[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

15. GCL Specific

Function: SYSTEM (string)
Package:LISP

GCL specific: Executes a Shell command as if STRING is an input to the Shell. Not all versions of GCL support this function.

Variable: *IGNORE-MAXIMUM-PAGES*
Package:LISP GCL specific: Tells the GCL memory manager whether (non-NIL) or not (NIL) it should expand memory whenever the maximum allocatable pages have been used up.

Function: MACHINE-VERSION ()
Package:LISP

Returns a string that identifies the machine version of the machine on which GCL is currently running.

Function: BY ()
Package:LISP

GCL specific: Exits from GCL.

Macro: DEFCFUN
Package:LISP

Syntax:
 
(defcfun header n {element}*)

GCL specific: Defines a C-language function which calls Lisp functions and/or handles Lisp objects. HEADER gives the header of the C function as a string. Non-negative-integer is the number of the main stack entries used by the C function, primarily for protecting Lisp objects from being garbage-collected. Each ELEMENT may give a C code fragment as a string, or it may be a list ((symbol {arg}*) {place}*) which, when executed, calls the Lisp function named by SYMBOL with the specified arguments and saves the value(s) to the specified places. The DEFCFUN form has the above meanings only after compiled; The GCL interpreter simply ignores this form.

An example which defines a C function list2 of two arguments, but which calls the 'lisp' function CONS by name, and refers to the constant 'NIL. Note to be loaded by load the function should be static.

(defCfun "static object list2(x,y) object x,y;" 0 "object z;" ('NIL z) ((CONS y z) z) ((CONS x z) z) "return(z);" )

In lisp the operations in the body would be (setq z 'nil) (setq z (cons y z)) (setq z (cons x z))

Syntax:
 
        (defCfun header non-negative-integer
                { string
                  | ( function-symbol { value }* )
                  | (( function-symbol  { value }* ) { place }* ) })


value:
place:
         { C-expr | ( C-type C-expr ) }

C-function-name:
C-expr:
         { string | symbol }
 
C-type:
         { object | int | char | float | double }

Macro: CLINES
Package:LISP

Syntax:
 
(clines {string}*)

GCL specific: The GCL compiler embeds STRINGs into the intermediate C language code. The interpreter ignores this form.

Function: ALLOCATE (type number &optional (really-allocate nil))
Package:LISP

GCL specific: Sets the maximum number of pages for the type class of the GCL implementation type TYPE to NUMBER. If REALLY-ALLOCATE is given a non-NIL value, then the specified number of pages will be allocated immediately.

Function: GBC (x)
Package:LISP

GCL specific: Invokes the garbage collector (GC) with the collection level specified by X. NIL as the argument causes GC to collect cells only. T as the argument causes GC to collect everything.

Function: SAVE (pathname)
Package:LISP

GCL specific: Saves the current GCL core image into a program file specified by PATHNAME. This function depends on the version of GCL. The function si::save-system is to be preferred in almost all circumstances. Unlike save, it makes the relocatable section permanent, and causes no future gc of currently loaded .o files.

Function: HELP* (string &optional (package 'lisp))
Package:LISP

GCL specific: Prints the documentation associated with those symbols in the specified package whose print names contain STRING as substring. STRING may be a symbol, in which case the print-name of that symbol is used. If PACKAGE is NIL, then all packages are searched.

Macro: DEFLA
Package:LISP

Syntax:
 
(defla name lambda-list {decl | doc}* {form}*)

GCL specific: Used to DEFine Lisp Alternative. For the interpreter, DEFLA is equivalent to DEFUN, but the compiler ignores this form.

Function: PROCLAMATION (decl-spec)
Package:LISP

GCL specific: Returns T if the specified declaration is globally in effect; NIL otherwise. See the doc of DECLARE for possible DECL-SPECs.

Macro: DEFENTRY
Package:LISP

Syntax:
 
(defentry name arg-types c-function)

GCL specific: The compiler defines a Lisp function whose body consists of a calling sequence to the C language function specified by C-FUNCTION. The interpreter ignores this form. The ARG-TYPES specifies the C types of the arguments which C-FUNCTION requires. The list of allowed types is (object char int float double string). Code will be produced to coerce from a lisp object to the appropriate type before passing the argument to the C-FUNCTION. The c-function should be of the form (c-result-type c-fname) where c-result-type is a member of (void object char int float double string). c-fname may be a symbol (in which case it will be downcased) or a string. If c-function is not a list, then (object c-function) is assumed. In order for C code to be loaded in by load you should declare any variables and functions to be static. If you will link them in at build time, of course you are allowed to define new externals.

 
  Sample usage:
--File begin-----
;; JOE takes X a lisp string and Y a fixnum and returns a character.
(clines "#include \"foo.ch\"")
(defentry joe (string int) (char "our_c_fun"))
---File end------
---File foo.ch---
/* C function for extracting the i'th element of a string */
static char our_c_fun(p,i)
char *p;
int i;
   {
	return p[i];
   }
-----File end---

One must be careful of storage allocation issues when passing a string. If the C code invokes storage allocation (either by calling malloc or make_cons etc), then there is a possibility of a garbage collection, so that if the string passed was not constructed with :static t when its array was constructed, then it could move. If the C function may allocate storage, then you should pass a copy:
 
(defun safe-c-string (x)
  (let* ((n (length x))
         (a (make-array (+ n 1) :element-type 'string-char
           :static t :fill-pointer n)))
    (si::copy-array-portion x y 0 0 n)
    (setf (aref a n) (code-char 0)))
    a)

Function: COPY-ARRAY-PORTION (x,y,i1,i2,n1)
Package:SI Copy elements from X to Y starting at X[i1] to Y[i2] and doing N1 elements if N1 is supplied otherwise, doing the length of X - I1 elements. If the types of the arrays are not the same, this has implementation dependent results.

Function: BYE ( &optional (exit-status 0))
Package:LISP

GCL specific: Exits from GCL with exit-status.

Function: USE-FAST-LINKS (turn-on)
Package:LISP

GCL specific: If TURN-ON is not nil, the fast link mechanism is enabled, so that ordinary function calls will not appear in the invocation stack, and calls will be much faster. This is the default. If you anticipate needing to see a stack trace in the debugger, then you should turn this off.

15.1 Bignums  


[ < ] [ > ]   [ << ] [ Up ] [ >> ]         [Top] [Contents] [Index] [ ? ]

15.1 Bignums

A directory mp was added to hold the new multi precision arithmetic code. The layout and a fair amount of code in the mp directory is an enhanced version of gpari version 34. The gpari c code was rewritten to be more efficient, and gcc assembler macros were added to allow inlining of operations not possible to do in C. On a 68K machine, this allows the C version to be as efficient as the very carefully written assembler in the gpari distribution. For the main machines, an assembler file (produced by gcc) based on this new method, is included. This is for sites which do not have gcc, or do not wish to compile the whole system with gcc.

Bignum arithmetic is much faster now. Many changes were made to cmpnew also, to add 'integer' as a new type. It differs from variables of other types, in that storage is associated to each such variable, and assignments mean copying the storage. This allows a function which does a good deal of bignum arithmetic, to do very little consing in the heap. An example is the computation of PI-INV in scratchpad, which calculates the inverse of pi to a prescribed number of bits accuracy. That function is now about 20 times faster, and no longer causes garbage collection. In versions of GCL where HAVE_ALLOCA is defined, the temporary storage growth is on the C stack, although this often not so critical (for example it makes virtually no difference in the PI-INV example, since in spite of the many operations, only one storage allocation takes place. Below is the actual code for PI-INV

On a sun3/280 (cli.com)

Here is the comparison of lucid and gcl before and after on that pi-inv. Times are in seconds with multiples of the gcl/akcl time in parentheses.

On a sun3/280 (cli.com)

 
pi-inv   akcl-566  franz        lucid         old kcl/akcl
----------------------------------------
10000      3.3     9.2(2.8 X)  15.3 (4.6X)    92.7   (29.5 X)
20000      12.7    31.0(2.4 X) 62.2 (4.9X)    580.0  (45.5 X)


(defun pi-inv (bits &aux (m 0))
  (declare (integer bits m))
  (let* ((n (+ bits (integer-length bits) 11))
         (tt (truncate (ash 1 n) 882))
         (d (* 4 882 882))
         (s 0))
    (declare (integer s d tt n))
    (do ((i 2 (+ i 2))
         (j 1123 (+ j 21460)))
        ((zerop tt) (cons s (- (+ n 2))))
      (declare (integer i j))
        (setq s (+ s (* j tt))
              m (- (* (- i 1) (- (* 2 i) 1) (- (* 2 i) 3)))
              tt (truncate (* m tt) (* d (the integer (expt i 3))))))))


[ << ] [ >> ]           [Top] [Contents] [Index] [ ? ]

This document was generated by Camm Maguire on March, 4 2002 using texi2html