RE: GAS

RE: GAS

[Zagorodnev, Grigory]

Hi, Nick!
Hash statistics says that increasing the size we really reduce average
chain length and number of hash comparisons. However I was not able to
measure any comptime change on 1.3MB of x86 assembly source code.
Hopefully bigger sources will expose the gain. Therefore we may keep
default hash size value unchanged.

Also there is a little problem with the patch: new command line option
has no affect on symbol table and local symbol table hashes because
routine parse_args gets called after symbol_begin in "as.c", where these
hashes allocated. 

Best regards!
- Grigory

>-----Original Message-----
>From: Nick Clifton [mailto:nickc@redhat.com]
>Sent: Wednesday, April 06, 2005 3:15 PM
>To: Zagorodnev, Grigory
>Cc: Stas Kiselev; binutils@sources.redhat.com
>Subject: Re: GAS
>
>Hi Grigory,
>
>> Similar issue was reported against 'ld' 1.5 years ago. Simple change
of
>> default hash size from 4K to 64K gave 2x link time improvement.
>> http://lists.gnu.org/archive/html/bug-gnu-utils/2003-08/msg00153.html
>>
>> Thus the patch may look like this:
>>
>> $ diff gas/hash.c.org gas/hash.c
>>
>> 38c38
>> < #define DEFAULT_SIZE (4051)
>> ---
>>>#define DEFAULT_SIZE (65537)
>
>We might as well do this properly.  Please could you try the attached
>patch which as well as changing the default hash size adds two new
>switches: --hash-size=<NUMBER> and --reduce-memory-overheads which
>behave in much the same way as their linker namesakes.
>
>If the patch works for you and there are no objections then I will
apply
>it next week.
>
>Cheers
>   Nick
>
>gas/ChangeLog
>
>2005-04-06  Nick Clifton  <nickc@redhat.com>
>
>	* hash.c (DEFAULT_SIZE): Delete.  Replace with:
>	(gas_hash_table_size): New static variable.
>	(set_gas_hash_table_size): New function:  Records a requested
size
>	for the hash tables.
>	(get_gas_hash_table_size): New function: Return a prime number
>	near the requested size of the hash table.
>	(hash_new): Use get_gas_hash_table_size.
>	* hash.h: Add a prototype for set_gas_hash_table_size.
>	* as.c (show_usage): Add description of new switches:
--hash-size
>	and --reduce-memory-overheads.
>	(option_values): Add OPTION_HASH_TABLE_SIZE and
>	OPTION_REDUCE_MEMORY_OVERHEADS.
>	(std_longpopts): Add entries for the new options.
>	(parse_args): Handle the new options.
>	* Makefile.am: Add a dependency of as.c on hash.h.
>	* Makefile.in: Regenerate.
>	* doc/as.texinfo: Document the new switches.
>	* NEWS: Mention the new switches.

Re: GAS

[Nick Clifton]

[-- Attachment #1: Type: text/plain, Size: 886 bytes --]

Hi Griogry,

> Also there is a little problem with the patch: new command line option
> has no affect on symbol table and local symbol table hashes because
> routine parse_args gets called after symbol_begin in "as.c", where these
> hashes allocated. 

Ah - good point.  There was also a small problem with non-BFD assemblers 
in that the patch used the bfd_size_type.  Therefore I am applying the 
attached patch to fix both of these problems.

I checked - moving the parse_args before the symbol_begin does not 
appear to have any adverse effects.

Cheers
   Nick

gas/ChangeLog
2005-04-14  Nick Clifton  <nickc@redhat.com>

	* as.c (main): Move parse_args before symbol_begin and frag_init
	so that the hash table size can be set before it is used.
	* hash.c: Use an unsigned long type for the size of the hash
	tables.
	* hash.h (set_gas_hash_table_size): Update the prototype.	




[-- Attachment #2: gas.patch --]
[-- Type: text/plain, Size: 3872 bytes --]

Index: gas/as.c
===================================================================
RCS file: /cvs/src/src/gas/as.c,v
retrieving revision 1.59
diff -c -3 -p -r1.59 as.c
*** gas/as.c	12 Apr 2005 08:42:38 -0000	1.59
--- gas/as.c	14 Apr 2005 08:56:36 -0000
*************** the GNU General Public License.  This pr
*** 885,891 ****
  
  	case OPTION_HASH_TABLE_SIZE:
  	  {
! 	    bfd_size_type new_size;
  
              new_size = strtoul (optarg, NULL, 0);
              if (new_size)
--- 885,891 ----
  
  	case OPTION_HASH_TABLE_SIZE:
  	  {
! 	    unsigned long new_size;
  
              new_size = strtoul (optarg, NULL, 0);
              if (new_size)
*************** main (int argc, char ** argv)
*** 1114,1123 ****
  #endif
  
    PROGRESS (1);
    symbol_begin ();
    frag_init ();
    subsegs_begin ();
-   parse_args (&argc, &argv);
    read_begin ();
    input_scrub_begin ();
    expr_begin ();
--- 1114,1125 ----
  #endif
  
    PROGRESS (1);
+   /* Call parse_args before any of the init/begin functions
+      so that switches like --hash-size can be honored.  */
+   parse_args (&argc, &argv);
    symbol_begin ();
    frag_init ();
    subsegs_begin ();
    read_begin ();
    input_scrub_begin ();
    expr_begin ();
Index: gas/hash.c
===================================================================
RCS file: /cvs/src/src/gas/hash.c,v
retrieving revision 1.14
diff -c -3 -p -r1.14 hash.c
*** gas/hash.c	12 Apr 2005 08:42:38 -0000	1.14
--- gas/hash.c	14 Apr 2005 08:56:36 -0000
*************** struct hash_control {
*** 73,92 ****
     switch --reduce-memory-overheads, or set to other values by using
     the --hash-size=<NUMBER> switch.  */
  
! static unsigned int gas_hash_table_size = 65537;
  
  void
! set_gas_hash_table_size (unsigned int size)
  {
    gas_hash_table_size = size;
  }
  
  /* FIXME: This function should be amalgmated with bfd/hash.c:bfd_hash_set_default_size().  */
! static unsigned int
  get_gas_hash_table_size (void)
  {
    /* Extend this prime list if you want more granularity of hash table size.  */
!   static const unsigned int hash_size_primes[] =
      {
        1021, 4051, 8599, 16699, 65537
      };
--- 73,92 ----
     switch --reduce-memory-overheads, or set to other values by using
     the --hash-size=<NUMBER> switch.  */
  
! static unsigned long gas_hash_table_size = 65537;
  
  void
! set_gas_hash_table_size (unsigned long size)
  {
    gas_hash_table_size = size;
  }
  
  /* FIXME: This function should be amalgmated with bfd/hash.c:bfd_hash_set_default_size().  */
! static unsigned long
  get_gas_hash_table_size (void)
  {
    /* Extend this prime list if you want more granularity of hash table size.  */
!   static const unsigned long hash_size_primes[] =
      {
        1021, 4051, 8599, 16699, 65537
      };
*************** get_gas_hash_table_size (void)
*** 107,115 ****
  struct hash_control *
  hash_new (void)
  {
!   unsigned int size;
    struct hash_control *ret;
-   unsigned int alloc;
  
    size = get_gas_hash_table_size ();
  
--- 107,115 ----
  struct hash_control *
  hash_new (void)
  {
!   unsigned long size;
!   unsigned long alloc;
    struct hash_control *ret;
  
    size = get_gas_hash_table_size ();
  
Index: gas/hash.h
===================================================================
RCS file: /cvs/src/src/gas/hash.h,v
retrieving revision 1.7
diff -c -3 -p -r1.7 hash.h
*** gas/hash.h	12 Apr 2005 08:42:38 -0000	1.7
--- gas/hash.h	14 Apr 2005 08:56:36 -0000
*************** struct hash_control;
*** 26,32 ****
  
  /* Set the size of the hash table used.  */
  
! void set_gas_hash_table_size (unsigned int);
  
  /* Create a hash table.  This return a control block.  */
  
--- 26,32 ----
  
  /* Set the size of the hash table used.  */
  
! void set_gas_hash_table_size (unsigned long);
  
  /* Create a hash table.  This return a control block.  */
  

Re[2]: GAS

[Stas Kiselev]

Hi Nick,
Ok, I have already solved problems with hash. Now I am working with reading .asn
and writing .obj strategies. What can be changed in order to reduce time ?


P.S. What a dataflow of GAS ?

Thank you, Stanislav

> 
> Hi Griogry,
> 
> > Also there is a little problem with the patch: new command line option
> > has no affect on symbol table and local symbol table hashes because
> > routine parse_args gets called after symbol_begin in "as.c", where these
> > hashes allocated. 
> 
> Ah - good point.  There was also a small problem with non-BFD assemblers 
> in that the patch used the bfd_size_type.  Therefore I am applying the 
> attached patch to fix both of these problems.
> 
> I checked - moving the parse_args before the symbol_begin does not 
> appear to have any adverse effects.
> 
> Cheers
>    Nick
> 
> gas/ChangeLog
> 2005-04-14  Nick Clifton  <nickc@redhat.com>
> 
> 	* as.c (main): Move parse_args before symbol_begin and frag_init
> 	so that the hash table size can be set before it is used.
> 	* hash.c: Use an unsigned long type for the size of the hash
> 	tables.
> 	* hash.h (set_gas_hash_table_size): Update the prototype.	
> 
> 
> 
> 
> 
> Index: gas/as.c
> ===================================================================
> RCS file: /cvs/src/src/gas/as.c,v
> retrieving revision 1.59
> diff -c -3 -p -r1.59 as.c
> *** gas/as.c	12 Apr 2005 08:42:38 -0000	1.59
> --- gas/as.c	14 Apr 2005 08:56:36 -0000
> *************** the GNU General Public License.  This pr
> *** 885,891 ****
>   
>   	case OPTION_HASH_TABLE_SIZE:
>   	  {
> ! 	    bfd_size_type new_size;
>   
>               new_size = strtoul (optarg, NULL, 0);
>               if (new_size)
> --- 885,891 ----
>   
>   	case OPTION_HASH_TABLE_SIZE:
>   	  {
> ! 	    unsigned long new_size;
>   
>               new_size = strtoul (optarg, NULL, 0);
>               if (new_size)
> *************** main (int argc, char ** argv)
> *** 1114,1123 ****
>   #endif
>   
>     PROGRESS (1);
>     symbol_begin ();
>     frag_init ();
>     subsegs_begin ();
> -   parse_args (&argc, &argv);
>     read_begin ();
>     input_scrub_begin ();
>     expr_begin ();
> --- 1114,1125 ----
>   #endif
>   
>     PROGRESS (1);
> +   /* Call parse_args before any of the init/begin functions
> +      so that switches like --hash-size can be honored.  */
> +   parse_args (&argc, &argv);
>     symbol_begin ();
>     frag_init ();
>     subsegs_begin ();
>     read_begin ();
>     input_scrub_begin ();
>     expr_begin ();
> Index: gas/hash.c
> ===================================================================
> RCS file: /cvs/src/src/gas/hash.c,v
> retrieving revision 1.14
> diff -c -3 -p -r1.14 hash.c
> *** gas/hash.c	12 Apr 2005 08:42:38 -0000	1.14
> --- gas/hash.c	14 Apr 2005 08:56:36 -0000
> *************** struct hash_control {
> *** 73,92 ****
>      switch --reduce-memory-overheads, or set to other values by using
>      the --hash-size=<NUMBER> switch.  */
>   
> ! static unsigned int gas_hash_table_size = 65537;
>   
>   void
> ! set_gas_hash_table_size (unsigned int size)
>   {
>     gas_hash_table_size = size;
>   }
>   
>   /* FIXME: This function should be amalgmated with bfd/hash.c:bfd_hash_set_default_size().  */
> ! static unsigned int
>   get_gas_hash_table_size (void)
>   {
>     /* Extend this prime list if you want more granularity of hash table size.  */
> !   static const unsigned int hash_size_primes[] =
>       {
>         1021, 4051, 8599, 16699, 65537
>       };
> --- 73,92 ----
>      switch --reduce-memory-overheads, or set to other values by using
>      the --hash-size=<NUMBER> switch.  */
>   
> ! static unsigned long gas_hash_table_size = 65537;
>   
>   void
> ! set_gas_hash_table_size (unsigned long size)
>   {
>     gas_hash_table_size = size;
>   }
>   
>   /* FIXME: This function should be amalgmated with bfd/hash.c:bfd_hash_set_default_size().  */
> ! static unsigned long
>   get_gas_hash_table_size (void)
>   {
>     /* Extend this prime list if you want more granularity of hash table size.  */
> !   static const unsigned long hash_size_primes[] =
>       {
>         1021, 4051, 8599, 16699, 65537
>       };
> *************** get_gas_hash_table_size (void)
> *** 107,115 ****
>   struct hash_control *
>   hash_new (void)
>   {
> !   unsigned int size;
>     struct hash_control *ret;
> -   unsigned int alloc;
>   
>     size = get_gas_hash_table_size ();
>   
> --- 107,115 ----
>   struct hash_control *
>   hash_new (void)
>   {
> !   unsigned long size;
> !   unsigned long alloc;
>     struct hash_control *ret;
>   
>     size = get_gas_hash_table_size ();
>   
> Index: gas/hash.h
> ===================================================================
> RCS file: /cvs/src/src/gas/hash.h,v
> retrieving revision 1.7
> diff -c -3 -p -r1.7 hash.h
> *** gas/hash.h	12 Apr 2005 08:42:38 -0000	1.7
> --- gas/hash.h	14 Apr 2005 08:56:36 -0000
> *************** struct hash_control;
> *** 26,32 ****
>   
>   /* Set the size of the hash table used.  */
>   
> ! void set_gas_hash_table_size (unsigned int);
>   
>   /* Create a hash table.  This return a control block.  */
>   
> --- 26,32 ----
>   
>   /* Set the size of the hash table used.  */
>   
> ! void set_gas_hash_table_size (unsigned long);
>   
>   /* Create a hash table.  This return a control block.  */
>   
> 
> 

Re: GAS

[Nick Clifton]

Hi Stanislav,

> Ok, I have already solved problems with hash. Now I am working with reading .asn
> and writing .obj strategies. What can be changed in order to reduce time ?

Sorry - I have not looked into that aspect of the assembler - you will 
have to profile it yourself to find out where the hot spots are.

Cheers
   Nick

RE: Re[2]: GAS

[Zagorodnev, Grigory]

>The idea is: huge input file implies huge symbol tables, that results
>more hash function collisions. Thus increasing the DEFAULT_SIZE of hash
>we can reduce the number of collisions and therefore speedup execution.
>Stas, would you try this on your sources, please?

Similar issue was reported against 'ld' 1.5 years ago. Simple change of
default hash size from 4K to 64K gave 2x link time improvement.
http://lists.gnu.org/archive/html/bug-gnu-utils/2003-08/msg00153.html

Thus the patch may look like this:

$ diff gas/hash.c.org gas/hash.c

38c38
< #define DEFAULT_SIZE (4051)
---
> #define DEFAULT_SIZE (65537)


- Grigory

Re: GAS

[Nick Clifton]

[-- Attachment #1: Type: text/plain, Size: 1636 bytes --]

Hi Grigory,

> Similar issue was reported against 'ld' 1.5 years ago. Simple change of
> default hash size from 4K to 64K gave 2x link time improvement.
> http://lists.gnu.org/archive/html/bug-gnu-utils/2003-08/msg00153.html
> 
> Thus the patch may look like this:
> 
> $ diff gas/hash.c.org gas/hash.c
> 
> 38c38
> < #define DEFAULT_SIZE (4051)
> ---
>>#define DEFAULT_SIZE (65537)

We might as well do this properly.  Please could you try the attached 
patch which as well as changing the default hash size adds two new 
switches: --hash-size=<NUMBER> and --reduce-memory-overheads which 
behave in much the same way as their linker namesakes.

If the patch works for you and there are no objections then I will apply 
it next week.

Cheers
   Nick

gas/ChangeLog

2005-04-06  Nick Clifton  <nickc@redhat.com>

	* hash.c (DEFAULT_SIZE): Delete.  Replace with:
	(gas_hash_table_size): New static variable.
	(set_gas_hash_table_size): New function:  Records a requested size
	for the hash tables.
	(get_gas_hash_table_size): New function: Return a prime number
	near the requested size of the hash table.
	(hash_new): Use get_gas_hash_table_size.
	* hash.h: Add a prototype for set_gas_hash_table_size.
	* as.c (show_usage): Add description of new switches: --hash-size
	and --reduce-memory-overheads.
	(option_values): Add OPTION_HASH_TABLE_SIZE and
	OPTION_REDUCE_MEMORY_OVERHEADS.
	(std_longpopts): Add entries for the new options.
	(parse_args): Handle the new options.
	* Makefile.am: Add a dependency of as.c on hash.h.
	* Makefile.in: Regenerate.
	* doc/as.texinfo: Document the new switches.
	* NEWS: Mention the new switches.


[-- Attachment #2: gas.hash-size.patch --]
[-- Type: text/plain, Size: 11570 bytes --]

Index: gas/Makefile.am
===================================================================
RCS file: /cvs/src/src/gas/Makefile.am,v
retrieving revision 1.104
diff -c -3 -p -r1.104 Makefile.am
*** gas/Makefile.am	21 Mar 2005 15:35:30 -0000	1.104
--- gas/Makefile.am	6 Apr 2005 11:06:00 -0000
*************** BMKDEP = #DO NOT PUT ANYTHING BETWEEN TH
*** 2436,2442 ****
  #MKDEP    DO NOT PUT ANYTHING BETWEEN THIS LINE AND THE MATCHING WARNING BELOW.
  app.o: app.c $(INCDIR)/symcat.h
  as.o: as.c $(INCDIR)/symcat.h subsegs.h $(INCDIR)/obstack.h \
!   output-file.h sb.h macro.h dwarf2dbg.h dw2gencfi.h \
    $(INCDIR)/elf/dwarf2.h $(BFDVER_H)
  atof-generic.o: atof-generic.c $(INCDIR)/symcat.h $(INCDIR)/safe-ctype.h
  bignum-copy.o: bignum-copy.c $(INCDIR)/symcat.h
--- 2436,2442 ----
  #MKDEP    DO NOT PUT ANYTHING BETWEEN THIS LINE AND THE MATCHING WARNING BELOW.
  app.o: app.c $(INCDIR)/symcat.h
  as.o: as.c $(INCDIR)/symcat.h subsegs.h $(INCDIR)/obstack.h \
!   output-file.h sb.h macro.h dwarf2dbg.h dw2gencfi.h hash.h \
    $(INCDIR)/elf/dwarf2.h $(BFDVER_H)
  atof-generic.o: atof-generic.c $(INCDIR)/symcat.h $(INCDIR)/safe-ctype.h
  bignum-copy.o: bignum-copy.c $(INCDIR)/symcat.h

Index: gas/NEWS
===================================================================
RCS file: /cvs/src/src/gas/NEWS,v
retrieving revision 1.74
diff -c -3 -p -r1.74 NEWS
*** gas/NEWS	1 Apr 2005 14:17:16 -0000	1.74
--- gas/NEWS	6 Apr 2005 11:06:01 -0000
***************
*** 1,5 ****
--- 1,9 ----
  -*- text -*-
  
+ * Added --reduce-memory-overheads switch to reduce the size of the hash
+   tables used, at the expense of longer assembly times, and
+   --hash-size=<NUMBER> to set the size of the hash tables used by gas.
+ 
  * Support the .f_floating, .d_floating, .g_floating and .h_floating directives
    for the VAX target in order to be more compatible with the VAX MACRO
    assembler.

Index: gas/as.c
===================================================================
RCS file: /cvs/src/src/gas/as.c,v
retrieving revision 1.58
diff -c -3 -p -r1.58 as.c
*** gas/as.c	3 Mar 2005 01:29:52 -0000	1.58
--- gas/as.c	6 Apr 2005 11:06:02 -0000
***************
*** 42,47 ****
--- 42,48 ----
  #include "macro.h"
  #include "dwarf2dbg.h"
  #include "dw2gencfi.h"
+ #include "hash.h"
  
  #ifdef BFD_ASSEMBLER
  #include "bfdver.h"
*************** Options:\n\
*** 292,297 ****
--- 293,300 ----
    fprintf (stream, _("\
    --gdwarf-2              generate DWARF2 debugging information\n"));
    fprintf (stream, _("\
+   --hash-size=<value>     set the hash table size close to <value>\n"));
+   fprintf (stream, _("\
    --help                  show this message and exit\n"));
    fprintf (stream, _("\
    --target-help           show target specific options\n"));
*************** Options:\n\
*** 314,319 ****
--- 317,326 ----
    fprintf (stream, _("\
    -R                      fold data section into text section\n"));
    fprintf (stream, _("\
+   --reduce-memory-overheads \n\
+                           prefer smaller memory use at the cost of longer\n\
+                           assembly times\n"));
+   fprintf (stream, _("\
    --statistics            print various measured statistics from execution\n"));
    fprintf (stream, _("\
    --strip-local-absolute  strip local absolute symbols\n"));
*************** parse_args (int * pargc, char *** pargv)
*** 426,431 ****
--- 433,440 ----
        OPTION_EXECSTACK,
        OPTION_NOEXECSTACK,
        OPTION_ALTERNATE,
+       OPTION_HASH_TABLE_SIZE,
+       OPTION_REDUCE_MEMORY_OVERHEADS,
        OPTION_WARN_FATAL
      /* When you add options here, check that they do
         not collide with OPTION_MD_BASE.  See as.h.  */
*************** parse_args (int * pargc, char *** pargv)
*** 457,462 ****
--- 466,472 ----
      ,{"gen-debug", no_argument, NULL, 'g'}
      ,{"gstabs", no_argument, NULL, OPTION_GSTABS}
      ,{"gstabs+", no_argument, NULL, OPTION_GSTABS_PLUS}
+     ,{"hash-size", required_argument, NULL, OPTION_HASH_TABLE_SIZE}
      ,{"help", no_argument, NULL, OPTION_HELP}
      /* New option for extending instruction set (see also -t above).
         The "-t file" or "--itbl file" option extends the basic set of
*************** parse_args (int * pargc, char *** pargv)
*** 478,483 ****
--- 488,494 ----
      ,{"mri", no_argument, NULL, 'M'}
      ,{"nocpp", no_argument, NULL, OPTION_NOCPP}
      ,{"no-warn", no_argument, NULL, 'W'}
+     ,{"reduce-memory-overheads", no_argument, NULL, OPTION_REDUCE_MEMORY_OVERHEADS}
      ,{"statistics", no_argument, NULL, OPTION_STATISTICS}
      ,{"strip-local-absolute", no_argument, NULL, OPTION_STRIP_LOCAL_ABSOLUTE}
      ,{"version", no_argument, NULL, OPTION_VERSION}
*************** the GNU General Public License.  This pr
*** 865,870 ****
--- 876,899 ----
  	case 'X':
  	  /* -X means treat warnings as errors.  */
  	  break;
+ 
+ 	case OPTION_REDUCE_MEMORY_OVERHEADS:
+ 	  /* The only change we make at the moment is to reduce
+ 	     the size of the hash tables that we use.  */
+ 	  set_gas_hash_table_size (4051);
+ 	  break;
+ 
+ 	case OPTION_HASH_TABLE_SIZE:
+ 	  {
+ 	    bfd_size_type new_size;
+ 
+             new_size = strtoul (optarg, NULL, 0);
+             if (new_size)
+               set_gas_hash_table_size (new_size);
+             else
+               as_fatal (_("--hash-size needs a numeric argument"));
+ 	    break;
+ 	  }
  	}
      }
  
Index: gas/hash.c
===================================================================
RCS file: /cvs/src/src/gas/hash.c,v
retrieving revision 1.13
diff -c -3 -p -r1.13 hash.c
*** gas/hash.c	3 Mar 2005 01:29:53 -0000	1.13
--- gas/hash.c	6 Apr 2005 11:06:02 -0000
***************
***************
*** 33,42 ****
  #include "safe-ctype.h"
  #include "obstack.h"
  
- /* The default number of entries to use when creating a hash table.  */
- 
- #define DEFAULT_SIZE (4051)
- 
  /* An entry in a hash table.  */
  
  struct hash_entry {
--- 33,38 ----
*************** struct hash_control {
*** 72,77 ****
--- 68,107 ----
  #endif /* HASH_STATISTICS */
  };
  
+ /* The default number of entries to use when creating a hash table.
+    Note this value can be reduced to 4051 by using the command line
+    switch --reduce-memory-overheads, or set to other values by using
+    the --hash-size=<NUMBER> switch.  */
+ 
+ static unsigned int gas_hash_table_size = 65537;
+ 
+ void
+ set_gas_hash_table_size (unsigned int size)
+ {
+   gas_hash_table_size = size;
+ }
+ 
+ /* FIXME: This function should be amalgmated with bfd/hash.c:bfd_hash_set_default_size().  */
+ static unsigned int
+ get_gas_hash_table_size (void)
+ {
+   /* Extend this prime list if you want more granularity of hash table size.  */
+   static const unsigned int hash_size_primes[] =
+     {
+       1021, 4051, 8599, 16699, 65537
+     };
+   unsigned int index;
+ 
+   /* Work out the best prime number near the hash_size.
+      FIXME: This could be a more sophisticated algorithm,
+      but is it really worth implementing it ?   */
+   for (index = 0; index < ARRAY_SIZE (hash_size_primes) - 1; ++index)
+     if (gas_hash_table_size <= hash_size_primes[index])
+       break;
+ 
+   return hash_size_primes[index];
+ }
+ 
  /* Create a hash table.  This return a control block.  */
  
  struct hash_control *
*************** hash_new (void)
*** 81,92 ****
    struct hash_control *ret;
    unsigned int alloc;
  
!   size = DEFAULT_SIZE;
  
!   ret = (struct hash_control *) xmalloc (sizeof *ret);
    obstack_begin (&ret->memory, chunksize);
    alloc = size * sizeof (struct hash_entry *);
!   ret->table = (struct hash_entry **) obstack_alloc (&ret->memory, alloc);
    memset (ret->table, 0, alloc);
    ret->size = size;
  
--- 111,122 ----
    struct hash_control *ret;
    unsigned int alloc;
  
!   size = get_gas_hash_table_size ();
  
!   ret = xmalloc (sizeof *ret);
    obstack_begin (&ret->memory, chunksize);
    alloc = size * sizeof (struct hash_entry *);
!   ret->table = obstack_alloc (&ret->memory, alloc);
    memset (ret->table, 0, alloc);
    ret->size = size;
  
Index: gas/hash.h
===================================================================
RCS file: /cvs/src/src/gas/hash.h,v
retrieving revision 1.6
diff -c -3 -p -r1.6 hash.h
*** gas/hash.h	3 Mar 2005 01:29:53 -0000	1.6
--- gas/hash.h	6 Apr 2005 11:06:02 -0000
***************
*** 24,29 ****
--- 24,33 ----
  
  struct hash_control;
  
+ /* Set the size of the hash table used.  */
+ 
+ void set_gas_hash_table_size (unsigned int);
+ 
  /* Create a hash table.  This return a control block.  */
  
  extern struct hash_control *hash_new (void);

Index: gas/doc/as.texinfo
===================================================================
RCS file: /cvs/src/src/gas/doc/as.texinfo,v
retrieving revision 1.118
diff -c -3 -p -r1.118 as.texinfo
*** gas/doc/as.texinfo	28 Mar 2005 22:34:17 -0000	1.118
--- gas/doc/as.texinfo	6 Apr 2005 11:06:04 -0000
*************** gcc(1), ld(1), and the Info entries for 
*** 227,240 ****
  @smallexample
  @c man begin SYNOPSIS
  @value{AS} [@b{-a}[@b{cdhlns}][=@var{file}]] [@b{--alternate}] [@b{-D}]
!  [@b{--defsym} @var{sym}=@var{val}] [@b{-f}] [@b{-g}] [@b{--gstabs}] [@b{--gstabs+}]
!  [@b{--gdwarf-2}] [@b{--help}] [@b{-I} @var{dir}] [@b{-J}] [@b{-K}] [@b{-L}]
!  [@b{--listing-lhs-width}=@var{NUM}] [@b{--listing-lhs-width2}=@var{NUM}]
!  [@b{--listing-rhs-width}=@var{NUM}] [@b{--listing-cont-lines}=@var{NUM}]
!  [@b{--keep-locals}] [@b{-o} @var{objfile}] [@b{-R}] [@b{--statistics}] [@b{-v}]
!  [@b{-version}] [@b{--version}] [@b{-W}] [@b{--warn}] [@b{--fatal-warnings}] 
!  [@b{-w}] [@b{-x}] [@b{-Z}] [@b{--target-help}] [@var{target-options}] 
!  [@b{--}|@var{files} @dots{}]
  @c
  @c Target dependent options are listed below.  Keep the list sorted.
  @c Add an empty line for separation. 
--- 227,241 ----
  @smallexample
  @c man begin SYNOPSIS
  @value{AS} [@b{-a}[@b{cdhlns}][=@var{file}]] [@b{--alternate}] [@b{-D}]
!  [@b{--defsym} @var{sym}=@var{val}] [@b{-f}] [@b{-g}] [@b{--gstabs}]
!  [@b{--gstabs+}] [@b{--gdwarf-2}] [@b{--help}] [@b{-I} @var{dir}] [@b{-J}]
!  [@b{-K}] [@b{-L}] [@b{--listing-lhs-width}=@var{NUM}]
!  [@b{--listing-lhs-width2}=@var{NUM}] [@b{--listing-rhs-width}=@var{NUM}]
!  [@b{--listing-cont-lines}=@var{NUM}] [@b{--keep-locals}] [@b{-o}
!  @var{objfile}] [@b{-R}] [@b{--reduce-memory-overheads}] [@b{--statistics}]
!  [@b{-v}] [@b{-version}] [@b{--version}] [@b{-W}] [@b{--warn}]
!  [@b{--fatal-warnings}] [@b{-w}] [@b{-x}] [@b{-Z}] [@b{--target-help}]
!  [@var{target-options}] [@b{--}|@var{files} @dots{}]
  @c
  @c Target dependent options are listed below.  Keep the list sorted.
  @c Add an empty line for separation. 
*************** Name the object-file output from @comman
*** 554,559 ****
--- 555,572 ----
  @item -R
  Fold the data section into the text section.
  
+ @kindex --hash-size=@var{number}
+ Set the default size of GAS's hash tables to a prime number close to
+ @var{number}.  Increasing this value can reduce the length of time it takes the
+ assembler to perform its tasks, at the expense of increasing the assembler's
+ memory requirements.  Similarly reducing this value can reduce the memory
+ requirements at the expense of speed.
+ 
+ @item --reduce-memory-overheads
+ This option reduces GAS's memory requirements, at the expense of making the
+ assembly processes slower.  Currently this switch is a synonym for
+ @samp{--hash-size=4051}, but in the future it may have other effects as well.
+ 
  @item --statistics
  Print the maximum space (in bytes) and total time (in seconds) used by
  assembly.

RE: GAS

[Zagorodnev, Grigory]

Stas,
Have you tried profiling gas to expose hotspots? 200mb of (plain?) code
seems to be good material for this.

---
Grigory Zagorodnev
Intel Corporation

>-----Original Message-----
>From: binutils-owner@sources.redhat.com [mailto:binutils-
>owner@sources.redhat.com] On Behalf Of Stas Kiselev
>Sent: Monday, April 04, 2005 6:52 PM
>To: binutils@sources.redhat.com
>Subject: GAS
>
>Hello,
>
>   I use AS for generating object files where time is
>critical.
>I mean I have to speed up gas as much as possible. As for
>input I give AS
>very big .asm files with size more than 200mb. so it works
>very slow. I
>tried to cut input file to several small .asm files - AS
>became work more
>faster but in general AS work very slow. How can I speed up
>it ? What kind
>of weak points can I find and fix ?
>
>P.S. Where can I get whole description of AS architecture?
>for instance I have
>descriptinon for lcc compilator - "A retargetable C
>compiler: design and implementation"
>
>Thank you, Stanislav

GAS

[Stas Kiselev]

Hello,

   I use AS for generating object files where time is
critical.
I mean I have to speed up gas as much as possible. As for
input I give AS
very big .asm files with size more than 200mb. so it works
very slow. I
tried to cut input file to several small .asm files - AS
became work more
faster but in general AS work very slow. How can I speed up
it ? What kind
of weak points can I find and fix ?

P.S. Where can I get whole description of AS architecture?
for instance I have
descriptinon for lcc compilator - "A retargetable C
compiler: design and implementation"

Thank you, Stanislav

RE: GAS

[Arun Vishwanathan]

No universal assembly language format. But it should not be difficult
to adapt to syntaxes once you learn one format. Intel one is probably
simpler so you can start with that. 

This maybe a good links to start with if you are programming on Linux.
http://linuxassembly.org/
http://www.tldp.org/HOWTO/Assembly-HOWTO/

Regards,
-------------------------------------------------------------
"Only two things are infinite, the universe and human stupidity;
and i am not sure about the former" 
					-Albert Einstein
-------------------------------------------------------------
Arun Vishwanathan 
 


> -----Original Message-----
> From: binutils-owner@sources.redhat.com [mailto:binutils-
> owner@sources.redhat.com] On Behalf Of Bill Cunningham
> Sent: Wednesday, May 05, 2004 7:19 PM
> To: binutils@sources.redhat.com
> Subject: GAS
> 
>     I all I just signed up to this list and I've been wanting to learn
> assembly as well as C. The GAS assembly format is what I want to learn
and
> Intel's IA-32 assembly language format. Is assembly language syntax
> different, or is there a universal assembly format I should learn?
> 
>     Bill
> 

GAS

[Bill Cunningham]

    I all I just signed up to this list and I've been wanting to learn
assembly as well as C. The GAS assembly format is what I want to learn and
Intel's IA-32 assembly language format. Is assembly language syntax
different, or is there a universal assembly format I should learn?

    Bill