Prev: inheriting a main method
Next: Adding int to a float
From: Jim Janney on 21 Sep 2009 19:56
Jim Janney <jjanney(a)shell.xmission.com> writes:

> Thomas Pornin <pornin(a)bolet.org> writes:
>
>> According to rossum <rossum48(a)coldmail.com>:
>>> When coding in C or C++ I use 'register' to hint to the compiler that
>>> something might usefully be kept in a place with easy/quick access:
>>
>> It seems that modern C and C++ compilers tend to ignore such hints.
>> The net effect of "register" is now more something like: "punish me
>> if I use the '&' operator on that variable".
>
> This does have some practical value; the compiler can reasonably
> assume that a register variable is not aliased.
>
> FWIW, I think Datalight C was the first PC compiler to do global
> dataflow analysis and allocate registers based on that, but the big
> players, Borland and Microsoft, were quick to copy it. That would be
> somewhere around 1990.

I just found an old manual for Datalight 3.0 (yes, I'm a pack rat) and
it's copyright 1985-1987. I'm guessing 3.0 was the first one with the
global optimizer and it came out in 1987. It compiled K&R C with a
few extensions, notably the volatile type specifier.

--
Jim Janney
From: Patricia Shanahan on 21 Sep 2009 21:22
Jim Janney wrote:
> Thomas Pornin <pornin(a)bolet.org> writes:
>
>> According to rossum <rossum48(a)coldmail.com>:
>>> When coding in C or C++ I use 'register' to hint to the compiler that
>>> something might usefully be kept in a place with easy/quick access:
>> It seems that modern C and C++ compilers tend to ignore such hints.
>> The net effect of "register" is now more something like: "punish me
>> if I use the '&' operator on that variable".
>
> This does have some practical value; the compiler can reasonably
> assume that a register variable is not aliased.

I don't understand this. It's been a long time since I last worked on a
C compiler, but as far as I can remember it was quite easy to detect
local variables that did not have the "&" operator applied to them.

In what cases would it be difficult for the compiler to detect the lack
of aliases for a variable that could be marked "register"?

Patricia
From: Jim Janney on 22 Sep 2009 00:18
Patricia Shanahan <pats(a)acm.org> writes:

> Jim Janney wrote:
>> Thomas Pornin <pornin(a)bolet.org> writes:
>>
>>> According to rossum <rossum48(a)coldmail.com>:
>>>> When coding in C or C++ I use 'register' to hint to the compiler that
>>>> something might usefully be kept in a place with easy/quick access:
>>> It seems that modern C and C++ compilers tend to ignore such hints.
>>> The net effect of "register" is now more something like: "punish me
>>> if I use the '&' operator on that variable".
>>
>> This does have some practical value; the compiler can reasonably
>> assume that a register variable is not aliased.
>
> I don't understand this. It's been a long time since I last worked on a
> C compiler, but as far as I can remember it was quite easy to detect
> local variables that did not have the "&" operator applied to them.
>
> In what cases would it be difficult for the compiler to detect the lack
> of aliases for a variable that could be marked "register"?
>

Hmm. Now that you put it that way, I can't think of any. Never mind :-)

--
Jim Janney

First  |  Prev  | 
Pages: 1 2 3 4 5 6 7 8
Prev: inheriting a main method
Next: Adding int to a float