Re: Computer architects leaving Intel...

Liste des GroupesRevenir à c arch 
Sujet : Re: Computer architects leaving Intel...
De : anton (at) *nospam* mips.complang.tuwien.ac.at (Anton Ertl)
Groupes : comp.arch
Date : 31. Aug 2024, 16:03:47
Autres entêtes
Organisation : Institut fuer Computersprachen, Technische Universitaet Wien
Message-ID : <2024Aug31.170347@mips.complang.tuwien.ac.at>
References : 1 2 3 4 5 6 7
User-Agent : xrn 10.11
Bernd Linsel <bl1-thispartdoesnotbelonghere@gmx.com> writes:
Maybe my previous post was not clear enough: It's not a general UB
detector that I'd like to have integrated into the compiler (there are
static checker tools available that can nearly perfectly do that);

Undefined behaviour is something that is exercised at run-time.
That's why the "undefined behaviour sanitizers" insert run-time
checks.  And of course they only detect the behaviour when it is
actually exercised.  I.e., they usually will not detect overflowable
buffers, because your usual test inputs don't exercise those.

What do you mean with the static checker tools you mention?

instead, I'd like to get a warning when the compiler does something
other than you would expect when reading the code in a "do what I mean"
manner.

Of course the fans of compilers that do what nobody means found a
counterargument long ago: They claim that compilers would need psychic
powers to know what you mean.  So one way to specify what I guess you
mean with 'read the code in a "do what I mean" manner' is the
behaviour that the the compiler exhibits without "knowledge" coming
from the assumption that there is no undefined behaviour in the
program.  For a longer discussion read
<https://www.complang.tuwien.ac.at/papers/ertl17kps.pdf>.

And yes, compilers could actually produce information about
differences between such a compilation and a compilation where the
compiler assumes that undefined behaviour does not happen.

One way to use such information is if you then intend to run the
compiler in "Assume That Undefined Behaviour Does Not Happen" mode for
production code: check *every* case where the resulting code behaves
differently.  If the behaviour of the ATUBDNH compiler is not
according to your intentions, change the source code to avoid
undefined behaviour in such cases, forcing the ATUBDNH compiler to
behave as you intend.  If the behaviour of the ATUBDNH compiler is as
you intended, you can keep the source code as-is (but then you get the
same warning the next time 'round).  Or you can change the source code
in a way that results in the compiler not needing to ATUBDNH in order
to produce the code you would like (see below for examples).

Another way to use such information is if you intend to run the
compiler in don't-ATUBDNH mode for production code.  In that case you
only need to look at a few cases: those occuring in the
most-frequently executed code.  Again, for each difference there are
two cases: If your intention is only reflected in the don't-ATUBDNH
code, you don't have to do anything, or change the code such that the
warning goes away in the future (without changing the code).  If your
intention is also covered by the ATUBDNH case, you can change the code
to actually perform the optimization also in the don't-ATUBDNH
compiler.

Here are examples: Wang et al. [Section 3.3 of wang+12], found that in
all of SPECint 2006 there were only two places where the ATUBDNH made
a measurable difference to performance.  These were two inner loops.

In one case the code is

int k;
int *ic, *is;
...
for (k = 1; k <= M; k++) {
  ...
  ic[k] += is[k];
  ...
}

and the don't-ATUBDNH variant has a sign extension after the "k++"
that the ATUBDNH does not have.  Wang et al. suggest changing the type
of k to size_t to avoid this sign-extension operation.  After that
change ATUBDNH makes no difference to this loop.

The other loop is

quantum_reg *reg;
...
// reg->size: int
// reg->node[i].state: unsigned long long
for (i = 0; i < reg->size; i++)
  reg->node[i].state = ...;

Here ATUBDNH pulls the load of reg->size out of the loop (it assumes
that reg->size does not alias with reg->node[i].state).  Wang et
al. solved that by assigning reg->size to a variable outside the loop,
i.e., something like:

quantum_reg *reg;
...
long reg_size = reg->size
for (i = 0; i < reg_size; i++)
  reg->node[i].state = ...;


But once we are at that, why stop at optimizations suggestions coming
from ATUBDNH.  E.g., consider a loop similar to the second loop:

quantum_reg *reg;
...
// reg->size: int
// reg->node[i].state: int <==== HERE'S THE DIFFERENCE
for (i = 0; i < reg->size; i++)
  reg->node[i].state = ...;

In this case ATUBDNH would not allow pulling reg->size out of the
loop, yet you don't intend to ever alias reg->size with
reg->node[i].state.  A compiler could actually guess your intention,
and suggest that you may want to pull reg->size out (plus also mention
the caveats about possible aliasing).

So once we are there, we no longer need ATUBDNH, we just need
don't-ATUBDNH and a compiler option that produces manual-optimization
suggestions, ordered by the expected payoff (probably it's a good idea
to use profile data for this ordering).

I personally try to turn GCC into don't-ATUBDNH as far as possible
with options like "-fno-delete-null-pointer-checks
-fno-strict-aliasing -fno-strict-overflow".

@InProceedings{wang+12,
  author = {Xi Wang and Haogang Chen and Alvin Cheung and Zhihao Jia and Nickolai Zeldovich and M. Frans Kaashoek},
  title = {Undefined Behavior: What Happened to My Code?},
  booktitle = {Asia-Pacific Workshop on Systems (APSYS'12)},
  OPTpages = {},
  year = {2012},
  url1 = {http://homes.cs.washington.edu/~akcheung/getFile.php?file=apsys12.pdf},
  url2 = {http://people.csail.mit.edu/nickolai/papers/wang-undef-2012-08-21.pdf},
  OPTannote = {}
}

- anton
--
'Anyone trying for "industrial quality" ISA should avoid undefined behavior.'
  Mitch Alsup, <c17fcd89-f024-40e7-a594-88a85ac10d20o@googlegroups.com>

Date Sujet#  Auteur
27 Aug 24 * Computer architects leaving Intel...539Thomas Koenig
27 Aug 24 +- Re: Computer architects leaving Intel...1Michael S
27 Aug 24 +- Re: Computer architects leaving Intel...1Stephen Fuld
27 Aug 24 `* Re: Computer architects leaving Intel...536John Dallman
28 Aug 24  +* Re: Computer architects leaving Intel...529BGB
28 Aug 24  i`* Re: Computer architects leaving Intel...528MitchAlsup1
28 Aug 24  i `* Re: Computer architects leaving Intel...527BGB
28 Aug 24  i  +* Re: Computer architects leaving Intel...2Robert Finch
28 Aug 24  i  i`- Re: Computer architects leaving Intel...1BGB
28 Aug 24  i  `* Re: Computer architects leaving Intel...524MitchAlsup1
29 Aug 24  i   `* Re: Computer architects leaving Intel...523BGB
29 Aug 24  i    +* Re: Computer architects leaving Intel...511MitchAlsup1
29 Aug 24  i    i`* Re: Computer architects leaving Intel...510BGB
30 Aug 24  i    i +* Re: Computer architects leaving Intel...499John Dallman
30 Aug 24  i    i i+* Re: Computer architects leaving Intel...11Thomas Koenig
30 Aug 24  i    i ii+- Re: Computer architects leaving Intel...1Michael S
30 Aug 24  i    i ii+* Re: Computer architects leaving Intel...8Anton Ertl
30 Aug 24  i    i iii+* Re: Computer architects leaving Intel...2Michael S
30 Aug 24  i    i iiii`- Re: Computer architects leaving Intel...1Anton Ertl
30 Aug 24  i    i iii`* Re: Computer architects leaving Intel...5John Dallman
30 Aug 24  i    i iii `* Re: Computer architects leaving Intel...4Brett
30 Aug 24  i    i iii  +- Re: Computer architects leaving Intel...1John Dallman
2 Sep 24  i    i iii  `* Re: Computer architects leaving Intel...2Terje Mathisen
2 Sep 24  i    i iii   `- Re: Computer architects leaving Intel...1Thomas Koenig
30 Aug 24  i    i ii`- Re: Computer architects leaving Intel...1BGB
30 Aug 24  i    i i`* Re: Computer architects leaving Intel...487Anton Ertl
30 Aug 24  i    i i +* Re: Computer architects leaving Intel...302John Dallman
30 Aug 24  i    i i i`* Re: Computer architects leaving Intel...301David Brown
30 Aug 24  i    i i i +* Re: Computer architects leaving Intel...293Anton Ertl
30 Aug 24  i    i i i i`* Re: Computer architects leaving Intel...292Bernd Linsel
31 Aug 24  i    i i i i +- Re: Computer architects leaving Intel...1Thomas Koenig
31 Aug 24  i    i i i i `* Re: Computer architects leaving Intel...290Thomas Koenig
31 Aug 24  i    i i i i  +- Re: Computer architects leaving Intel...1Thomas Koenig
31 Aug 24  i    i i i i  `* Re: Computer architects leaving Intel...288Bernd Linsel
31 Aug 24  i    i i i i   +- Re: Computer architects leaving Intel...1Thomas Koenig
31 Aug 24  i    i i i i   +* Re: Computer architects leaving Intel...2Thomas Koenig
31 Aug 24  i    i i i i   i`- Re: Computer architects leaving Intel...1Bernd Linsel
31 Aug 24  i    i i i i   `* Re: Computer architects leaving Intel...284Anton Ertl
31 Aug 24  i    i i i i    +* Re: Computer architects leaving Intel...279Thomas Koenig
31 Aug 24  i    i i i i    i+* Re: Computer architects leaving Intel...157Bernd Linsel
31 Aug 24  i    i i i i    ii+* Re: Computer architects leaving Intel...153MitchAlsup1
1 Sep 24  i    i i i i    iii`* Re: Computer architects leaving Intel...152Stephen Fuld
2 Sep 24  i    i i i i    iii `* Re: Computer architects leaving Intel...151Terje Mathisen
2 Sep 24  i    i i i i    iii  `* Re: Computer architects leaving Intel...150Stephen Fuld
3 Sep 24  i    i i i i    iii   +* Re: Computer architects leaving Intel...139David Brown
3 Sep 24  i    i i i i    iii   i+* Re: Computer architects leaving Intel...108Stephen Fuld
4 Sep 24  i    i i i i    iii   ii`* Re: Computer architects leaving Intel...107David Brown
4 Sep 24  i    i i i i    iii   ii +* Re: Computer architects leaving Intel...103Terje Mathisen
4 Sep 24  i    i i i i    iii   ii i+* Re: Computer architects leaving Intel...101David Brown
4 Sep 24  i    i i i i    iii   ii ii+* Re: Computer architects leaving Intel...97jseigh
4 Sep 24  i    i i i i    iii   ii iii`* Re: Computer architects leaving Intel...96David Brown
4 Sep 24  i    i i i i    iii   ii iii `* Re: Computer architects leaving Intel...95Brett
4 Sep 24  i    i i i i    iii   ii iii  +- Re: Computer architects leaving Intel...1Thomas Koenig
4 Sep 24  i    i i i i    iii   ii iii  +- Re: Computer architects leaving Intel...1MitchAlsup1
5 Sep 24  i    i i i i    iii   ii iii  +* Re: Computer architects leaving Intel...8BGB
5 Sep 24  i    i i i i    iii   ii iii  i`* Re: Computer architects leaving Intel...7MitchAlsup1
5 Sep 24  i    i i i i    iii   ii iii  i `* Re: Computer architects leaving Intel...6David Brown
5 Sep 24  i    i i i i    iii   ii iii  i  `* Re: Computer architects leaving Intel...5Niklas Holsti
5 Sep 24  i    i i i i    iii   ii iii  i   `* Re: Computer architects leaving Intel...4David Brown
6 Sep 24  i    i i i i    iii   ii iii  i    `* Re: Computer architects leaving Intel...3BGB
6 Sep 24  i    i i i i    iii   ii iii  i     `* Re: Computer architects leaving Intel...2David Brown
9 Sep 24  i    i i i i    iii   ii iii  i      `- Re: Computer architects leaving Intel...1BGB
5 Sep 24  i    i i i i    iii   ii iii  +* Re: Computer architects leaving Intel...83David Brown
5 Sep 24  i    i i i i    iii   ii iii  i`* Re: Computer architects leaving Intel...82Terje Mathisen
5 Sep 24  i    i i i i    iii   ii iii  i +* Re: Computer architects leaving Intel...79David Brown
5 Sep 24  i    i i i i    iii   ii iii  i i+* Re: Computer architects leaving Intel...2Thomas Koenig
7 Sep 24  i    i i i i    iii   ii iii  i ii`- Re: Computer architects leaving Intel...1Tim Rentsch
5 Sep 24  i    i i i i    iii   ii iii  i i+* Re: Computer architects leaving Intel...74Terje Mathisen
5 Sep 24  i    i i i i    iii   ii iii  i ii+* Re: Computer architects leaving Intel...16David Brown
9 Sep 24  i    i i i i    iii   ii iii  i iii`* Re: Computer architects leaving Intel...15Terje Mathisen
9 Sep 24  i    i i i i    iii   ii iii  i iii +* Re: Computer architects leaving Intel...12David Brown
9 Sep 24  i    i i i i    iii   ii iii  i iii i`* Re: Computer architects leaving Intel...11Brett
10 Sep 24  i    i i i i    iii   ii iii  i iii i +* Re: Computer architects leaving Intel...5Terje Mathisen
10 Sep 24  i    i i i i    iii   ii iii  i iii i i`* Re: Computer architects leaving Intel...4Brett
10 Sep 24  i    i i i i    iii   ii iii  i iii i i +* Re: Computer architects leaving Intel...2Michael S
11 Sep 24  i    i i i i    iii   ii iii  i iii i i i`- Re: Computer architects leaving Intel...1Brett
11 Sep 24  i    i i i i    iii   ii iii  i iii i i `- Re: Computer architects leaving Intel...1Terje Mathisen
10 Sep 24  i    i i i i    iii   ii iii  i iii i `* Re: Computer architects leaving Intel...5David Brown
10 Sep 24  i    i i i i    iii   ii iii  i iii i  +* Re: Computer architects leaving Intel...3Anton Ertl
10 Sep 24  i    i i i i    iii   ii iii  i iii i  i`* Re: Computer architects leaving Intel...2David Brown
10 Sep 24  i    i i i i    iii   ii iii  i iii i  i `- Re: Computer architects leaving Intel...1Stefan Monnier
10 Sep 24  i    i i i i    iii   ii iii  i iii i  `- Re: Computer architects leaving Intel...1BGB
9 Sep 24  i    i i i i    iii   ii iii  i iii `* Re: Computer architects leaving Intel...2Michael S
10 Sep 24  i    i i i i    iii   ii iii  i iii  `- Re: Computer architects leaving Intel...1Michael S
5 Sep 24  i    i i i i    iii   ii iii  i ii+* Re: Computer architects leaving Intel...45Bernd Linsel
6 Sep 24  i    i i i i    iii   ii iii  i iii+- Re: Computer architects leaving Intel...1David Brown
9 Sep 24  i    i i i i    iii   ii iii  i iii+* Re: Computer architects leaving Intel...2Terje Mathisen
9 Sep 24  i    i i i i    iii   ii iii  i iiii`- Re: Computer architects leaving Intel...1Tim Rentsch
14 Sep 24  i    i i i i    iii   ii iii  i iii`* Re: Computer architects leaving Intel...41Kent Dickey
14 Sep 24  i    i i i i    iii   ii iii  i iii +* Re: Computer architects leaving Intel...32Anton Ertl
14 Sep 24  i    i i i i    iii   ii iii  i iii i+* Re: Computer architects leaving Intel...29MitchAlsup1
14 Sep 24  i    i i i i    iii   ii iii  i iii ii`* Re: Computer architects leaving Intel...28Thomas Koenig
15 Sep 24  i    i i i i    iii   ii iii  i iii ii `* Re: Computer architects leaving Intel...27David Brown
16 Sep 24  i    i i i i    iii   ii iii  i iii ii  +* Re: Computer architects leaving Intel...5Thomas Koenig
16 Sep 24  i    i i i i    iii   ii iii  i iii ii  i`* Re: Computer architects leaving Intel...4David Brown
16 Sep 24  i    i i i i    iii   ii iii  i iii ii  i `* Re: Computer architects leaving Intel...3Thomas Koenig
17 Sep 24  i    i i i i    iii   ii iii  i iii ii  i  +- Re: Upwards and downwards compatible, Computer architects leaving Intel...1John Levine
17 Sep 24  i    i i i i    iii   ii iii  i iii ii  i  `- Re: Computer architects leaving Intel...1David Brown
16 Sep 24  i    i i i i    iii   ii iii  i iii ii  `* Re: Computer architects leaving Intel...21Terje Mathisen
16 Sep 24  i    i i i i    iii   ii iii  i iii ii   `* Re: Computer architects leaving Intel...20David Brown
16 Sep 24  i    i i i i    iii   ii iii  i iii ii    +* Re: Computer architects leaving Intel...14Michael S
17 Sep 24  i    i i i i    iii   ii iii  i iii ii    `* Re: Computer architects leaving Intel...5Terje Mathisen
15 Sep 24  i    i i i i    iii   ii iii  i iii i`* Re: Computer architects leaving Intel...2BGB
14 Sep 24  i    i i i i    iii   ii iii  i iii +* Re: Computer architects leaving Intel...3Thomas Koenig
16 Sep 24  i    i i i i    iii   ii iii  i iii `* Re: Computer architects leaving Intel...5Tim Rentsch
6 Sep 24  i    i i i i    iii   ii iii  i ii+* Re: Computer architects leaving Intel...3Tim Rentsch
7 Sep 24  i    i i i i    iii   ii iii  i ii`* Re: Computer architects leaving Intel...9Chris M. Thomasson
5 Sep 24  i    i i i i    iii   ii iii  i i`* Re: Computer architects leaving Intel...2MitchAlsup1
5 Sep 24  i    i i i i    iii   ii iii  i `* Re: Computer architects leaving Intel...2MitchAlsup1
7 Sep 24  i    i i i i    iii   ii iii  `- Re: Computer architects leaving Intel...1Tim Rentsch
4 Sep 24  i    i i i i    iii   ii ii`* Re: Computer architects leaving Intel...3Thomas Koenig
6 Sep 24  i    i i i i    iii   ii i`- Re: Computer architects leaving Intel...1Chris M. Thomasson
4 Sep 24  i    i i i i    iii   ii +- Re: Computer architects leaving Intel...1jseigh
13 Sep 24  i    i i i i    iii   ii `* Re: Computer architects leaving Intel...2Stephen Fuld
3 Sep 24  i    i i i i    iii   i`* Re: Computer architects leaving Intel...30Stefan Monnier
3 Sep 24  i    i i i i    iii   `* Re: Computer architects leaving Intel...10Terje Mathisen
31 Aug 24  i    i i i i    ii`* Re: Computer architects leaving Intel...3Thomas Koenig
1 Sep 24  i    i i i i    i`* Re: Computer architects leaving Intel...121David Brown
1 Sep 24  i    i i i i    +* Re: Computer architects leaving Intel...3John Dallman
3 Sep 24  i    i i i i    `- Re: Computer architects leaving Intel...1Stefan Monnier
30 Aug 24  i    i i i +- Re: Computer architects leaving Intel...1MitchAlsup1
30 Aug 24  i    i i i +* Re: Computer architects leaving Intel...4Stefan Monnier
30 Aug 24  i    i i i `* Re: Computer architects leaving Intel...2John Dallman
8 Sep 24  i    i i `* Re: Computer architects leaving Intel...184Tim Rentsch
30 Aug 24  i    i `* Re: Computer architects leaving Intel...10MitchAlsup1
31 Aug 24  i    `* Re: Computer architects leaving Intel...11Paul A. Clayton
29 Aug 24  `* Re: Computer architects leaving Intel...6Anton Ertl

Haut de la page

Les messages affichés proviennent d'usenet.

NewsPortal