------------------------------------------------------
Dell, Inc.
SPEC CPU2000 Flags Description
Intel C/C++/Visual FORTRAN Compilers Version 9.1
Windows and Linux
------------------------------------------------------
------------------------------------------------------
General Options (C/C++/FORTRAN)
------------------------------------------------------
-arch:<keyword>
Determines the version of the architecture for which the
compiler generates instructions keyword Is the processor type.
Possible values are:
SSE: Optimizes for Intel Pentium 4 processors with
Streaming SIMD Extensions (SSE).
SSE2: Optimizes for Intel Pentium 4 processors with
Streaming SIMD Extensions 2 (SSE2).
-fast This option maximizes speed across the entire program by
including the following Options (IA32, EM64T)
-O3 -Qipo -Qprec-div- -QxP
-O{1|2|3} Optimization-level options:
1: optimize for speed, but disable some optimizations which
increase code size for a small speed benefit. Includes
inline expansion for intrinsic functions, global
optimizations, string pooling optimizations.
2: This is the default level of optimization.
Optimizes for speed. The -O2 option includes O1
optimizations and in addition enables inlining of
intrinsics and more speed optimizations.
3: Builds on -01 and -02 optimizations by enabling high-level
optimization. This level does not guarantee higher performance
unless loop and memory access transformation take place. In
conjunction with -QaxK/-QxK and QaxW/QxW, this switch causes
the compiler to perform more aggressive data dependency
analysis than for -O2. This may result in longer compilation
times.
-Oa[-] Assume [do not assume] no aliasing in program.
-Qansi_alias
Enable/disable use of ANSI aliasing rules in
optimizations; user asserts that the program adheres to
these rules. The default for C++ is -Qansi_alias-
which is that aliasing rules are not assumed. The default for
the Fortran compiler is -Qansi_alias. For C++, the -Qansi_alias
flag will enable optimizations that would otherwise be
prevented by potential aliasing.
-Qax<processor> (Windows)
-ax<processor> (Linux)
Directs the compiler to generate processor-specific code if there is a
performance benefit, while also generating generic IA-32 code.
<processor> is the processor for which you want to target your program.
Possible values are the same as described for -Qx[].
-Qipo Multi-file ip optimizations that includes:
- inline function expansion
- interprocedural constant propogation
- dead code elimination
- propagation of function characteristics
- passing arguments in registers
- loop-invariant code motion
-Qoption,<string>,options
Passes options to a specified tool.
<string> Is the name of the tool:
c - Indicates the Intel C++ compiler.
cpp (or fpp) - Indicates the Intel C++ preprocessor.
cxxinc - Indicates C++ header files.
cinc - Indicates C header files.
asm - Indicates the assembler.
link - Indicates the linker.
prof - Indicates the profiler.
masm - Indicates the Microsoft assembler. (Windows only)
Linux and Mac OS:
as - Indicates the assembler.
gas - Indicates the GNU assembler.
ld - Indicates the loader.
gld - Indicates the GNU loader.
lib - Indicates an additional library.
crt - Indicates the crt%.o files linked into executables
to contain the place to start execution.
<options> One or more comma-separated valid options for the designated tool.
-Qoption can be used with the -Qipo flag to refine IPO.
The valid options that can be used for this purpose are:
-ip_args_in_regs=0 Disables the passing of arguments in registers.
-ip_ninl_max_stats=n
Sets the valid max number of intermediate language
statements for a function that is expanded in line.
The number n is a positive integer. The number of
intermediate language statements usually exceeds the
actual number of source language statements. The default
value for n is 230. The compiler uses a larger limit for
user inline functions.
-ip_ninl_min_stats=n
Sets the valid min number of intermedidate language
statements for a function that is expanded in line.
The number n is a positive integer. The default values for
ip_ninl_min_stats are: IA-32 compiler: ip_ninl_min_stats = 7
-ip_ninl_max_total_stats=n Sets the maximum increase in size of
a function, measured in intermediate language statements,
due to inlining. n is a positive integer whose default
value is 2000.
-Qprec-div[-]
Improves precision of floating point divides.
-Qprof_gen (Windows)
-prof_gen (Linux)
Instrument program for profiling for the first phase of
two-phase profile guided optimization.
-Qprof_use (Windows)
-prof_use (Linux)
Instructs the compiler to produce a profile-optimized
executable and merges available dynamic information (.dyn)
files into a pgopti.dpi file. If you perform multiple
executions of the instrumented program, -Qprof_use merges
the dynamic information files again and overwrites the
previous pgopti.dpi file. Without any other options,
the current directory is searched for .dyn files.
-Qrcd The Intel compiler uses the -Qrcd option to improve the
performance of code that requires floating-point-to-integer
conversions.
The system default floating point rounding mode is
round-to-nearest. This means that values are rounded during
floating point calculations. However, the C language requires
floating point values to be truncated when a conversion to an
integer is involved. To do this, the compiler must change the
rounding mode to truncation before each floating
point-to-integer conversion and change it back afterwards.
The -Qrcd option disables the change to truncation of the
rounding mode for all floating point calculations, including
floating point-to-integer conversions. Turning on this option
can improve performance, but floating point conversions to
integer may not conform to C semantics.
-Qunroll[n]
Specifies the maximum number of times to unroll a loop. Omit n
to let the compiler decide whether to perform unrolling or not.
Use n = 0 to disable unroller.
If n is not specified, the compiler automatically chooses the maximum
number of times to unroll a loop.
-Qx<codes>
Generate specialized code for processor specified by <codes>
while also generating generic code.
<codes> includes one or more of the following:
K: Intel Pentium III or copmatible Intel processors (SSE)
W: Intel Pentium 4 and compatible Intel processors (SSE2)
N: Intel Pentium 4 and compatible Intel processors (SSE2)
Enables new optimizations in addition to Intel processor-specific
optimizations.
P: Intel Pentium 4 processors and compatible Intel processors with SSE3
B Intel Pentium M and compatible Intel processors (SSE2)
SSE is Streamng SIMD Extensions
Additional Notes on <codes> N and P:
------------------------------------
The N and P options target your program to run on Intel Pentium 4
and compatible Intel processors. The resulting code might
contain unconditional use of features that are not supported
on other processors. Programs, where the function main() is
compiled with this option, will detect non compatible processors
and generate an error message during execution.
These options also enable new optimizations in addition to Intel
processor-specific optimizations including advanced data layout and
code restructuring optimizations to improve memory accesses for
Intel processors.
------------------------------------------------------
Flags Specific to C/C++
------------------------------------------------------
-Oi[-] Enables inline expansion of intrinsic functions.
-Oi- disables inlining of all intrinsic functions.
-Qcxx_features
Enables standard C++ features without disabling Microsoft features and within
the bounds of what is provided in the Microsoft headers and libraries.
This option also enables -GR and -GX.
-GR Enables C++ Runtime Type Information (RTTI).
-GX Enables the full C++ Exception Handling unwind semantics.
------------------------------------------------------
Flags Specific to FORTRAN
------------------------------------------------------
-Qauto Causes all variables to be allocated on the stack, rather than
in local static storage.
-Qscalar_rep[-]
Enables (DEFAULT) [disables] scalar replacement performed
during loop transformations.
------------------------------------------------------
General Options and Libraries
------------------------------------------------------
The starting tokens "/" and "-" are both equivalent for flags passed to the
compiler. For example, -QxW and /QxW are identical switches.
+FDO PASS1=-Qprof_gen PASS2=-Qprof_use
Using feedback-directed optimization, a profile is generated
on the first pass of compilation and used on the second pass.
shlW32M.lib
MicroQuill SmartHeap Library available from www.microquill.com
------------------------------------------------------
Benchmark-Specific Portability Options
------------------------------------------------------
-DSPEC__CPU2000_LP64
Compile using LP64 programming model.
176.gcc:
-Dalloca=_alloca So as to use the built-in optimized alloca.
/F10000000 176.gcc uses alloca and this option tells the
linker to pre-allocate 10MB of stack. The default
amount of stack allocated is not enough and 176.gcc
crashes with a run-time error
178.galgel:
-FI (Linux)
/FI Fixed-format F90 source code.
/F32000000 Same as with 176.gcc, pre-allocates a 32MB stack
186.crafty:
-DNT_i386 Specifies that it is a Windows NT Intel processor-based
system which makes the compiler use "_int64"
as the 64-bit variable that 186.crafty needs.
253.perlbmk:
-DSPEC_CPU2000_NTOS
This enables the code changes for porting to
Windows get included.
-DPERLDLL On Windows, we need a perl.exe instead of a
perl.exe and perl.dll. This pre-defines ensures
that the changes necessary to get a single, UNIX-style
executible without getting the indirect calls that
can cause a 10% performance degradation. This allows
the Windows-based executible to be as close as possible
to the Unix-based one.
/MT Use the static multi-threaded library else it
will not compile.
254.gap:
-DSYS_HAS_CALLOC_PROTO
-DSYS_HAS_MALLOC_PROTO
These two pre-defines tell of the existence of
malloc and calloc prototypes.
------------------------------------------------------
BIOS Settings
------------------------------------------------------
"Snoop Filter"
The Snoop Filter (SF) offers significant performance enhancements on several
workstation benchmarks by eliminating traffic on the snooped frontside bus of the
processor being snooped. By removing snoops from the snooped bus, the full
bandwidth is available for other transactions. Supporting concurrent snoops effectively
reduces performance degradation attributable to multiple snoop stalls.