<html lang="en">
<head>
<title>DEC Alpha Options - Using the GNU Compiler Collection (GCC)</title>
<meta http-equiv="Content-Type" content="text/html">
<meta name="description" content="Using the GNU Compiler Collection (GCC)">
<meta name="generator" content="makeinfo 4.13">
<link title="Top" rel="start" href="index.html#Top">
<link rel="up" href="Submodel-Options.html#Submodel-Options" title="Submodel Options">
<link rel="prev" href="Darwin-Options.html#Darwin-Options" title="Darwin Options">
<link rel="next" href="FR30-Options.html#FR30-Options" title="FR30 Options">
<link href="http://www.gnu.org/software/texinfo/" rel="generator-home" title="Texinfo Homepage">
<!--
Copyright (C) 1988-2015 Free Software Foundation, Inc.
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License, Version 1.3 or
any later version published by the Free Software Foundation; with the
Invariant Sections being ``Funding Free Software'', the Front-Cover
Texts being (a) (see below), and with the Back-Cover Texts being (b)
(see below). A copy of the license is included in the section entitled
``GNU Free Documentation License''.
(a) The FSF's Front-Cover Text is:
A GNU Manual
(b) The FSF's Back-Cover Text is:
You have freedom to copy and modify this GNU Manual, like GNU
software. Copies published by the Free Software Foundation raise
funds for GNU development.-->
<meta http-equiv="Content-Style-Type" content="text/css">
<style type="text/css"><!--
pre.display { font-family:inherit }
pre.format { font-family:inherit }
pre.smalldisplay { font-family:inherit; font-size:smaller }
pre.smallformat { font-family:inherit; font-size:smaller }
pre.smallexample { font-size:smaller }
pre.smalllisp { font-size:smaller }
span.sc { font-variant:small-caps }
span.roman { font-family:serif; font-weight:normal; }
span.sansserif { font-family:sans-serif; font-weight:normal; }
--></style>
</head>
<body>
<div class="node">
<a name="DEC-Alpha-Options"></a>
<p>
Next: <a rel="next" accesskey="n" href="FR30-Options.html#FR30-Options">FR30 Options</a>,
Previous: <a rel="previous" accesskey="p" href="Darwin-Options.html#Darwin-Options">Darwin Options</a>,
Up: <a rel="up" accesskey="u" href="Submodel-Options.html#Submodel-Options">Submodel Options</a>
<hr>
</div>
<h4 class="subsection">3.17.11 DEC Alpha Options</h4>
<p>These ‘<samp><span class="samp">-m</span></samp>’ options are defined for the DEC Alpha implementations:
<dl>
<dt><code>-mno-soft-float</code><dt><code>-msoft-float</code><dd><a name="index-mno_002dsoft_002dfloat-1603"></a><a name="index-msoft_002dfloat-1604"></a>Use (do not use) the hardware floating-point instructions for
floating-point operations. When <samp><span class="option">-msoft-float</span></samp> is specified,
functions in <samp><span class="file">libgcc.a</span></samp> are used to perform floating-point
operations. Unless they are replaced by routines that emulate the
floating-point operations, or compiled in such a way as to call such
emulations routines, these routines issue floating-point
operations. If you are compiling for an Alpha without floating-point
operations, you must ensure that the library is built so as not to call
them.
<p>Note that Alpha implementations without floating-point operations are
required to have floating-point registers.
<br><dt><code>-mfp-reg</code><dt><code>-mno-fp-regs</code><dd><a name="index-mfp_002dreg-1605"></a><a name="index-mno_002dfp_002dregs-1606"></a>Generate code that uses (does not use) the floating-point register set.
<samp><span class="option">-mno-fp-regs</span></samp> implies <samp><span class="option">-msoft-float</span></samp>. If the floating-point
register set is not used, floating-point operands are passed in integer
registers as if they were integers and floating-point results are passed
in <code>$0</code> instead of <code>$f0</code>. This is a non-standard calling sequence,
so any function with a floating-point argument or return value called by code
compiled with <samp><span class="option">-mno-fp-regs</span></samp> must also be compiled with that
option.
<p>A typical use of this option is building a kernel that does not use,
and hence need not save and restore, any floating-point registers.
<br><dt><code>-mieee</code><dd><a name="index-mieee-1607"></a>The Alpha architecture implements floating-point hardware optimized for
maximum performance. It is mostly compliant with the IEEE floating-point
standard. However, for full compliance, software assistance is
required. This option generates code fully IEEE-compliant code
<em>except</em> that the <var>inexact-flag</var> is not maintained (see below).
If this option is turned on, the preprocessor macro <code>_IEEE_FP</code> is
defined during compilation. The resulting code is less efficient but is
able to correctly support denormalized numbers and exceptional IEEE
values such as not-a-number and plus/minus infinity. Other Alpha
compilers call this option <samp><span class="option">-ieee_with_no_inexact</span></samp>.
<br><dt><code>-mieee-with-inexact</code><dd><a name="index-mieee_002dwith_002dinexact-1608"></a>This is like <samp><span class="option">-mieee</span></samp> except the generated code also maintains
the IEEE <var>inexact-flag</var>. Turning on this option causes the
generated code to implement fully-compliant IEEE math. In addition to
<code>_IEEE_FP</code>, <code>_IEEE_FP_EXACT</code> is defined as a preprocessor
macro. On some Alpha implementations the resulting code may execute
significantly slower than the code generated by default. Since there is
very little code that depends on the <var>inexact-flag</var>, you should
normally not specify this option. Other Alpha compilers call this
option <samp><span class="option">-ieee_with_inexact</span></samp>.
<br><dt><code>-mfp-trap-mode=</code><var>trap-mode</var><dd><a name="index-mfp_002dtrap_002dmode-1609"></a>This option controls what floating-point related traps are enabled.
Other Alpha compilers call this option <samp><span class="option">-fptm </span><var>trap-mode</var></samp>.
The trap mode can be set to one of four values:
<dl>
<dt>‘<samp><span class="samp">n</span></samp>’<dd>This is the default (normal) setting. The only traps that are enabled
are the ones that cannot be disabled in software (e.g., division by zero
trap).
<br><dt>‘<samp><span class="samp">u</span></samp>’<dd>In addition to the traps enabled by ‘<samp><span class="samp">n</span></samp>’, underflow traps are enabled
as well.
<br><dt>‘<samp><span class="samp">su</span></samp>’<dd>Like ‘<samp><span class="samp">u</span></samp>’, but the instructions are marked to be safe for software
completion (see Alpha architecture manual for details).
<br><dt>‘<samp><span class="samp">sui</span></samp>’<dd>Like ‘<samp><span class="samp">su</span></samp>’, but inexact traps are enabled as well.
</dl>
<br><dt><code>-mfp-rounding-mode=</code><var>rounding-mode</var><dd><a name="index-mfp_002drounding_002dmode-1610"></a>Selects the IEEE rounding mode. Other Alpha compilers call this option
<samp><span class="option">-fprm </span><var>rounding-mode</var></samp>. The <var>rounding-mode</var> can be one
of:
<dl>
<dt>‘<samp><span class="samp">n</span></samp>’<dd>Normal IEEE rounding mode. Floating-point numbers are rounded towards
the nearest machine number or towards the even machine number in case
of a tie.
<br><dt>‘<samp><span class="samp">m</span></samp>’<dd>Round towards minus infinity.
<br><dt>‘<samp><span class="samp">c</span></samp>’<dd>Chopped rounding mode. Floating-point numbers are rounded towards zero.
<br><dt>‘<samp><span class="samp">d</span></samp>’<dd>Dynamic rounding mode. A field in the floating-point control register
(<var>fpcr</var>, see Alpha architecture reference manual) controls the
rounding mode in effect. The C library initializes this register for
rounding towards plus infinity. Thus, unless your program modifies the
<var>fpcr</var>, ‘<samp><span class="samp">d</span></samp>’ corresponds to round towards plus infinity.
</dl>
<br><dt><code>-mtrap-precision=</code><var>trap-precision</var><dd><a name="index-mtrap_002dprecision-1611"></a>In the Alpha architecture, floating-point traps are imprecise. This
means without software assistance it is impossible to recover from a
floating trap and program execution normally needs to be terminated.
GCC can generate code that can assist operating system trap handlers
in determining the exact location that caused a floating-point trap.
Depending on the requirements of an application, different levels of
precisions can be selected:
<dl>
<dt>‘<samp><span class="samp">p</span></samp>’<dd>Program precision. This option is the default and means a trap handler
can only identify which program caused a floating-point exception.
<br><dt>‘<samp><span class="samp">f</span></samp>’<dd>Function precision. The trap handler can determine the function that
caused a floating-point exception.
<br><dt>‘<samp><span class="samp">i</span></samp>’<dd>Instruction precision. The trap handler can determine the exact
instruction that caused a floating-point exception.
</dl>
<p>Other Alpha compilers provide the equivalent options called
<samp><span class="option">-scope_safe</span></samp> and <samp><span class="option">-resumption_safe</span></samp>.
<br><dt><code>-mieee-conformant</code><dd><a name="index-mieee_002dconformant-1612"></a>This option marks the generated code as IEEE conformant. You must not
use this option unless you also specify <samp><span class="option">-mtrap-precision=i</span></samp> and either
<samp><span class="option">-mfp-trap-mode=su</span></samp> or <samp><span class="option">-mfp-trap-mode=sui</span></samp>. Its only effect
is to emit the line ‘<samp><span class="samp">.eflag 48</span></samp>’ in the function prologue of the
generated assembly file.
<br><dt><code>-mbuild-constants</code><dd><a name="index-mbuild_002dconstants-1613"></a>Normally GCC examines a 32- or 64-bit integer constant to
see if it can construct it from smaller constants in two or three
instructions. If it cannot, it outputs the constant as a literal and
generates code to load it from the data segment at run time.
<p>Use this option to require GCC to construct <em>all</em> integer constants
using code, even if it takes more instructions (the maximum is six).
<p>You typically use this option to build a shared library dynamic
loader. Itself a shared library, it must relocate itself in memory
before it can find the variables and constants in its own data segment.
<br><dt><code>-mbwx</code><dt><code>-mno-bwx</code><dt><code>-mcix</code><dt><code>-mno-cix</code><dt><code>-mfix</code><dt><code>-mno-fix</code><dt><code>-mmax</code><dt><code>-mno-max</code><dd><a name="index-mbwx-1614"></a><a name="index-mno_002dbwx-1615"></a><a name="index-mcix-1616"></a><a name="index-mno_002dcix-1617"></a><a name="index-mfix-1618"></a><a name="index-mno_002dfix-1619"></a><a name="index-mmax-1620"></a><a name="index-mno_002dmax-1621"></a>Indicate whether GCC should generate code to use the optional BWX,
CIX, FIX and MAX instruction sets. The default is to use the instruction
sets supported by the CPU type specified via <samp><span class="option">-mcpu=</span></samp> option or that
of the CPU on which GCC was built if none is specified.
<br><dt><code>-mfloat-vax</code><dt><code>-mfloat-ieee</code><dd><a name="index-mfloat_002dvax-1622"></a><a name="index-mfloat_002dieee-1623"></a>Generate code that uses (does not use) VAX F and G floating-point
arithmetic instead of IEEE single and double precision.
<br><dt><code>-mexplicit-relocs</code><dt><code>-mno-explicit-relocs</code><dd><a name="index-mexplicit_002drelocs-1624"></a><a name="index-mno_002dexplicit_002drelocs-1625"></a>Older Alpha assemblers provided no way to generate symbol relocations
except via assembler macros. Use of these macros does not allow
optimal instruction scheduling. GNU binutils as of version 2.12
supports a new syntax that allows the compiler to explicitly mark
which relocations should apply to which instructions. This option
is mostly useful for debugging, as GCC detects the capabilities of
the assembler when it is built and sets the default accordingly.
<br><dt><code>-msmall-data</code><dt><code>-mlarge-data</code><dd><a name="index-msmall_002ddata-1626"></a><a name="index-mlarge_002ddata-1627"></a>When <samp><span class="option">-mexplicit-relocs</span></samp> is in effect, static data is
accessed via <dfn>gp-relative</dfn> relocations. When <samp><span class="option">-msmall-data</span></samp>
is used, objects 8 bytes long or smaller are placed in a <dfn>small data area</dfn>
(the <code>.sdata</code> and <code>.sbss</code> sections) and are accessed via
16-bit relocations off of the <code>$gp</code> register. This limits the
size of the small data area to 64KB, but allows the variables to be
directly accessed via a single instruction.
<p>The default is <samp><span class="option">-mlarge-data</span></samp>. With this option the data area
is limited to just below 2GB. Programs that require more than 2GB of
data must use <code>malloc</code> or <code>mmap</code> to allocate the data in the
heap instead of in the program's data segment.
<p>When generating code for shared libraries, <samp><span class="option">-fpic</span></samp> implies
<samp><span class="option">-msmall-data</span></samp> and <samp><span class="option">-fPIC</span></samp> implies <samp><span class="option">-mlarge-data</span></samp>.
<br><dt><code>-msmall-text</code><dt><code>-mlarge-text</code><dd><a name="index-msmall_002dtext-1628"></a><a name="index-mlarge_002dtext-1629"></a>When <samp><span class="option">-msmall-text</span></samp> is used, the compiler assumes that the
code of the entire program (or shared library) fits in 4MB, and is
thus reachable with a branch instruction. When <samp><span class="option">-msmall-data</span></samp>
is used, the compiler can assume that all local symbols share the
same <code>$gp</code> value, and thus reduce the number of instructions
required for a function call from 4 to 1.
<p>The default is <samp><span class="option">-mlarge-text</span></samp>.
<br><dt><code>-mcpu=</code><var>cpu_type</var><dd><a name="index-mcpu-1630"></a>Set the instruction set and instruction scheduling parameters for
machine type <var>cpu_type</var>. You can specify either the ‘<samp><span class="samp">EV</span></samp>’
style name or the corresponding chip number. GCC supports scheduling
parameters for the EV4, EV5 and EV6 family of processors and
chooses the default values for the instruction set from the processor
you specify. If you do not specify a processor type, GCC defaults
to the processor on which the compiler was built.
<p>Supported values for <var>cpu_type</var> are
<dl>
<dt>‘<samp><span class="samp">ev4</span></samp>’<dt>‘<samp><span class="samp">ev45</span></samp>’<dt>‘<samp><span class="samp">21064</span></samp>’<dd>Schedules as an EV4 and has no instruction set extensions.
<br><dt>‘<samp><span class="samp">ev5</span></samp>’<dt>‘<samp><span class="samp">21164</span></samp>’<dd>Schedules as an EV5 and has no instruction set extensions.
<br><dt>‘<samp><span class="samp">ev56</span></samp>’<dt>‘<samp><span class="samp">21164a</span></samp>’<dd>Schedules as an EV5 and supports the BWX extension.
<br><dt>‘<samp><span class="samp">pca56</span></samp>’<dt>‘<samp><span class="samp">21164pc</span></samp>’<dt>‘<samp><span class="samp">21164PC</span></samp>’<dd>Schedules as an EV5 and supports the BWX and MAX extensions.
<br><dt>‘<samp><span class="samp">ev6</span></samp>’<dt>‘<samp><span class="samp">21264</span></samp>’<dd>Schedules as an EV6 and supports the BWX, FIX, and MAX extensions.
<br><dt>‘<samp><span class="samp">ev67</span></samp>’<dt>‘<samp><span class="samp">21264a</span></samp>’<dd>Schedules as an EV6 and supports the BWX, CIX, FIX, and MAX extensions.
</dl>
<p>Native toolchains also support the value ‘<samp><span class="samp">native</span></samp>’,
which selects the best architecture option for the host processor.
<samp><span class="option">-mcpu=native</span></samp> has no effect if GCC does not recognize
the processor.
<br><dt><code>-mtune=</code><var>cpu_type</var><dd><a name="index-mtune-1631"></a>Set only the instruction scheduling parameters for machine type
<var>cpu_type</var>. The instruction set is not changed.
<p>Native toolchains also support the value ‘<samp><span class="samp">native</span></samp>’,
which selects the best architecture option for the host processor.
<samp><span class="option">-mtune=native</span></samp> has no effect if GCC does not recognize
the processor.
<br><dt><code>-mmemory-latency=</code><var>time</var><dd><a name="index-mmemory_002dlatency-1632"></a>Sets the latency the scheduler should assume for typical memory
references as seen by the application. This number is highly
dependent on the memory access patterns used by the application
and the size of the external cache on the machine.
<p>Valid options for <var>time</var> are
<dl>
<dt>‘<samp><var>number</var></samp>’<dd>A decimal number representing clock cycles.
<br><dt>‘<samp><span class="samp">L1</span></samp>’<dt>‘<samp><span class="samp">L2</span></samp>’<dt>‘<samp><span class="samp">L3</span></samp>’<dt>‘<samp><span class="samp">main</span></samp>’<dd>The compiler contains estimates of the number of clock cycles for
“typical” EV4 & EV5 hardware for the Level 1, 2 & 3 caches
(also called Dcache, Scache, and Bcache), as well as to main memory.
Note that L3 is only valid for EV5.
</dl>
</dl>
</body></html>