RE: 16-bit int

RE: 16-bit int

[Göran Steen]

Thank you for your response.

I want to make sure that my code snippets, if they are compiled and run on a machine with 16-bit int, still works. I don't have access to such a machine, so I want to compile and run them with 16-bit int on my machine that defaults to 32-bit int. Especially the intermediate results' size are interesting. What happens if they are truncated to 16-bit?

There will not be any problem with library compatibility.

BR /göran

-----Original Message-----
From: David Brown [mailto:david@westcontrol.com]
Sent: den 9 augusti 2012 15:37
To: Göran Steen
Cc: gcc-help@gcc.gnu.org
Subject: Re: 16-bit int

On 09/08/2012 13:26, Göran Steen wrote:
> Hi!
>
> I use gcc version 4.3.4, where int is 32-bit variables. Is it possible 
> to setup compiler to compile int as 16-bit variables?
>
> With best regards / Med vänlig hälsning / Mit freundlichen Grüßen / 
> Saudações
>
> Göran Steen


The size of an int depends on the target - gcc supports dozens of targets.  Most have 32-bit ints, but some have 16-bit ints and at least one has a compile-time option to support 8-bit ints (though that goes against C standards, and is deprecated on current builds).  There are probably also targets with 64-bit ints.

So step one in asking for help here is to tell us your target.

Step two is to tell us what you are hoping to achieve.  Almost certainly, there is no way to change the int size - and even if there happens to be a command-line switch for the given target, it is probably not a good idea (you'll get in a horrible mess with library compatibility, for example).  And even if it is possible, it is highly unlikely to be advantageous.  Tell us what you really want to achieve here, and people can give you advice towards that.

mvh.,

David

Re: 16-bit int

[David Brown]

On 10/08/2012 09:15, Göran Steen wrote:
> Thank you for your response.
>
> I want to make sure that my code snippets, if they are compiled and
> run on a machine with 16-bit int, still works. I don't have access to
> such a machine, so I want to compile and run them with 16-bit int on
> my machine that defaults to 32-bit int. Especially the intermediate
> results' size are interesting. What happens if they are truncated to
> 16-bit?
>

The answer here is simple - #include <stdint.h>, and use types like 
int_fast16_t.  If intermediary results might need 32 bits, cast to 
int_fast32_t as needed.  On a target with 32-bit ints, both these types 
will normally be 32-bit (though int_fast16_t could theoretically be 
16-bit), and the cast will do nothing.  On a target with 16-bit ints, 
you will get 16-bit and 32-bit ints respectively.

This will give you optimal code for all sizes of target ints, while also 
being correct on each target.

mvh.,

David


> There will not be any problem with library compatibility.
>
> BR /göran
>
> -----Original Message----- From: David Brown
> [mailto:david@westcontrol.com] Sent: den 9 augusti 2012 15:37 To:
> Göran Steen Cc: gcc-help@gcc.gnu.org Subject: Re: 16-bit int
>
> On 09/08/2012 13:26, Göran Steen wrote:
>> Hi!
>>
>> I use gcc version 4.3.4, where int is 32-bit variables. Is it
>> possible to setup compiler to compile int as 16-bit variables?
>>
>> With best regards / Med vänlig hälsning / Mit freundlichen Grüßen
>> / Saudações
>>
>> Göran Steen
>
>
> The size of an int depends on the target - gcc supports dozens of
> targets.  Most have 32-bit ints, but some have 16-bit ints and at
> least one has a compile-time option to support 8-bit ints (though
> that goes against C standards, and is deprecated on current builds).
> There are probably also targets with 64-bit ints.
>
> So step one in asking for help here is to tell us your target.
>
> Step two is to tell us what you are hoping to achieve.  Almost
> certainly, there is no way to change the int size - and even if there
> happens to be a command-line switch for the given target, it is
> probably not a good idea (you'll get in a horrible mess with library
> compatibility, for example).  And even if it is possible, it is
> highly unlikely to be advantageous.  Tell us what you really want to
> achieve here, and people can give you advice towards that.
>
> mvh.,
>
> David
>

Re: 16-bit int

[David Brown]

On 10/08/2012 09:15, Göran Steen wrote:
> Thank you for your response.
>
> I want to make sure that my code snippets, if they are compiled and
> run on a machine with 16-bit int, still works. I don't have access to
> such a machine, so I want to compile and run them with 16-bit int on
> my machine that defaults to 32-bit int. Especially the intermediate
> results' size are interesting. What happens if they are truncated to
> 16-bit?
>

The answer here is simple - #include <stdint.h>, and use types like 
int_fast16_t.  If intermediary results might need 32 bits, cast to 
int_fast32_t as needed.  On a target with 32-bit ints, both these types 
will normally be 32-bit (though int_fast16_t could theoretically be 
16-bit), and the cast will do nothing.  On a target with 16-bit ints, 
you will get 16-bit and 32-bit ints respectively.

This will give you optimal code for all sizes of target ints, while also 
being correct on each target.

mvh.,

David


> There will not be any problem with library compatibility.
>
> BR /göran
>
> -----Original Message----- From: David Brown
> [mailto:david@westcontrol.com] Sent: den 9 augusti 2012 15:37 To:
> Göran Steen Cc: gcc-help@gcc.gnu.org Subject: Re: 16-bit int
>
> On 09/08/2012 13:26, Göran Steen wrote:
>> Hi!
>>
>> I use gcc version 4.3.4, where int is 32-bit variables. Is it
>> possible to setup compiler to compile int as 16-bit variables?
>>
>> With best regards / Med vänlig hälsning / Mit freundlichen Grüßen
>> / Saudações
>>
>> Göran Steen
>
>
> The size of an int depends on the target - gcc supports dozens of
> targets.  Most have 32-bit ints, but some have 16-bit ints and at
> least one has a compile-time option to support 8-bit ints (though
> that goes against C standards, and is deprecated on current builds).
> There are probably also targets with 64-bit ints.
>
> So step one in asking for help here is to tell us your target.
>
> Step two is to tell us what you are hoping to achieve.  Almost
> certainly, there is no way to change the int size - and even if there
> happens to be a command-line switch for the given target, it is
> probably not a good idea (you'll get in a horrible mess with library
> compatibility, for example).  And even if it is possible, it is
> highly unlikely to be advantageous.  Tell us what you really want to
> achieve here, and people can give you advice towards that.
>
> mvh.,
>
> David
>

RE: 16-bit int

[Göran Steen]

Thank you for your response.

Sorry, this will not help. I want to test the snippets, not change them. Besides, I don't know if the compiler that will be used for the 16-bit machine will support any of the types you suggest, so I only want to use standard types.

BR /göran

-----Original Message-----
From: David Brown [mailto:david@westcontrol.com] 
Sent: den 10 augusti 2012 10:12
To: Göran Steen
Cc: gcc-help@gcc.gnu.org
Subject: Re: 16-bit int

On 10/08/2012 09:15, Göran Steen wrote:
> Thank you for your response.
>
> I want to make sure that my code snippets, if they are compiled and 
> run on a machine with 16-bit int, still works. I don't have access to 
> such a machine, so I want to compile and run them with 16-bit int on 
> my machine that defaults to 32-bit int. Especially the intermediate 
> results' size are interesting. What happens if they are truncated to 
> 16-bit?
>

The answer here is simple - #include <stdint.h>, and use types like int_fast16_t.  If intermediary results might need 32 bits, cast to int_fast32_t as needed.  On a target with 32-bit ints, both these types will normally be 32-bit (though int_fast16_t could theoretically be 16-bit), and the cast will do nothing.  On a target with 16-bit ints, you will get 16-bit and 32-bit ints respectively.

This will give you optimal code for all sizes of target ints, while also being correct on each target.

mvh.,

David


> There will not be any problem with library compatibility.
>
> BR /göran
>
> -----Original Message----- From: David Brown 
> [mailto:david@westcontrol.com] Sent: den 9 augusti 2012 15:37 To:
> Göran Steen Cc: gcc-help@gcc.gnu.org Subject: Re: 16-bit int
>
> On 09/08/2012 13:26, Göran Steen wrote:
>> Hi!
>>
>> I use gcc version 4.3.4, where int is 32-bit variables. Is it 
>> possible to setup compiler to compile int as 16-bit variables?
>>
>> With best regards / Med vänlig hälsning / Mit freundlichen Grüßen / 
>> Saudações
>>
>> Göran Steen
>
>
> The size of an int depends on the target - gcc supports dozens of 
> targets.  Most have 32-bit ints, but some have 16-bit ints and at 
> least one has a compile-time option to support 8-bit ints (though that 
> goes against C standards, and is deprecated on current builds).
> There are probably also targets with 64-bit ints.
>
> So step one in asking for help here is to tell us your target.
>
> Step two is to tell us what you are hoping to achieve.  Almost 
> certainly, there is no way to change the int size - and even if there 
> happens to be a command-line switch for the given target, it is 
> probably not a good idea (you'll get in a horrible mess with library 
> compatibility, for example).  And even if it is possible, it is highly 
> unlikely to be advantageous.  Tell us what you really want to achieve 
> here, and people can give you advice towards that.
>
> mvh.,
>
> David
>

Re: 16-bit int

[David Brown]

Could you please not top-post?  It makes it very difficult to follow the 
order of the thread.  Remember that some people see these emails in 
isolation, rather than threaded in an email client.  I've re-ordered the 
mail below.

On 10/08/2012 10:35, Göran Steen wrote:
> -----Original Message----- From: David Brown
> [mailto:david@westcontrol.com] Sent: den 10 augusti 2012 10:12 To:
> Göran Steen Cc: gcc-help@gcc.gnu.org Subject: Re: 16-bit int
>
> On 10/08/2012 09:15, Göran Steen wrote:
>> Thank you for your response.
>>
>> I want to make sure that my code snippets, if they are compiled
>> and run on a machine with 16-bit int, still works. I don't have
>> access to such a machine, so I want to compile and run them with
>> 16-bit int on my machine that defaults to 32-bit int. Especially
>> the intermediate results' size are interesting. What happens if
>> they are truncated to 16-bit?
>>
>
> The answer here is simple - #include <stdint.h>, and use types like
> int_fast16_t.  If intermediary results might need 32 bits, cast to
> int_fast32_t as needed.  On a target with 32-bit ints, both these
> types will normally be 32-bit (though int_fast16_t could
> theoretically be 16-bit), and the cast will do nothing.  On a target
> with 16-bit ints, you will get 16-bit and 32-bit ints respectively.
>
> This will give you optimal code for all sizes of target ints, while
> also being correct on each target.
>
> mvh.,
>
> David

 > Thank you for your response.
 >
 > Sorry, this will not help. I want to test the snippets, not change
 > them. Besides, I don't know if the compiler that will be used for the
 > 16-bit machine will support any of the types you suggest, so I only
 > want to use standard types.
 >
 > BR /göran

If you need to test the snippets on different targets, then you need to 
have compilers and targets available.  You still haven't said which 
target(s) you are using, but unless it is one like the M68k then there 
are no compiler switches to help you.

Note that there is no point in testing anything until you have done what 
you can to make sure the code is correctly written.  Otherwise your 
tests may work on some targets by coincidence, but not on others, or may 
be sub-optimal on some targets.

Any C99 compiler - and almost all pre-C99 compilers - will have 
<stdint.h>.  I don't know of any compiler less than 15 years old that 
doesn't come with a <stdint.h>, and many people using such compilers 
have written their own <stdint.h>.

The int_fast16_t and int_fast32_t types (and the unsigned versions) are 
mandatory in <stdint.h>, so you can take it for granted that /all/ 
compilers support them.  This is unlike the fixed-size types (like 
int16_t) that will be defined if and only if the target supports types 
of exactly that size (some architectures don't support the smaller types).

mvh.,

David

RE: 16-bit int

[Göran Steen]

-----Original Message-----
From: David Brown [mailto:david@westcontrol.com] 
Sent: den 10 augusti 2012 10:45
To: Göran Steen
Cc: gcc-help@gcc.gnu.org
Subject: Re: 16-bit int


Could you please not top-post?  It makes it very difficult to follow the order of the thread.  Remember that some people see these emails in isolation, rather than threaded in an email client.  I've re-ordered the mail below.

On 10/08/2012 10:35, Göran Steen wrote:
> -----Original Message----- From: David Brown 
> [mailto:david@westcontrol.com] Sent: den 10 augusti 2012 10:12 To:
> Göran Steen Cc: gcc-help@gcc.gnu.org Subject: Re: 16-bit int
>
> On 10/08/2012 09:15, Göran Steen wrote:
>> Thank you for your response.
>>
>> I want to make sure that my code snippets, if they are compiled and 
>> run on a machine with 16-bit int, still works. I don't have access to 
>> such a machine, so I want to compile and run them with 16-bit int on 
>> my machine that defaults to 32-bit int. Especially the intermediate 
>> results' size are interesting. What happens if they are truncated to 
>> 16-bit?
>>
>
> The answer here is simple - #include <stdint.h>, and use types like 
> int_fast16_t.  If intermediary results might need 32 bits, cast to 
> int_fast32_t as needed.  On a target with 32-bit ints, both these 
> types will normally be 32-bit (though int_fast16_t could theoretically 
> be 16-bit), and the cast will do nothing.  On a target with 16-bit 
> ints, you will get 16-bit and 32-bit ints respectively.
>
> This will give you optimal code for all sizes of target ints, while 
> also being correct on each target.
>
> mvh.,
>
> David

 > Thank you for your response.
 >
 > Sorry, this will not help. I want to test the snippets, not change  > them. Besides, I don't know if the compiler that will be used for the  > 16-bit machine will support any of the types you suggest, so I only  > want to use standard types.
 >
 > BR /göran

If you need to test the snippets on different targets, then you need to have compilers and targets available.  You still haven't said which
target(s) you are using, but unless it is one like the M68k then there are no compiler switches to help you.

Note that there is no point in testing anything until you have done what you can to make sure the code is correctly written.  Otherwise your tests may work on some targets by coincidence, but not on others, or may be sub-optimal on some targets.

Any C99 compiler - and almost all pre-C99 compilers - will have <stdint.h>.  I don't know of any compiler less than 15 years old that doesn't come with a <stdint.h>, and many people using such compilers have written their own <stdint.h>.

The int_fast16_t and int_fast32_t types (and the unsigned versions) are mandatory in <stdint.h>, so you can take it for granted that /all/ compilers support them.  This is unlike the fixed-size types (like
int16_t) that will be defined if and only if the target supports types of exactly that size (some architectures don't support the smaller types).

mvh.,

David

Thank you for your response.

I want to test the snippets with 16-bit int, without having target system or compiler available. You wrote that there is no compiler switch to help me. Thank you for that answer.

BR /göran

Re: 16-bit int

[David Brown]

On 10/08/2012 11:05, Göran Steen wrote:
>
>
> -----Original Message----- From: David Brown
> [mailto:david@westcontrol.com] Sent: den 10 augusti 2012 10:45 To:
> Göran Steen Cc: gcc-help@gcc.gnu.org Subject: Re: 16-bit int
>
>
> Could you please not top-post?  It makes it very difficult to follow
> the order of the thread.  Remember that some people see these emails
> in isolation, rather than threaded in an email client.  I've
> re-ordered the mail below.
>
> On 10/08/2012 10:35, Göran Steen wrote:
>> -----Original Message----- From: David Brown
>> [mailto:david@westcontrol.com] Sent: den 10 augusti 2012 10:12 To:
>> Göran Steen Cc: gcc-help@gcc.gnu.org Subject: Re: 16-bit int
>>
>> On 10/08/2012 09:15, Göran Steen wrote:
>>> Thank you for your response.
>>>
>>> I want to make sure that my code snippets, if they are compiled
>>> and run on a machine with 16-bit int, still works. I don't have
>>> access to such a machine, so I want to compile and run them with
>>> 16-bit int on my machine that defaults to 32-bit int. Especially
>>> the intermediate results' size are interesting. What happens if
>>> they are truncated to 16-bit?
>>>
>>
>> The answer here is simple - #include <stdint.h>, and use types
>> like int_fast16_t.  If intermediary results might need 32 bits,
>> cast to int_fast32_t as needed.  On a target with 32-bit ints, both
>> these types will normally be 32-bit (though int_fast16_t could
>> theoretically be 16-bit), and the cast will do nothing.  On a
>> target with 16-bit ints, you will get 16-bit and 32-bit ints
>> respectively.
>>
>> This will give you optimal code for all sizes of target ints,
>> while also being correct on each target.
>>
>> mvh.,
>>
>> David
>
>> Thank you for your response.
>>
>> Sorry, this will not help. I want to test the snippets, not change
>> > them. Besides, I don't know if the compiler that will be used for
>> the  > 16-bit machine will support any of the types you suggest, so
>> I only  > want to use standard types.
>>
>> BR /göran
>
> If you need to test the snippets on different targets, then you need
> to have compilers and targets available.  You still haven't said
> which target(s) you are using, but unless it is one like the M68k
> then there are no compiler switches to help you.
>
> Note that there is no point in testing anything until you have done
> what you can to make sure the code is correctly written.  Otherwise
> your tests may work on some targets by coincidence, but not on
> others, or may be sub-optimal on some targets.
>
> Any C99 compiler - and almost all pre-C99 compilers - will have
> <stdint.h>.  I don't know of any compiler less than 15 years old that
> doesn't come with a <stdint.h>, and many people using such compilers
> have written their own <stdint.h>.
>
> The int_fast16_t and int_fast32_t types (and the unsigned versions)
> are mandatory in <stdint.h>, so you can take it for granted that
> /all/ compilers support them.  This is unlike the fixed-size types
> (like int16_t) that will be defined if and only if the target
> supports types of exactly that size (some architectures don't support
> the smaller types).
>
> mvh.,
>
> David
>
> Thank you for your response.
>
> I want to test the snippets with 16-bit int, without having target
> system or compiler available. You wrote that there is no compiler
> switch to help me. Thank you for that answer.
>
> BR /göran
>

Note that even though gcc doesn't have such a switch for many targets, 
there may be other 16-bit C compilers available that you can test with. 
  You /still/ haven't given any information about your systems - either 
the targets you are aiming for, or the systems you are using for 
testing.  But I am guessing (from your email headers) that you are using 
some sort of Windows system - it is quite possible that you can get a 
16-bit DOS compiler to run for your testing.

You can also test by adding some "int16_t" casts into your code - that 
will cause values to be truncated explicitly to 16 bits.

I know you don't want to change the code, but there is no way around 
this - if you want to write code that is portable across different 
integer sizes, you have to write code in the correct way.  Bad code that 
happens to test okay on one system is still bad code.

And you really should find some way to test the code on real targets - 
otherwise you are just guessing.

mvh.,

David

Re: 16-bit int

[Jonathan Wakely]

On 10 August 2012 09:35, Göran Steen wrote:
> Thank you for your response.
>
> Sorry, this will not help. I want to test the snippets, not change them. Besides, I don't know if the compiler that will be used for the 16-bit machine will support any of the types you suggest, so I only want to use standard types.

Those are standard types.

But the bottom line is that on most targets you can't change the size
of int. You still haven't said what target you're using, but chances
are it's not possible.  You could look into running a simulator for a
target with 16-bit int instead.

Re: 16-bit int

[Vincent Lefevre]

On 2012-08-10 10:12:05 +0200, David Brown wrote:
> On 10/08/2012 09:15, Göran Steen wrote:
> >I want to make sure that my code snippets, if they are compiled and
> >run on a machine with 16-bit int, still works.

This is what I wish to do too.

> >I don't have access to such a machine, so I want to compile and run
> >them with 16-bit int on my machine that defaults to 32-bit int.

Note that if your code snippets use functions from the C library, this
won't work.

> >Especially the intermediate results' size are interesting. What
> >happens if they are truncated to 16-bit?
> 
> The answer here is simple - #include <stdint.h>, and use types like
> int_fast16_t.

Actually you need to use your *own* type, say my_int16_t, and typedef
it to int_fast16_t for the normal use of your code when <stdint.h> is
available, to int for the normal use of your code when <stdint.h> is
not available, and to int16_t (not int_fast16_t) for the tests.

Now if you want to also test code with calls to the C library, you'll
need to find a target with 16-bit int's to test everything.

-- 
Vincent Lefèvre <vincent@vinc17.net> - Web: <http://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <http://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)

Re: 16-bit int

[Vincent Lefevre]

On 2012-08-10 09:05:13 +0000, Göran Steen wrote:
> Any C99 compiler - and almost all pre-C99 compilers - will have
> <stdint.h>. I don't know of any compiler less than 15 years old that
> doesn't come with a <stdint.h>, and many people using such compilers
> have written their own <stdint.h>.

I thought that Microsoft's compiler didn't have <stdint.h>.
I've also heard that <inttypes.h> is more common.

> The int_fast16_t and int_fast32_t types (and the unsigned versions)
> are mandatory in <stdint.h>, so you can take it for granted that
> /all/ compilers support them. This is unlike the fixed-size types
> (like int16_t) that will be defined if and only if the target
> supports types of exactly that size (some architectures don't
> support the smaller types).

But int_fast16_t is useless to test whether code can be affected
by 16-bit truncation on platforms for which int_fast16_t is really
a 16-bit type. For tests, int16_t is necessary. Now the user may
want to know what targets provide this type.

-- 
Vincent Lefèvre <vincent@vinc17.net> - Web: <http://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <http://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)

Re: 16-bit int

[Ian Lance Taylor]

On Sat, Aug 11, 2012 at 3:47 AM, Vincent Lefevre <vincent+gcc@vinc17.org> wrote:
> On 2012-08-10 09:05:13 +0000, Göran Steen wrote:
>> Any C99 compiler - and almost all pre-C99 compilers - will have
>> <stdint.h>. I don't know of any compiler less than 15 years old that
>> doesn't come with a <stdint.h>, and many people using such compilers
>> have written their own <stdint.h>.
>
> I thought that Microsoft's compiler didn't have <stdint.h>.
> I've also heard that <inttypes.h> is more common.

It's true that <inttypes.h> is older than <stdint.h>.  But as Göran
said, <stdint.h> is in C99, a standard that was released 13 years ago.
 I have to assume that even Microsoft compilers support <stdint.h> by
now.

Ian

Re: 16-bit int

[Tim Prince]

On 8/11/2012 11:32 AM, Ian Lance Taylor wrote:
> On Sat, Aug 11, 2012 at 3:47 AM, Vincent Lefevre <vincent+gcc@vinc17.org> wrote:
>> On 2012-08-10 09:05:13 +0000, Göran Steen wrote:
>>> Any C99 compiler - and almost all pre-C99 compilers - will have
>>> <stdint.h>. I don't know of any compiler less than 15 years old that
>>> doesn't come with a <stdint.h>, and many people using such compilers
>>> have written their own <stdint.h>.
>> I thought that Microsoft's compiler didn't have <stdint.h>.
>> I've also heard that <inttypes.h> is more common.
> It's true that <inttypes.h> is older than <stdint.h>.  But as Göran
> said, <stdint.h> is in C99, a standard that was released 13 years ago.
>   I have to assume that even Microsoft compilers support <stdint.h> by
> now.
>
> Ian
I haven't been able to find an authoritative source, but someone said 
recently that Microsoft had stated they would some day support those C99 
features which are common with C++ (VS2010 doesn't, and VS2012 isn't 
widely available, but I don't hold my breath).
Microsoft never committed to bring C99 support to the level of any other 
specified compiler, nor did other vendors of Microsoft-compatible 
compilers commit to the same level as on linux, for example.
For example, I haven't found a way to engage the __restrict extension of 
MSVC to emulate (even to the documented extent) the corresponding 
feature of C99.
VS2010 has stdint.h (at least at current service pack level), but not 
inttypes.h.  VS2008 doesn't, and it's still under full support and in 
wide professional use.  As previously mentioned, some use open source 
replacements, even though that is antithetical to the Microsoft proposition.

-- 
Tim Prince

Re: 16-bit int

[David Brown]

On 11/08/2012 12:47, Vincent Lefevre wrote:
> On 2012-08-10 09:05:13 +0000, Göran Steen wrote:
>> Any C99 compiler - and almost all pre-C99 compilers - will have
>> <stdint.h>. I don't know of any compiler less than 15 years old that
>> doesn't come with a <stdint.h>, and many people using such compilers
>> have written their own <stdint.h>.
>
> I thought that Microsoft's compiler didn't have <stdint.h>.
> I've also heard that <inttypes.h> is more common.
>
>> The int_fast16_t and int_fast32_t types (and the unsigned versions)
>> are mandatory in <stdint.h>, so you can take it for granted that
>> /all/ compilers support them. This is unlike the fixed-size types
>> (like int16_t) that will be defined if and only if the target
>> supports types of exactly that size (some architectures don't
>> support the smaller types).
>
> But int_fast16_t is useless to test whether code can be affected
> by 16-bit truncation on platforms for which int_fast16_t is really
> a 16-bit type. For tests, int16_t is necessary. Now the user may
> want to know what targets provide this type.
>

The use of "int_fast16_t" was to get correct and optimal code, whether 
the target is 16-bit or 32-bit.  You are right that this will not test 
whether the code will work on 16-bit targets if it is compiled on a 
32-bit target - using int16_t will help more for that (but it won't give 
guarantees, unless you are absolutely sure there are no hidden int 
promotions).

Re: 16-bit int

[Vincent Lefevre]

On 2012-08-09 15:51:44 -0700, Ian Lance Taylor wrote:
> On Thu, Aug 9, 2012 at 3:38 PM, Vincent Lefevre <vincent+gcc@vinc17.org> wrote:
> >
> > There are many processors I don't know. What I'm interested in is
> > information like: is there any target that has some given int type
> > size?
> 
> In the GCC source code:
> 
> grep INT_TYPE_SIZE gcc/config/*/*.h

Thanks. Actually

grep 'define[[:space:]]*INT_TYPE_SIZE' gcc/config/*/*.h

to get exactly what I need. And

grep 'define[[:space:]]*LONG_TYPE_SIZE' gcc/config/*/*.h
grep 'define[[:space:]]*LONG_LONG_TYPE_SIZE' gcc/config/*/*.h

for the long type and long long type respectively. So, I can see that
one just has: 16/32/64, 32/32/64 and 32/64/64.

-- 
Vincent Lefèvre <vincent@vinc17.net> - Web: <http://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <http://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)

Re: 16-bit int

[David Brown]

On 10/08/2012 08:45, Georg-Johann Lay wrote:
> David Brown schrieb:
>> On 09/08/2012 13:26, Göran Steen wrote:
>>>
>>> I use gcc version 4.3.4, where int is 32-bit variables. Is it
>>> possible to setup compiler to compile int as 16-bit variables?
>
> The m86k port has 32-bit int, -mshort turns it to 16 bit.
>
>> The size of an int depends on the target - gcc supports dozens of
>> targets.  Most have 32-bit ints, but some have 16-bit ints and at
>> least one has a compile-time option to support 8-bit ints (though that
>> goes against C standards, and is deprecated on current builds).  There
>> are probably also targets with 64-bit ints.
>
> You mean the avr port? int is 16 bits wide and -mint8 implements the
> non-standard 8-bit int.  That option was broken for quite some time,
> see PR46261. But that PR is fixed now.
>
> Where in the release notes did you read about deprecation of -mint8?
>

I thought -mint8 had been broken for so long, and that there were no 
plans to fix it or support it in avrlibc, especially as the main reason 
for its existence (inefficient code generated for uint8_t and int8_t 
with normal 16-bit ints) is much less of a problem with newer versions 
of avr-gcc.

I guess I incorrectly extrapolated from this that the option was 
deprecated - I certainly hadn't noticed that the PR had been fixed.

Thanks for the correction.

Re: 16-bit int

[Jose-Marcio Martins da Cruz]

Ian Lance Taylor wrote:
> On Thu, Aug 9, 2012 at 3:38 PM, Vincent Lefevre <vincent+gcc@vinc17.org> wrote:
>>
>> There are many processors I don't know. What I'm interested in is
>> information like: is there any target that has some given int type
>> size?
>
> In the GCC source code:
>
> grep INT_TYPE_SIZE gcc/config/*/*.h
>
>>> There are also a number of out-of-tree gcc ports.
>>
>> Similar information on them would be interesting too.
>
> Pretty hard to come by, though.  Many out-of-tree ports are
> out-of-tree for a reason.


Usually, a regular compiler user just needs to know the size of int on the particular 
architecture he's working on, not on every processor/OS existing on the wild. So, the 
simpler is just to run a program like :

#include <stdio.h>
int main()
{
   printf("sizeof int = %d\n", sizeof(int));
}


If someone needs, for some valid reason, that the integer variables be 16-bit wide, the 
good practice to write clean, portable and maintainable code is to use "int16_t" integer 
type instead of just "int".

JM

-- 

  Envoyé de ma machine à écrire.
  ---------------------------------------------------------------

Re: 16-bit int

[Georg-Johann Lay]

David Brown schrieb:
> On 09/08/2012 13:26, Göran Steen wrote:
>>
>> I use gcc version 4.3.4, where int is 32-bit variables. Is it
>> possible to setup compiler to compile int as 16-bit variables?

The m86k port has 32-bit int, -mshort turns it to 16 bit.

> The size of an int depends on the target - gcc supports dozens of 
> targets.  Most have 32-bit ints, but some have 16-bit ints and at least 
> one has a compile-time option to support 8-bit ints (though that goes 
> against C standards, and is deprecated on current builds).  There are 
> probably also targets with 64-bit ints.

You mean the avr port? int is 16 bits wide and -mint8 implements the
non-standard 8-bit int.  That option was broken for quite some time,
see PR46261. But that PR is fixed now.

Where in the release notes did you read about deprecation of -mint8?

Johann

Re: 16-bit int

[Ian Lance Taylor]

On Thu, Aug 9, 2012 at 3:38 PM, Vincent Lefevre <vincent+gcc@vinc17.org> wrote:
>
> There are many processors I don't know. What I'm interested in is
> information like: is there any target that has some given int type
> size?

In the GCC source code:

grep INT_TYPE_SIZE gcc/config/*/*.h

>> There are also a number of out-of-tree gcc ports.
>
> Similar information on them would be interesting too.

Pretty hard to come by, though.  Many out-of-tree ports are
out-of-tree for a reason.

Ian

Re: 16-bit int

[Vincent Lefevre]

On 2012-08-09 21:58:36 +0200, David Brown wrote:
> There is probably some list in the gcc internals documents, and of course
> you could see it in the source configuration files if you want to go there.
> But you can get a fair idea from the manual on the target-specific options,
> as every target will have a few options of their own:
> 
> <http://gcc.gnu.org/onlinedocs/gcc/Submodel-Options.html>
> 
> This doesn't list the sizes of int or other target-specific information.
> However, usually you know what target processor you are going to use, and
> usually you know the basic size of the processor (if not, then you really
> should find out before starting to use it!).

There are many processors I don't know. What I'm interested in is
information like: is there any target that has some given int type
size?

> There are also a number of out-of-tree gcc ports.

Similar information on them would be interesting too.

-- 
Vincent Lefèvre <vincent@vinc17.net> - Web: <http://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <http://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)

Re: 16-bit int

[David Brown]

On 09/08/12 20:52, Vincent Lefevre wrote:
> On 2012-08-09 15:37:05 +0200, David Brown wrote:
>> The size of an int depends on the target - gcc supports dozens of targets.
>> Most have 32-bit ints, but some have 16-bit ints and at least one has a
>> compile-time option to support 8-bit ints (though that goes against C
>> standards, and is deprecated on current builds).  There are probably also
>> targets with 64-bit ints.
>
> Couldn't the list of supported targets, with type size and other
> similar information, be available on some web page?
>

There is probably some list in the gcc internals documents, and of 
course you could see it in the source configuration files if you want to 
go there.  But you can get a fair idea from the manual on the 
target-specific options, as every target will have a few options of 
their own:

<http://gcc.gnu.org/onlinedocs/gcc/Submodel-Options.html>

This doesn't list the sizes of int or other target-specific information. 
  However, usually you know what target processor you are going to use, 
and usually you know the basic size of the processor (if not, then you 
really should find out before starting to use it!).

There are also a number of out-of-tree gcc ports.

Re: 16-bit int

[Vincent Lefevre]

On 2012-08-09 15:37:05 +0200, David Brown wrote:
> The size of an int depends on the target - gcc supports dozens of targets.
> Most have 32-bit ints, but some have 16-bit ints and at least one has a
> compile-time option to support 8-bit ints (though that goes against C
> standards, and is deprecated on current builds).  There are probably also
> targets with 64-bit ints.

Couldn't the list of supported targets, with type size and other
similar information, be available on some web page?

-- 
Vincent Lefèvre <vincent@vinc17.net> - Web: <http://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <http://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)

Re: 16-bit int

[Paulo J. Matos]

On 09/08/12 12:26, Göran Steen wrote:
> Hi!
>
> I use gcc version 4.3.4, where int is 32-bit variables. Is it possible to setup compiler to compile int as 16-bit variables?
>
> With best regards / Med vänlig hälsning / Mit freundlichen Grüßen / Saudações
>

David answered your question as it came across... however if I 
re-interpret your question as wanting an int type that is 16bits, then 
you can try int16_t from stdint.h.

Cheers,


-- 
PMatos

Re: 16-bit int

[David Brown]

On 09/08/2012 13:26, Göran Steen wrote:
> Hi!
>
> I use gcc version 4.3.4, where int is 32-bit variables. Is it
> possible to setup compiler to compile int as 16-bit variables?
>
> With best regards / Med vänlig hälsning / Mit freundlichen Grüßen /
> Saudações
>
> Göran Steen


The size of an int depends on the target - gcc supports dozens of 
targets.  Most have 32-bit ints, but some have 16-bit ints and at least 
one has a compile-time option to support 8-bit ints (though that goes 
against C standards, and is deprecated on current builds).  There are 
probably also targets with 64-bit ints.

So step one in asking for help here is to tell us your target.

Step two is to tell us what you are hoping to achieve.  Almost 
certainly, there is no way to change the int size - and even if there 
happens to be a command-line switch for the given target, it is probably 
not a good idea (you'll get in a horrible mess with library 
compatibility, for example).  And even if it is possible, it is highly 
unlikely to be advantageous.  Tell us what you really want to achieve 
here, and people can give you advice towards that.

mvh.,

David

16-bit int

[Göran Steen]

Hi!

I use gcc version 4.3.4, where int is 32-bit variables. Is it possible to setup compiler to compile int as 16-bit variables?

With best regards / Med vänlig hälsning / Mit freundlichen Grüßen / Saudações

Göran Steen
Senior Software Developer
EIS Mälardalen
 
EIS by Semcon AB
Patentgatan 8
112 67 STOCKHOLM
Sweden
 
Phone 08-56290697
Mobile 073-6840004
E-mail goran.steen@eis.semcon.com
 
www.semcon.com
www.facebook.com/semcon