code. You have some very long lines. I am
assuming that you are using free form source, even though all your
program statements are indented. [If you are putting TAB characters in
your source file, DON'T do that.] Perhaps some of your source
lines are being truncated? It makes more sense, and it makes it much
- Re: Error message from random number generation
Hello again,
Thanks for your responses. I apologize in advance for my ignorance on
some of the things you've mentioned - I'm not a programmer, just a
stats student who was advised to use Fortran for my research :)
The subroutine (called mvn_rng_mkl) I've used is similar to ones I
have used to sample from normal and uniform distributions, which I've
- Re: Error message from random number generation
It's surprising that you code got that far.
The USE statement is incorrect, so it should have failed in compilation.
However, assuming that you correct that, it's telling you what the
error is, namely, that the first argument to vslDeleteStream is wrong.
Do you have an interface for it? And for the other external procedures
- HOT GIRLS TOP AND BOTTOM SEXY MOVIE
HOT GIRLS TOP AND BOTTOM SEXY MOVIE
[link]
[link]
- Re: Error message from random number generation
vslDeleteStream is an Intel Math Kernel Library (MKL) routine. I
suggest that you ask for assistance at the Intel MKL user forum at
[link]
- Merge Sorting
Back in the good old days of the 1960 there was a routine in Fortran 66 that could take data from one tape, sort it in chunks and send the result to another tape. It would then open the new tape and merge the files into new sorted files on another tape. This was repeated until all of the chunks were merged back together.
- Re: Static array versus allocatable array in derived type
No
Yes/No : I profiled the code in counting for allocate/deallocate and
in measuring the CPU-TIME of different parts. But I did not use a
specific profiling tool like gprof because the routines I want to
profile are too fast (the overhead of a profile tool would be too
high).
I tried 3 different compilers (ifort, g95, gfortran). Did you look at
- Re: Static array versus allocatable array in derived type
And did changing those 5 INTENTs make any difference?
BTW, did you do a profile on the code to find where the time is being spent?
Also, did you try another compiler (as I suggested)?
In my test on your code, there was no difference in time between your codes.
- Re: compaq VF consistently crashes on "add files to project"
I have run into the same problem and was wondering if you tried
uninstalling recent Windows XP updates. Did it work?
Thanks,
Frank
- Re: Error message from random number generation
...
....
...
...
...
What you haven't included is the calling protocol/argument list for the
routine vslDeleteStr that raised an error. Something like you passed an
integer instead of real or vice versa would easily cause the problem.
It's also interesting there's the reference to "vsldeletestream" in the
- Re: Error message from random number generation
The error messages seem to be coming from a routine called
vsldeletestr, but your code below doesn't reference or contain
a routine with that name. Possible could it be a shortened version
of vsldelerestream?
Anyhow, the obvious problem is that something is wrong with the
first argument to vsldeletestr ;).
- Error message from random number generation
Hello,
I am a relatively new user to Fortran, and have pretty basic
programming skills (my research is in statistics). I was looking for
some guidance with a program I'm working on, and came across this
forum. I'm hoping someone here may be able to point me in the right
direction with the problem I'm having.
- www.dudes-mall.com Ed hardy,(tshirt$13,swim strunk$25,jean$30,handbag$34,cap$13,sunglass$12,shoes$25)nike shoes:$32,handbag:$35,NFL:$20,jean:$30,air force one shoes,lv,gucci,D&G,bbc,coachUGG boot:$50,.free shipping!
Get Nike Shoes at Super Cheap Prices
Discount Ed hardy tshirt ([link])
Discount Ed hardy swimming suit ([link])
Discount Ed hardy jean ([link])
Discount Ed hardy shoes ([link])
Discount Ed hardy handbag ([link])
Discount Ed hardy other porduct ([link])
- Re: deferred type parameters, again
Once again thanks for sharing your expertise. gfortran's source
code (not including the runtime library) is 128 KLOC. I've started
adding what some might consider to be trite comments. For example,
the 5 lines of code that now check C403 is prefaced with /* C403:
not a declaration. */
REMOVE:kargl[206] gfc4x -c t.f90
- (paypal payment)( www.brand-clothing.com ) sell Air max 90 shoes
(paypal payment)( [link] ) discount wholesaler Air max
95
shoes
(paypal payment)( [link] ) wholesale Air Max 87 shoes
(paypal payment)( [link] ) discount wholesaler Air MAX
LTD
shoes
(paypal payment)( [link] ) sell Air max 90 shoes
(paypal payment)( [link] ) discount wholesaler Air Max
- Re: deferred type parameters, again
Ok. Went and looked. Found it, but it is a bit far away from where you
were probably looking. That's largely because it isn't specific to the
ALLOCATE statement. It is a much more general restriction. See C403
" A colon may be used as a type-param-value only in the declaration of
an entity or component that has the POINTER or ALLOCATABLE attribute."
- Re: linking dll with g95
In article ,
No, there isn't. That is part of the justification for the unnecessary
restrictions on setjmp/longjmp.
In theory, that is correct. In practice, it is less so.
Any program that can tell the difference is relying on undefined
behaviour. But - hey! - ALL realistic C programs have to do that.
- Re: deferred type parameters, again
Well, I can't find a prohibition. The number constraints in 6.3.1
seem to miss this case and the surrounding text does not give
a prohibition. There may very well be one, but I can't find it.
From a common-sense standpoint, I had come to essentially this
conclusion.
Yes. The declaration of fmt should have included ALLOCATABLE.
- Re: linking dll with g95
I suppose the C/C++ Standard like the Fortran one doesn't address the
"how" an implementation manages its business?
I was just surprised that what I thought was traditional in C owing to
the language syntax would go away and seem to me in my ignorance to
cause widespread incompatibilities in C code by switching from "callee
- Re: linking dll with g95
dpb wrote:
stdcall causes problems with C varargs. With cdecl, nothing special is
required because the caller knows how much was sent in and how much
needs to be cleaned up off the stack, whereas with stdcall it would be
necessary to do something to let the callee know how much to clean (e.g.
a hidden length argument).
- Re: Question on MOVE_ALLOC
It will have that kind of effect, but it isn't as though the
implementation actually has to do anything for that. That will just be
the effect of moving the top-level allocation. I'm still feeling that I
haven't managed to communicate what is happening. When you move the
allocation, that has the effect of moving everything, without actually
- Re: Question on MOVE_ALLOC
Richard,
Thank you for your response. Just so I'm clear on this: Will the
call
to MOVE_ALLOC(A,B) move the allocation of A(K)%MEMBER to B(K)%MEMBER
for each valid index K of the original A? In other words, it is not
necessary to include code like
do k=1,n
call move_alloc(a(k)%member, b(k)%member)
- Re: Question on MOVE_ALLOC
No. That's not the way it works. The effect is somewhat like that, but
not quite. Sounds to me like you have the wrong model for what is going
on.
Data is not copied at all, deep or otherwise. Conceptually, there are 3
steps.
1. First, if b was previously allocated, it is deallocated. That
includes all the usual consequences of any deallocation. Any allocatable
- Re: deferred type parameters, again
I'm feeling to lazy to go look for a prohibition right now. I'd sort of
expect to find one somewhere and would consider it a minor bug in the
standard if it isn't explicitly prohibited. I would say that it was
prohibited anyway, if only because there is a general thing somewhere
way up front that prohibits things where the standard does not establish
- Re: linking dll with g95
...
OK, I'm showing my ignorance here... :)
I thought there was something inherent in C about callee cleaning up and
the Win32 API stdcall was an aberration????
- Re: Use of deferred type parameter and ALLOCATE
Yes, I think you are correct in both cases.
Correct. Perhaps slightly odd, as you can implicitly reallocate it, but
an explicit allocation also requires an explicit deallocation. I'm not
entirely sure how philosophically consistent this is, but you have it
correct, as far as I can tell.
Also correct. One might plausibly imagine implicit allocation being
- deferred type parameters, again
In developing a testcase for the gfortran testsuite, I
stumbled upon the following code:
program test
character(len=:) fmt
allocate(character(len=:) :: fmt)
end program test
This certainly looks to be invalid, but under Sec. 6.3.1
I can't find a prohibition or any clear guidance on how
the above would be interpreted.
- Question on MOVE_ALLOC
Does MOVE_ALLOC perform deep copies? In other words,
will code like the following populate B with the all the same
data as A and deallocate all the members of A as well as
A itself?
type t
real, allocatable:: member(:)
...
end type t
type(t), allocatable:: a(:), b(:)
! allocate and populate A
- Re: Use of deferred type parameter and ALLOCATE
Reviving an old thread.
First, thanks Richard and Colin for your comments. If I
understanding
correctly, the following is a conforming code.
program test
implicit none
character(:), allocatable :: fmt
!
! Allocate 3 storage units. Assign 'abc' into these units.
! Print contents of fmt, which is 'abc'.
- www.sbbshoe.com cheaper g-satr Kidrobot hoodies ed hardy star red monkey gino green global True Religion ED-hardy Kidrobot jeans hoodies china supplier wholesaler exporters, manufacture
[link]
g-satr Kidrobot hoodies ed hardy star red monkey gino
green global True Religion ED-hardy Kidrobot jeans hoodies
wholesale - Offers for g-satr Kidrobot hoodies ed hardy star red
monkey gino green global True Religion ED-hardy Kidrobot jeans
hoodies wholesale - exporters, manufacture
- Re: linking dll with g95
Caller always cleans up in x64. The stack comes to the callee aligned
8 mod 16. Caller is responsible for creating the 32-byte parameter
save area just above the return address. See
[link] for a concise
comparison of calling conventions.
- Re: paypal payment)( www.niketrade08.cn ) sell Air max 90 shoes
please look our website ,have more mode shoes clothing hat cap bags !
[link]
Wholesale Sell Discount:
Dsquared 2 shoes,Dsquared jeans t-shirt Dsquared jacket [link]
Abercrombie & Fitch 2009 Polo A&F t-shirt bags jeans jacket apparel
[link]
air max 90 87 88 89 ltd 91 95 97 tn shoes [link]
- Re: linking dll with g95
...
Does the responsibility for stack cleanup change as well between 32- and
64-bit Windows? That seems bizarre if so...
- Re: linking dll with g95
Sure, it's fine for Win64, but I thought 32-bit had been discussed in
the thread as well.
- Using g95 with the NAG Library
Quoting from NAGNews
Volume 81 - June/July 2009
'Question: "Can I use the g95 compiler with the NAG Library?"
Answer: Good program developers know that it's always wise to test
their code using more than one compiler, for the simple reason that
different compilers can find different problems. Code that one
- Re: linking dll with g95
On 1 jul, 10:43, Catherine Rees Lay
wrote:
Luckily, you can still smile.
Regards,
Arjen
- Re: linking dll with g95
You should make sure you use the right version though. Often specifying
the "wrong" calling convention will appear to work but fail somewhere
completely different much later on in the program, or on the seventeenth
time you call it, or when there's a J in the month, or on your client's
computer, when it works just fine on yours...
- Re: Static array versus allocatable array in derived type
I know that very well but when a derided type variable contains
allocatable or pointer components, it is often more efficient (as
underlined by R. Maine) to take care about deallocation and
reallocation which are "automatic" in case of INTENT(OUT) : an
allocatable array already allocated is immediately deallocated when
- Re: Static array versus allocatable array in derived type
I know that very well but when a derided type variable contains
allocatable or pointer components, it is often more efficient (as
underlined by R. Maine) to take care about deallocation and
reallocation which are "automatic" in case of INTENT(OUT) : an
allocatable array already allocated is immediately deallocated when
- Re: Static array versus allocatable array in derived type
I know that very well but when a derided type variable contains
allocatable or pointer components, it is often more efficient (as
underlined by R. Maine) to take care about deallocation and
reallocation which are "automatic" in case of INTENT(OUT) : an
allocatable array already allocated is immediately deallocated when
- Re: hex and octal constants in various languages
On Mon, 15 Jun 2009 11:00:26 -0600, "James Van Buskirk"
the bitpattern of a hex constant to a floating-point representation>
Nit: C integer constants, unless suffixed L or LL, are the 'lowest'
type at least as high as int sufficient to contain the value.
- Re: Static array versus allocatable array in derived type
I know that, and I expect that the OP does also.
I was commenting in the fact that the OP changed his
INTENT(OUT) dummy arguments to INOUT,
which was not a good step.
And vice versa if the dummy argument is INTENT (OUT).
- Re: linking dll with g95
Not that misleading. For 64-bit Windows stdcall and cdecl are
indistinguishable.
- Re: linking dll with g95
The latter point is potentially misleading. The default calling
convention for Win32 APIs is stdcall, not cdecl.
- Re: FORTRAN 90/95 program reading .dat file (\\ force line continuation)
From the information above, your issue appears to be with FVCOM
(finite volume coastal ocean model), and how it works. I would not
expect a change in compilers to fix this, althought I'd suppose that
it (fvcom) could be relying on extensions or undocumented behaviour of
a specific compiler. That, hopefully, would not be the case.
|
Computational