{"id":59,"date":"2011-05-27T10:31:07","date_gmt":"2011-05-27T02:31:07","guid":{"rendered":"https:\/\/www.sulabs.net\/?p=59"},"modified":"2017-01-09T10:34:50","modified_gmt":"2017-01-09T02:34:50","slug":"intel-c-%e7%bc%96%e8%af%91%e5%99%a811-1-%e4%b8%ad%e6%96%87%e6%96%87%e6%a1%a3%e4%b8%8d%e5%ae%8c%e5%85%a8%e7%bf%bb%e8%af%91","status":"publish","type":"post","link":"https:\/\/www.sulabs.net\/?p=59","title":{"rendered":"Intel C++ \u7f16\u8bd1\u566811.1 \u4e2d\u6587\u6587\u6863(\u4e0d\u5b8c\u5168\u7ffb\u8bd1)"},"content":{"rendered":"

\u65e9\u671f\u5728\u5c1d\u8bd5icc \u7f16\u8bd1\u5668\u6d4b\u8bd5\u65f6\u987a\u4fbf\u5c06\u82f1\u6587\u6587\u6863\u505a\u4e86\u90e8\u5206\u4e2d\u6587\u7ffb\u8bd1\uff0c\u672f\u8bed\u592a\u5f3a\uff0c\u90e8\u5206\u7ffb\u8bd1\u4e5f\u4e0d\u80fd\u4fdd\u969c\u51c6\u786e\u6027\uff0c\u5c06\u5c31\u770b\u770b\u5427\u3002<\/p>\n

Intel(R) C++ \u7f16\u8bd1\u566811.1 \u4e2d\u6587\u5e2e\u52a9\u6587\u6863
\n=================================<\/p>\n

\u7528\u6cd5: icc [\u9009\u9879] \u6587\u4ef61 [\u6587\u4ef62...]
\nicpc [\u9009\u9879] \u6587\u4ef61 [\u6587\u4ef62...]<\/p>\n

\u8fd9\u4e9b\u53c2\u6570\u53ef\u4ee5\u4e3a\u7a7a\u6216\u5176\u5b83\u7f16\u8bd1\u9009\u9879<\/p>\n

\u6587\u4ef6(N) \u53ef\u4ee5\u662f C\/C++ \u6e90\u4ee3\u7801\u6587\u4ef6(\u4f8b\u5982 .C .c .cc .cpp .cxx .c++ .i .ii)\uff0c
\n\u96c6\u5408 (.s .S), \u5bf9\u8c61 (.o), \u9759\u6001\u5e93 (.a), \u6216\u8005\u5176\u5b83\u8fde\u63a5\u6587\u4ef6<\/p>\n

The icpc command uses the same compiler options as the icc
\nicpc \u547d\u4ee4\u7684\u4f7f\u7528\u4e0eicc \u7f16\u8bd1\u9009\u9879\u76f8\u540c<\/p>\n

command.\u00a0 Invoking the compiler using icpc compiles .c and .i files
\n\u53ef\u4ee5\u8c03\u7528icpc \u7f16\u8bd1\u5668\u6765\u7f16\u8bd1 .c \u548c .i \u6587\u4ef6\u4e3ac++\u3002<\/p>\n

as C++.\u00a0 Invoking the compiler using icc compiles .c and .i files as
\n\u8c03\u7528icc \u7f16\u8bd1\u5668\u6765\u7f16\u8bd1.c \u548c .i \u6587\u4ef6\u4e3a C<\/p>\n

C.\u00a0 Using icpc always links in C++ libraries.\u00a0 Using icc only
\n\u4f7f\u7528icpc \u7f16\u8bd1\u5668\u5fc5\u987b\u8fde\u63a5C++ \u5e93\uff0c\u4f7f\u7528icc \u7f16\u8bd1\u5668\u53ea\u9700\u8981\u5728\u547d\u4ee4\u884c\u4e0b
\nlinks in C++ libraries if C++ source is provided on the command line.
\n\u8fde\u63a5 C++ \u5e93\u548c\u63d0\u4f9bC++\u6e90\u4ee3\u7801
\nCommonly used options may be placed in the icc .cfg file.
\n\u5e38\u7528\u7684\u9009\u9879\u53ef\u80fd\u4f1a\u5b58\u653e\u5728 icc.cfg \u6587\u4ef6\u91cc<\/p>\n

Some options listed are only available on a specific system
\n\u4e00\u4e9b\u5217\u51fa\u7684\u9009\u9879\u53ea\u9002\u7528\u4e8e\u7279\u5b9a\u7684\u7cfb\u7edf
\ni32\u00a0\u00a0\u00a0 indicates the feature is available on systems based on IA-32
\narchitecture
\n\u8bf4\u660e\u8be5\u529f\u80fd\u53ea\u80fd\u7528\u572832\u4f4d\uff08IA-32\uff09\u6784\u67b6\u7684\u7cfb\u7edf\u4e0a\u9762<\/p>\n

i64em\u00a0 indicates the feature is available on systems using Intel(R) 64
\narchitecture
\n\u8bf4\u660e\u8be5\u529f\u80fd\u53ea\u80fd\u7528\u5728\u82f1\u7279\u5c1464\u4f4d\u6784\u67b6\u7684\u7cfb\u7edf\u4e0a\u9762<\/p>\n

\u7f16\u8bd1\u53c2\u6570\u5217\u8868
\n--------------------<\/p>\n

\u4f18\u5316\u53c2\u6570
\n------------<\/p>\n

-O1\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 optimize for maximum speed, but disable some optimizations which
\nincrease code size for a small speed benefit
\n\u5bf9\u901f\u5ea6\u8fdb\u884c\u6700\u5927\u4f18\u5316\uff0c\u4f46\u4f1a\u7981\u7528\u4e00\u4e9b\u4ee3\u7801\u4f53\u79ef\u4f18\u5316\u6765\u6362\u53d6\u901f\u5ea6\u63d0\u9ad8<\/p>\n

-O2\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 optimize for maximum speed (DEFAULT)
\n\u5bf9\u901f\u5ea6\u8fdb\u884c\u6700\u5927\u4f18\u5316(\u9ed8\u8ba4)<\/p>\n

-O3\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 optimize for maximum speed and enable more aggressive optimizations
\nthat may not improve performance on some programs
\n\u5bf9\u901f\u5ea6\u8fdb\u884c\u6700\u5927\u4f18\u5316\u5e76\u5f00\u542f\u9ad8\u7ea7\u4f18\u5316\uff0c\u5bf9\u90e8\u5206\u7a0b\u5e8f\u6027\u80fd\u53ef\u80fd\u65e0\u6cd5\u63d0\u9ad8<\/p>\n

-O\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 same as -O2
\n\u548c -O2 \u76f8\u540c<\/p>\n

-Os\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 enable speed optimizations, but disable some optimizations which
\nincrease code size for small speed benefit
\n\u5f00\u542f\u901f\u5ea6\u4f18\u5316\uff0c\u4f46\u4f1a\u7981\u7528\u4e00\u4e9b\u4ee3\u7801\u4f53\u79ef\u4f18\u5316\u6765\u6362\u53d6\u901f\u5ea6\u63d0\u9ad8<\/p>\n

-O0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 disable optimizations
\n\u7981\u7528\u4f18\u5316<\/p>\n

-fast\u00a0\u00a0\u00a0\u00a0 enable -xHOST -O3 -ipo -no-prec-div -static
\noptions set by -fast cannot be overridden with the exception of
\n-xHOST, list options separately to change behavior
\n\u7528-fast \u53ef\u540c\u65f6\u5f00\u542f -xHOST -O3 -ipo -no-prec-div -static \u9009\u9879\uff0c\u4f46\u4e0d\u80fd
\n\u548c-xHOST \u5171\u7528\uff0c\u5426\u5219\u5217\u51fa\u7684\u9009\u9879\u5c06\u72ec\u7acb\u4f7f\u7528
\n-fno-alias
\nassume no aliasing in program
\n\u4e0d\u91c7\u7528\u4ee3\u7801\u6df7\u6dc6(\u7528\u4e8e\u4fdd\u62a4\u4ee3\u7801\uff0c\u9632\u53cd\u7f16\u8bd1)
\n-fno-fnalias
\nassume no aliasing within functions, but assume aliasing across calls
\n\u4e0d\u91c7\u7528\u5185\u90e8\u51fd\u6570\u6df7\u6dc6\uff0c\u53ea\u6df7\u6dc6\u5916\u90e8\u8c03\u7528 \uff1f\uff1f\uff1f
\n-f[no-]builtin
\ndisable inline expansion of intrinsic functions
\n\u7981\u7528\u56fa\u6709\u51fd\u6570\u5728\u5185\u90e8\u6269\u5c55
\n-fno-builtin-<func>
\ndisable the <func> intrinsic
\n\u7981\u7528\u5185\u5efa\u51fd\u6570 <\u51fd\u6570\u540d>
\n-ffunction-sections
\nseparate functions for the linker (COMDAT)
\n\u5206\u79bb\u8fde\u63a5\u5668\u51fd\u6570
\n-fdata-sections
\nsame as -ffunction-sections
\n\u548c -ffunction-sections \u76f8\u540c
\n-nolib-inline
\ndisable inline expansion of intrinsic functions
\n\u7981\u7528 \u56fa\u6709\u51fd\u6570\u7684\u5185\u8054\u5c55\u5f00<\/p>\n

Code Generation
\n\u4ee3\u7801\u751f\u6210
\n---------------<\/p>\n

-x<code1>[,<code2>,...]
\ngenerate specialized code to run exclusively on processors
\nindicated by <code> as described below
\n\u751f\u6210\u6307\u5b9a\u7684\u4ee3\u7801\u8fd0\u884c\u5728\u4e13\u95e8\u7684\u5904\u7406\u5668\u4e0a
\n<code>\u5185\u8868\u793a\u7684\u53c2\u6570\u5982\u4e0b\u6240\u793a<\/p>\n

Host generate instructions for the highest instruction set and
\nprocessor available on the compilation host machine
\n\u4e3b\u673a\u751f\u6210\u6307\u4ee4\u7528\u4e8e\u5bf9\u5e94\u7684\u5904\u7406\u5668\u4e3b\u673a\u4e0a\u6700\u9ad8\u6307\u4ee4\u96c6\u8bbe\u7f6e<\/p>\n

SSE2 Intel Pentium 4 and compatible Intel processors.\u00a0 Enables new
\noptimizations in addition to Intel processor-specific
\noptimizations
\n\u82f1\u7279\u5c14\u5954\u817e4 \u548c\u517c\u5bb9\u7684\u82f1\u7279\u5c14\u5904\u7406\u5668\uff0c\u5f00\u542f\u65b0\u7684\u4f18\u5316\u7528\u4e8e\u82f1\u7279\u5c14\u6307\u5b9a\u5904\u7406\u5668\u4f18\u5316<\/p>\n

SSE3\u00a0\u00a0\u00a0 Intel(R) Core(TM) processor family with Streaming SIMD
\nExtensions 3 (Intel(R) SSE3) instruction support
\n\u5355\u6307\u4ee4\u591a\u6570\u636e\u6d41(Intel(R)SSE3)\u6307\u4ee4\u96c6\u652f\u6301\u7684\u82f1\u7279\u5c14(R) \u9177\u777f(TM) \u5904\u7406\u5668\u5bb6\u65cf<\/p>\n

SSE3_ATOM Can generate MOVBE instructions for Intel processors and
\ncan optimize for the Intel(R) Atom(TM) processor.
\n\u53ef\u4ee5\u751f\u6210MOVBE \u6307\u4ee4\u7528\u4e8e\u82f1\u7279\u5c14\u5904\u7406\u5668\u548c\u5bf9\u82f1\u7279\u5c14(R)\u51cc\u52a8(TM)\u5904\u7406\u5668<\/p>\n

SSSE3\u00a0\u00a0 Intel(R) Core(TM)2 processor family with Supplemental
\nStreaming SIMD Extensions 3 (SSSE3)
\n\u652f\u6301\u5355\u6307\u4ee4\u591a\u6570\u636e\u6d41\u6307\u4ee4\u96c6\u6269\u5c55\u7684\u82f1\u7279\u5c14\u9177\u777f2\u5904\u7406\u5668\u5bb6\u65cf<\/p>\n

SSE4.1\u00a0 Intel(R) 45nm Hi-k next generation Intel Core(TM)
\nmicroarchitecture with support for SSE4 Vectorizing
\nCompiler and Media Accelerator instructions
\n\u7528\u4e8e\u652f\u6301SSE4 \u77e2\u91cf\u5316\u7f16\u8bd1\u548c\u5a92\u4f53\u52a0\u901f\u6307\u4ee4\u7684 \u82f1\u7279\u5c14 45\u7eb3\u7c73Hi-k \u5de5\u827a\u7684\u4e0b\u4e00
\n\u4ee3\u82f1\u7279\u5c14\u9177\u777f\u5fae\u67b6\u6784<\/p>\n

SSE4.2\u00a0 Can generate Intel(R) SSE4 Efficient Accelerated String
\nand Text Processing instructions supported by Intel(R)
\nCore(TM) i7 processors. Can generate Intel(R) SSE4
\nVectorizing Compiler and Media Accelerator, Intel(R) SSSE3,
\nSSE3, SSE2, and SSE instructions and it can optimize for
\nthe Intel(R) Core(TM) processor family.
\n\u53ef\u4ee5\u751f\u6210\u82f1\u7279\u5c14 SSE4 \u9ad8\u6548\u5b57\u4e32\u548c\u6587\u672c\u5904\u7406\u52a0\u901f\u6307\u4ee4\u652f\u6301\u7684Intel \u9177\u777f i7 \u5904
\n\u7406\u5668\u53ef\u4ee5\u751f\u6210\u82f1\u7279\u5c14 SSE4 \u77e2\u91cf\u5316\u7f16\u8bd1\u548c\u5a92\u4f53\u52a0\u901f\uff0c\u82f1\u7279\u5c14SSE3,SSE2 \u548cSSE
\n\u6307\u4ee4\u96c6\u548c\u5b83\u80fd\u591f\u4f18\u5316\u7684 \u82f1\u7279\u5c14\u9177\u777f\u5904\u7406\u5668\u5bb6\u65cf<\/p>\n

AVX\u00a0\u00a0\u00a0\u00a0 Enable Intel(R) Advanced Vector Extensions instructions
\n\u542f\u7528\u82f1\u7279\u5c14\u9ad8\u7ea7\u5411\u91cf\u6269\u5c55\u6307\u4ee4<\/p>\n

-ax<code1>[,<code2>,...]
\ngenerate code specialized for processors specified by <codes>
\nwhile also generating generic IA-32 instructions.
\n\u751f\u6210\u4ee3\u7801\u6307\u5b9a\u4e0e\u7528\u4e8e\u6307\u5b9a\u7684\u5904\u7406\u5668\u5230 <codes> ,\u4e5f\u540c\u65f6\u751f\u6210\u901a\u7528\u7684IA-32 \u6307\u4ee4<\/p>\n

<codes> includes one or more of the following:
\n<codes > \u91cc\u9762\u5305\u542b\u4e00\u4e2a\u6216\u591a\u4e2a\u5982\u4e0b\u6240\u793a<\/p>\n

SSE2 Intel Pentium 4 and compatible Intel processors.\u00a0 Enables new
\noptimizations in addition to Intel processor-specific
\noptimizations
\n\u82f1\u7279\u5c14\u5954\u817e4 \u548c\u517c\u5bb9\u7684\u82f1\u7279\u5c14\u5904\u7406\u5668\uff0c\u5f00\u542f\u65b0\u7684\u4f18\u5316\u7528\u4e8e\u82f1\u7279\u5c14\u6307\u5b9a\u5904\u7406\u5668\u4f18\u5316<\/p>\n

SSE3\u00a0\u00a0\u00a0 Intel(R) Core(TM) processor family with Streaming SIMD
\nExtensions 3 (Intel(R) SSE3) instruction support
\n\u5355\u6307\u4ee4\u591a\u6570\u636e\u6d41(Intel(R)SSE3)\u6307\u4ee4\u96c6\u652f\u6301\u7684\u82f1\u7279\u5c14(R) \u9177\u777f(TM) \u5904\u7406\u5668\u5bb6\u65cf<\/p>\n

SSSE3\u00a0\u00a0 Intel(R) Core(TM)2 processor family with Supplemental
\nStreaming SIMD Extensions 3 (SSSE3)
\n\u652f\u6301\u5355\u6307\u4ee4\u591a\u6570\u636e\u6d41\u6307\u4ee4\u96c6\u6269\u5c55\u7684\u82f1\u7279\u5c14\u9177\u777f2\u5904\u7406\u5668\u5bb6\u65cf<\/p>\n

SSE4.1\u00a0 Intel(R) 45nm Hi-k next generation Intel Core(TM)
\nmicroarchitecture with support for Streaming SIMD
\nExtensions 4 (Intel(R) SSE4) Vectorizing
\nCompiler and Media Accelerator instructions
\n\u7528\u4e8e\u652f\u6301SSE4 \u77e2\u91cf\u5316\u7f16\u8bd1\u548c\u5a92\u4f53\u52a0\u901f\u6307\u4ee4\u7684 \u82f1\u7279\u5c14 45\u7eb3\u7c73Hi-k \u5de5\u827a\u7684\u4e0b\u4e00
\n\u4ee3\u82f1\u7279\u5c14\u9177\u777f\u5fae\u67b6\u6784<\/p>\n

SSE4.2\u00a0 Can generate Intel(R) SSE4 Efficient Accelerated String
\nand Text Processing instructions supported by Intel(R)
\nCore(TM) i7 processors. Can generate Intel(R) SSE4
\nVectorizing Compiler and Media Accelerator, Intel(R) SSSE3,
\nSSE3, SSE2, and SSE instructions and it can optimize for
\nthe Intel(R) Core(TM) processor family.
\n\u53ef\u4ee5\u751f\u6210\u82f1\u7279\u5c14 SSE4 \u9ad8\u6548\u5b57\u4e32\u548c\u6587\u672c\u5904\u7406\u52a0\u901f\u6307\u4ee4\u652f\u6301\u7684Intel \u9177\u777f i7 \u5904
\n\u7406\u5668\uff0c\u53ef\u4ee5\u751f\u6210\u82f1\u7279\u5c14 SSE4 \u77e2\u91cf\u5316\u7f16\u8bd1\u548c\u5a92\u4f53\u52a0\u901f\uff0c\u82f1\u7279\u5c14 SSE3, SSE2 \u548c
\nSSE\u6307\u4ee4\u96c6\u548c\u5b83\u80fd\u591f\u4f18\u5316\u7684\u82f1\u7279\u5c14\u9177\u777f\u5904\u7406\u5668\u5bb6\u65cf<\/p>\n

AVX\u00a0\u00a0\u00a0\u00a0 Enable Intel(R) Advanced Vector Extensions instructions
\n\u542f\u7528\u82f1\u7279\u5c14\u9ad8\u7ea7\u5411\u91cf\u6269\u5c55\u6307\u4ee4
\n-mcpu=<cpu>
\nsame as -mtune=<cpu>
\n\u548c -mtune=<cpu> \u76f8\u540c<\/p>\n

-mtune=<cpu>
\noptimize for a specific <cpu>
\n\u7279\u5b9a\u4f18\u5316<cpu>
\npentium3\u00a0 - optimize for Pentium(R) III processors
\n\u7528\u4e8e \u5954\u817e3 \u5904\u7406\u5668\u4f18\u5316
\npentium4\u00a0 - optimize for Pentium(R) 4 processor (DEFAULT)
\n\u7528\u4e8e \u5954\u817e4 \u5904\u7406\u5668\u4f18\u5316(\u9ed8\u8ba4)<\/p>\n

-march=<cpu>
\ngenerate code exclusively for a given <cpu>
\n\u5355\u72ec\u4e3aCPU\u751f\u6210\u4ee3\u7801<cpu>
\npentium3\u00a0 - streaming SIMD extensions
\n\u5355\u6307\u4ee4\u591a\u6570\u636e\u6d41\u6269\u5c55
\npentium4\u00a0 - Pentium(R) 4 New Instructions
\n\u5954\u817e 4 \u65b0\u6307\u4ee4<\/p>\n

-mia32\u00a0\u00a0\u00a0 generates code which will run on any Intel Pentium or later
\nprocessor.\u00a0 Disables any default extended instruction settings, and
\nany previously set extended instruction settings.
\n\u751f\u6210\u4ee3\u7801\u5c06\u8fd0\u884c\u4e8e\u4efb\u4f55\u82f1\u7279\u5c14\u5954\u817e\u6216\u66f4\u9ad8\u7684\u5904\u7406\u5668\uff0c\u7981\u7528\u4efb\u4f55\u9ed8\u8ba4\u6269\u5c55\u7684\u6307\u4ee4\u8bbe\u7f6e\u548c\u5148\u524d
\n\u8bbe\u7f6e\u7684\u6269\u5c55\u7684\u6307\u4ee4\u8bbe\u7f6e<\/p>\n

-msse\u00a0\u00a0\u00a0\u00a0 generate code for Intel Pentium III and compatible Intel processors
\n\u751f\u6210\u4ee3\u7801\u7528\u4e8e\u82f1\u7279\u5c14\u5954\u817eIII\u548c\u82f1\u7279\u5c14\u517c\u5bb9\u5904\u7406\u5668<\/p>\n

-msse2\u00a0\u00a0\u00a0 generate code for Intel Pentium 4 and compatible Intel processors
\n\u751f\u6210\u4ee3\u7801\u7528\u4e8e\u82f1\u7279\u5c14\u5954\u817e4\u548c\u82f1\u7279\u5c14\u517c\u5bb9\u5904\u7406\u5668<\/p>\n

-msse3\u00a0\u00a0\u00a0 generate code for Intel(R) Core(TM) Duo processors, Intel(R) Core(TM)
\nSolo processors, Intel Pentium 4 and compatible Intel processors with
\nStreaming SIMD Extensions 3 (Intel(R) SSE3) instruction support
\n\u751f\u6210\u4ee3\u7801\u7528\u4e8e\u5355\u6307\u4ee4\u591a\u6570\u636e\u6d41\u6307\u4ee4(SSE3)\u652f\u6301\u7684\u82f1\u7279\u5c14\u9177\u777f2\u5904\u7406\u5668\uff0c\u82f1\u7279\u5c14\u9177\u777f\u4f4e\u7535\u538b
\n\u7248\u5904\u7406\u5668\u82f1\u7279\u5c14\u5954\u817e\u56db\u548c\u517c\u5bb9\u7684\u82f1\u7279\u5c14\u5904\u7406\u5668<\/p>\n

-mssse3\u00a0\u00a0 Intel(R) Core(TM)2 processor family with Supplemental Streaming
\nSIMD Extensions 3 (SSSE3)
\n\u82f1\u7279\u5c14\u9177\u777f2\u5904\u7406\u5668\u5bb6\u65cf\u5e76\u652f\u6301\u5355\u6307\u4ee4\u591a\u6570\u636e\u6d41\u6269\u5c55(SSE3)<\/p>\n

-msse4.1\u00a0 Intel(R) 45nm Hi-k next generation Intel Core(TM) microarchitecture
\nwith support for Streaming SIMD Extensions 4 (Intel(R) SSE4)
\nVectorizing Compiler and Media Accelerator instructions
\n\u82f1\u7279\u5c14 45\u7eb3\u7c73Hi-k \u5de5\u827a\u7684\u4e0b\u4e00\u4ee3\u82f1\u7279\u5c14\u9177\u777f\u5fae\u67b6\u6784\u7528\u4e8e\u652f\u6301SSE4 \u77e2\u91cf\u5316\u7f16\u8bd1\u548c\u5a92\u4f53\u52a0\u901f\u6307\u4ee4\u7684
\n-minstruction=<keyword>
\nRefine instruction set output for the selected target processor
\n\u6307\u5b9a\u5df2\u9009\u62e9\u7684\u76ee\u6807\u5904\u7406\u5668\u7cbe\u7b80\u6307\u4ee4\u96c6\u8f93\u51fa\u8bbe\u7f6e
\n[no]movbe\u00a0 - Do\/do not generate MOVBE instructions with SSE3_ATOM
\n(requires -xSSE3_ATOM)
\n\u4e0d\u751f\u6210 MOVBE\u6307\u4ee4\u7528\u4e8e SSE3_ATOM\u00a0 (\u8981\u6c42 -xSSE3_ATOM)<\/p>\n

-f[no-]omit-frame-pointer
\nenable(DEFAULT)\/disable use of EBP as general purpose register.
\n-fno-omit-frame-pointer replaces -fp
\n\u5f00\u542f(\u9ed8\u8ba4)\/\u5173\u95ed EBP\u4f5c\u4e3a\u901a\u7528\u5bc4\u5b58\u5668<\/p>\n

-f[no-]exceptions
\nenable\/disable(DEFAULT) C++ exception handling table generation
\n\u5f00\u542f\/\u5173\u95ed(\u9ed8\u8ba4) \u751f\u6210C++ \u5f02\u5e38\u5904\u7406\u8868<\/p>\n

-fnon-call-exceptions
\nenable\/disable(DEFAULT) code that allows exceptions from trapping
\ninstructions to be caught
\n\u5f00\u542f\/\u5173\u95ed(\u9ed8\u8ba4)\u5141\u8bb8\u6355\u83b7\u4f8b\u5916\u7684\u4ee3\u7801\u5f02\u5e38<\/p>\n

Interprocedural Optimization (IPO)
\n\u95f4\u63a5\u4f18\u5316
\n----------------------------------<\/p>\n

-[no-]ip\u00a0 enable(DEFAULT)\/disable single-file IP optimization
\nwithin files
\n\u5f00\u542f(\u9ed8\u8ba4)\/\u7981\u7528\u5355\u6587\u4ef6\u95f4\u63a5\u4f18\u5316\u5185\u90e8\u6587\u4ef6<\/p>\n

-ipo[n]\u00a0\u00a0 enable multi-file IP optimization between files
\n\u5728\u6587\u4ef6\u4e2d\u5f00\u542f\u591a\u6587\u4ef6\u95f4\u63a5\u4f18\u5316<\/p>\n

-ipo-c\u00a0\u00a0\u00a0 generate a multi-file object file (ipo_out.o)
\n\u751f\u6210\u4e00\u4e2a\u591a\u6587\u4ef6\u7684\u76ee\u6807\u6587\u4ef6 (ipo_out.o)<\/p>\n

-ipo-S\u00a0\u00a0\u00a0 generate a multi-file assembly file (ipo_out.S)
\n\u751f\u6210\u591a\u6587\u4ef6\u96c6\u6587\u4ef6(ipo_out.S)<\/p>\n

-ip-no-inlining
\ndisable full and partial inlining
\n\u7981\u7528\u5168\u90e8\u548c\u90e8\u5206\u5185\u8054<\/p>\n

-ip-no-pinlining
\ndisable partial inlining
\n\u7981\u7528\u90e8\u5206\u5185\u8054<\/p>\n

-ipo-separate
\ncreate one object file for every source file (overrides -ipo[n])
\n\u4e3a\u6bcf\u4e2a\u6e90\u6587\u4ef6\u521b\u5efa\u4e00\u4e2a\u5bf9\u8c61(overrides -ipo[n])<\/p>\n

-ipo-jobs<n>
\nspecify the number of jobs to be executed simultaneously during the
\nIPO link phase
\n\u6307\u5b9a\u961f\u5217\u7f16\u53f7\uff0c\u4ee5\u540c\u65f6\u6267\u884cIPO\u8fde\u63a5\uff1f<\/p>\n

Advanced Optimizations
\n\u9ad8\u7ea7\u4f18\u5316
\n----------------------<\/p>\n

-unroll[n]
\nset maximum number of times to unroll loops.\u00a0 Omit n to use default
\nheuristics.\u00a0 Use n=0 to disable the loop unroller
\n\u8bbe\u7f6e\u5faa\u73af\u5c55\u5f00\u7684\u6700\u5927\u6b21\u6570\uff0c\u53bb\u6389 n \u5219\u4f7f\u7528\u9ed8\u8ba4\u542f\u53d1\uff0c\u4f7f\u7528 n=0 \u5219\u53bb\u6389unroller \u5faa\u73af<\/p>\n

-[no-]unroll-aggressive
\nenables more aggressive unrolling heuristics
\n\u5f00\u542f\u66f4\u591a\u7684\u5f3a\u5236\u5c55\u5f00\u542f\u53d1 \uff1f\uff1f<\/p>\n

-funroll-loops
\nunroll loops based on default heuristics
\n\u5c55\u5f00\u5faa\u73af\u5728\u9ed8\u8ba4\u542f\u53d1<\/p>\n

-[no-]scalar-rep
\nenable(DEFAULT)\/disable scalar replacement (requires -O3)
\n\u542f\u7528 \u7981\u7528 \u6807\u91cf\u66ff\u6362\uff08\u9700\u8981 -o3 \uff09
\n-[no-]ansi-alias
\nenable\/disable(DEFAULT) use of ANSI aliasing rules optimizations;
\nuser asserts that the program adheres to these rules<\/p>\n

-[no-]complex-limited-range
\nenable\/disable(DEFAULT) the use of the basic algebraic expansions of
\nsome complex arithmetic operations.\u00a0 This can allow for some
\nperformance improvement in programs which use a lot of complex
\narithmetic at the loss of some exponent range.
\n-[no-]alias-const
\nenable\/disable(DEFAULT) a heuristic stating that if two arguments to
\na function have pointer type, a pointer to const does not alias a
\npointer to non-const. Also known as the input\/output buffer rule, it
\nassumes that input and output buffer arguments do not overlap.
\n-fargument-alias
\narguments may alias each other and may alias global storage
\n-fargument-noalias
\narguments do not alias each other but may alias global storage
\n-fargument-noalias-global
\narguments do not alias each other and do not alias global storage
\n-ftls-model=<model>
\nchange thread-local storage model, where <model> can be the
\nfollowing: global-dynamic, local-dynamic, initial-exec or local-exec
\n-[no-]opt-multi-version-aggressive
\nenables more aggressive multi-versioning to check for pointer
\naliasing and scalar replacement
\n-opt-ra-region-strategy[=<keyword>]
\nselect the method that the register allocator uses to partition each
\nroutine into regions
\nroutine - one region per routine
\nblock\u00a0\u00a0 - one region per block
\ntrace\u00a0\u00a0 - one region per trace
\nloop\u00a0\u00a0\u00a0 - one region per loop
\ndefault - compiler selects best option
\n-[no-]vec
\nenables(DEFAULT)\/disables vectorization
\n\u5f00\u542f \u5173\u95ed \u5411\u91cf\u5316
\n-[no-]vec-guard-write
\nenables cache\/bandwidth optimization for stores under conditionals
\nwithin vector loops
\n\u5f00\u542f\u7f13\u5b58\/\u5e26\u5bbd\u4f18\u5316\u7528\u4e8e\uff1f\uff1f\uff1f
\n-vec-threshold[n]
\nsets a threshold for the vectorization of loops based on the
\nprobability of profitable execution of the vectorized loop in
\nparallel
\n-opt-malloc-options={0|1|2|3|4}
\nspecify malloc configuration parameters.\u00a0 Specifying a non-zero <n>
\nvalue will cause alternate configuration parameters to be set for
\nhow malloc allocates and frees memory
\n-opt-calloc
\nenable\/disable(DEFAULT) calls to fast calloc function
\n-opt-jump-tables=<arg>
\ncontrol the generation of jump tables
\ndefault - let the compiler decide when a jump table, a series of
\nif-then-else constructs or a combination is generated
\nlarge\u00a0\u00a0 - generate jump tables up to a certain pre-defined size
\n(64K entries)
\n<n>\u00a0\u00a0\u00a0\u00a0 - generate jump tables up to <n> in size
\nuse -no-opt-jump-tables to lower switch statements as chains of
\nif-then-else constructs
\n-fno-jump-tables
\ndo not generate jump tables for switches and if-then-else statements
\n-opt-block-factor=<n>
\nspecify blocking factor for loop blocking
\n-ffreestanding
\ncompile in a freestanding environment where the standard library
\nmay not be present
\n-opt-streaming-stores <arg>
\nspecifies whether streaming stores are generated
\nalways - enables generation of streaming stores under the
\nassumption that the application is memory bound
\nauto\u00a0\u00a0 - compiler decides when streaming stores are used (DEFAULT)
\nnever\u00a0 - disables generation of streaming stores
\n-ipp[=<arg>]
\nlink some or all of the Intel(R) Integrated Performance Primitives
\n(Intel(R) IPP) libraries and bring in the associated headers
\ncommon\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - link using the main libraries set.\u00a0 This is the
\ndefault value when -ipp is specified
\ncrypto\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - link using the main libraries set and the crypto
\nlibrary
\nnonpic\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - link using the version of the libraries that do not
\nhave position independent code
\nnonpic_crypto - link using the crypto library and the version of
\nthe libraries that do not have position independent
\ncode
\n-mkl[=<arg>]
\nlink to the Intel(R) Math Kernel Library (Intel(R) MKL) and bring
\nin the associated headers
\nparallel\u00a0\u00a0 - link using the threaded Intel(R) MKL libraries. This
\nis the default when -mkl is specified
\nsequential - link using the non-threaded Intel(R) MKL libraries
\ncluster\u00a0\u00a0\u00a0 - link using the Intel(R) MKL Cluster libraries plus
\nthe sequential Intel(R) MKL libraries
\n-tbb\u00a0\u00a0\u00a0\u00a0\u00a0 link to the Intel(R) Threading Building Blocks (Intel(R) TBB)
\nlibraries and bring in the associated headers
\n-[no-]opt-subscript-in-range
\nassumes no overflows in the intermediate computation of the
\nsubscripts
\n-[no-]use-intel-optimized-headers
\ntake advantage of the optimized header files<\/p>\n

Profile Guided Optimization (PGO)
\n\u914d\u7f6e\u5411\u5bfc\u4f18\u5316
\n---------------------------------<\/p>\n

-prof-dir <dir>
\nspecify directory for profiling output files (*.dyn and *.dpi)
\n\u6307\u5b9a\u76ee\u5f55\u7528\u4e8e\u8f93\u51fa\u6587\u4ef6(*.dyn and *.dpi)<\/p>\n

-prof-src-root <dir>
\nspecify project root directory for application source files to
\nenable relative path resolution during profile feedback on sources
\nbelow that directory
\n\u6307\u5b9a\u5e94\u7528\u7a0b\u5e8f\u7684\u6e90\u4ee3\u7801\u7684\u6839\u76ee\u5f55\u6587\u4ef6\uff0c\u53ef\u4ee5\u5728\u914d\u7f6e\u6587\u4ef6\u7684\u76f8\u5bf9\u8def\u5f84\u6765\u6e90\u53cd\u9988\u51b3\u8bae
\n??????????\u4f4e\u4e8e\u8be5\u76ee\u5f55<\/p>\n

-prof-src-root-cwd
\nspecify the current directory as the project root directory for
\napplication source files to enable relative path resolution during
\nprofile feedback on sources below that directory
\n\u6307\u5b9a\u4e3a\u9879\u76ee\u7684\u6839\u76ee\u5f55\u4e3a\u5f53\u524d\u76ee\u5f55\u5e94\u7528\u7a0b\u5e8f\u6e90\u4ee3\u7801\uff0c\u4ee5\u4fbf\u5728\u89e3\u51b3\u76f8\u5bf9\u8def\u5f84
\n??????????\u6587\u4ef6\u53cd\u9988\u4f4e\u4e8e\u8be5\u76ee\u5f55\u7684\u6765\u6e90<\/p>\n

-[no-]prof-src-dir
\nspecify whether directory names of sources should be
\nconsidered when looking up profile records within the .dpi file
\n\u5177\u4f53\u8bf4\u660e\u6765\u6e90\u7684\u76ee\u5f55\u540d\u5e94\u8be5\u662f\u67e5\u627e\u65f6\u8003\u8651\u8303\u56f4\u5185\u3002dpi\u7684\u6587\u4ef6\u8bb0\u5f55\u7684\u6587\u4ef6<\/p>\n

-prof-file <file>
\nspecify file name for profiling summary file
\n\u6307\u5b9a\u6587\u4ef6\u7684\u6587\u4ef6\u540d\u5206\u6790\u603b\u7ed3<\/p>\n

-[no-]prof-data-order
\nenable\/disable(DEFAULT) static data ordering with profiling
\n\u542f\u7528\/\u7981\u7528\uff08\u9ed8\u8ba4\uff09\u9759\u6001\u6570\u636e\u6392\u5e8f\u5206\u6790<\/p>\n

-[no-]prof-func-order
\nenable\/disable(DEFAULT) function ordering with profiling
\n\u542f\u7528\/\u7981\u7528\uff08\u9ed8\u8ba4\uff09\u51fd\u6570\u5206\u6790\u6392\u5e8f<\/p>\n

-[no-]prof-func-groups
\nenable(DEFAULT with PGO)\/disable function grouping
\n\u542f\u7528\uff08\u9ed8\u8ba4\u4e0e\u901a\u4f8b\uff09\/\u7981\u7528\u529f\u80fd\u5206\u7ec4<\/p>\n

-prof-gen[=keyword]
\ninstrument program for profiling.
\nOptional keyword may be srcpos or globdata
\n\u673a\u5668\u8bed\u8a00\u5206\u6790\uff0c\u53ef\u9009\u5173\u952e\u5b57 srcpos \u6216\u8005 globdata<\/p>\n

-no-prof-gen
\ndisable profiling instrumentation
\n\u7981\u7528\u5206\u6790\u4eea\u5668<\/p>\n

-prof-use[=<arg>]
\nenable use of profiling information during optimization
\n\u542f\u7528\u5206\u6790\u5728\u4f18\u5316\u8fc7\u7a0b\u4e2d\u4fe1\u606f\u7684\u4f7f\u7528
\nweighted\u00a0 - invokes profmerge with -weighted option to scale data
\nbased on run durations
\n[no]merge - enable(default)\/disable the invocation of the profmerge
\ntool
\n-no-prof-use
\ndisable use of profiling information during optimization
\n-opt-prefetch[=n]
\nenable levels of prefetch insertion, where 0 disables.
\nn may be 0 through 4 inclusive.\u00a0 Default is 2.
\n-no-opt-prefetch
\ndisable(DEFAULT) prefetch insertion.\u00a0 Equivalent to -opt-prefetch=0
\n-p\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 compile and link for function profiling with UNIX gprof tool
\nOn IA32 and Intel(r)64, -pg is also valid
\n-f[no-]instrument-functions
\ndetermine whether function entry and exit points are instrumented
\n-prof-hotness-threshold=<val>
\nset the hotness threshold for function grouping and function ordering
\nval indicates percentage of functions to be placed in hot region.
\nThis option requires -prof-use
\nand -prof-func-groups or -prof-func-order
\nOptimization Reports
\n\u4f18\u5316\u62a5\u544a
\n--------------------<\/p>\n

-vec-report[n]
\ncontrol amount of vectorizer diagnostic information
\nn=0\u00a0\u00a0\u00a0 no diagnostic information
\n\u4e0d\u663e\u793a\u8bca\u65ad\u4fe1\u606f
\nn=1\u00a0\u00a0\u00a0 indicate vectorized loops (DEFAULT)
\n\u663e\u793a\u77e2\u91cf\u5faa\u73af
\nn=2\u00a0\u00a0\u00a0 indicate vectorized\/non-vectorized loops
\n\u663e\u793a\u77e2\u91cf\/\u975e\u77e2\u91cf\u5faa\u73af\u5faa\u73af
\nn=3\u00a0\u00a0\u00a0 indicate vectorized\/non-vectorized loops and prohibiting
\ndata dependence information
\n\u663e\u793a\u77e2\u91cf\/\u975e\u5faa\u73af\u548c\u7981\u6b62\u77e2\u91cf\u6570\u636e\u76f8\u5173\u7684\u4fe1\u606f
\nn=4\u00a0\u00a0\u00a0 indicate non-vectorized loops
\n\u8868\u660e\u975e\u77e2\u91cf\u73af
\nn=5\u00a0\u00a0\u00a0 indicate non-vectorized loops and prohibiting data
\ndependence information
\n\u663e\u793a\u975e\u5faa\u73af\u548c\u7981\u6b62\u77e2\u91cf\u6570\u636e\u4f9d\u5b58\u4fe1\u606f
\n-opt-report [n]
\ngenerate an optimization report to stderr
\n\u751f\u6210\u4f18\u5316\u62a5\u544a\u5230stderr
\n0\u00a0\u00a0 disable optimization report output
\n\u7981\u7528\u4f18\u5316\u62a5\u544a\u8f93\u51fa
\n1\u00a0\u00a0 minimum report output
\n\u6700\u4f4e\u7684\u62a5\u544a\u8f93\u51fa
\n2\u00a0\u00a0 medium output (DEFAULT when enabled)
\n\u4e2d\u7b49\u8f93\u51fa\uff08\u9ed8\u8ba4\u542f\u7528\u65f6\uff09
\n3\u00a0\u00a0 maximum report output
\n\u6700\u5927\u8f93\u51fa\u62a5\u544a
\n-opt-report-file=<file>
\nspecify the filename for the generated report
\n\u6307\u5b9a\u7684\u6587\u4ef6\u540d\u751f\u6210\u7684\u62a5\u544a
\n-opt-report-phase=<phase>
\nspecify the phase that reports are generated against
\n\u6307\u5b9a\u751f\u6210\u7684\u62a5\u544a\u5bf9\u76f8
\n-opt-report-routine=<name>
\nreports on routines containing the given name
\n\u62a5\u544a\u5305\u542b\u7ed9\u5b9a\u540d\u79f0\u4f8b\u7a0b
\n-opt-report-help
\ndisplay the optimization phases available for reporting
\n\u663e\u793a\u53ef\u7528\u7684\u4f18\u5316\u62a5\u544a\u4fe1\u606f
\n-tcheck [mode]
\nenable analysis of threaded applications (requires Intel(R) Thread
\nChecker; cannot be used with compiler alone)
\ntci - instruments a program to perform a thread-count-independent
\nanalysis
\ntcd - instruments a program to perform a thread-count-dependent
\nanalysis (DEFAULT when mode is not used)
\napi - instruments a program at the api-imports level
\n-tprofile
\ngenerate instrumentation to analyze multi-threading performance
\n(requires Intel(R) Thread Profiler; cannot be used with compiler
\nalone)
\n-tcollect[=<lib>]
\ninserts instrumentation probes calling the Intel(R) Trace Collector
\nAPI.\u00a0 The library -l<lib> is linked in the default being -lVT
\n(requires Intel(R) Trace Collector)
\n-tcollect-filter file
\nEnable or disable the instrumentation of specified functions.
\n(requires Intel(R) Trace Collector)<\/p>\n

OpenMP* and Parallel Processing
\nOpenMP* \u548c\u5e76\u884c\u5904\u7406
\n-------------------------------<\/p>\n

-openmp\u00a0\u00a0 enable the compiler to generate multi-threaded code based on the
\nOpenMP* directives
\n-openmp-profile
\nenable analysis of OpenMP application when the Intel(R) Thread
\nProfiler is installed
\n-openmp-stubs
\nenables the user to compile OpenMP programs in sequential mode.\u00a0 The
\nOpenMP directives are ignored and a stub OpenMP library is linked
\n(sequential)
\n-openmp-report{0|1|2}
\ncontrol the OpenMP parallelizer diagnostic level
\n-openmp-lib <ver>
\nchoose which OpenMP library version to link with
\ncompat - use the GNU compatible OpenMP run-time libraries
\n(DEFAULT)
\n-openmp-link <library>
\nchoose whether to link with the static or dynamic OpenMP
\nlibraries.\u00a0 Default is dynamic.
\n-openmp-task <arg>
\nchoose which OpenMP tasking model to support
\nomp\u00a0\u00a0 - support OpenMP 3.0 tasking (DEFAULT)
\nintel - support Intel taskqueuing
\n-openmp-threadprivate <ver>
\nchoose which threadprivate implementation to use
\ncompat - use the GNU compatible thread local storage
\nlegacy - use the Intel compatible implementation
\n(DEFAULT)
\n-parallel
\nenable the auto-parallelizer to generate multi-threaded code for
\nloops that can be safely executed in parallel
\n-par-report{0|1|2|3}
\ncontrol the auto-parallelizer diagnostic level
\n-par-threshold[n]
\nset threshold for the auto-parallelization of loops where n is an
\ninteger from 0 to 100
\n-[no-]par-runtime-control
\nenable compiler to generate runtime control code for effective
\nautomatic parallelization
\n-par-schedule-static[=n]
\nSpecifies a scheduling algorithm for DO loop iteration.
\nDivides iterations into contiguous pieces.\u00a0 Size n if
\nspecified, equal sized pieces if not.
\n-par-schedule-static-balanced[=n]
\nDivides iterations into even-sized chunks.\u00a0 Size n if
\nspecified, equal sized pieces if not.
\n-par-schedule-static-steal[=n]
\nDivides iterations into even-sized chunks, but allows
\nthreads to steal parts of chunks from neighboring threads
\n-par-schedule-dynamic[=n]
\nSpecifies a scheduling algorithm for DO loop iteration.
\nAssigns iterations to threads in chunks dynamically.
\nChunk size is n iterations if specified, otherwise 1.
\n-par-schedule-guided[=n]
\nSpecifies a scheduling algorithm for DO loop iteration.
\nIndicates a minimum number of iterations.\u00a0 If specified,
\nn is the minimum number, otherwise 1.
\n-par-schedule-guided-analytical[=n]
\nDivides iterations by using exponential distribution or
\ndynamic distributions.
\n-par-schedule-runtime
\nSpecifies a scheduling algorithm for DO loop iteration.
\nDefers the scheduling decision until runtime.
\n-par-schedule-auto
\nLets the compiler or run-time system determine the
\nscheduling algorithm.
\n-par-affinity=[<modifier>,...]<type>[,<permute>][,<offset>]
\ntune application performance by setting different thread affinity
\n-par-num-threads=<n>
\ntune application performance by setting different number of threads<\/p>\n

Floating Point
\n\u6d6e\u70b9\uff08\u8fd0\u7b97\uff09
\n--------------<\/p>\n

-fp-model <name>
\nenable <name> floating point model variation
\n[no-]except - enable\/disable floating point semantics
\n\u542f\u7528\/\u7981\u7528\u6d6e\u70b9\u8bed\u4e49
\nfast[=1|2]\u00a0 - enables more aggressive floating point optimizations
\n\u53ef\u4ee5\u66f4\u79ef\u6781\u7684\u6d6e\u70b9\u4f18\u5316
\nprecise\u00a0\u00a0\u00a0\u00a0 - allows value-safe optimizations
\n\u5141\u8bb8\u503c\u5b89\u5168\u4f18\u5316
\nsource\u00a0\u00a0\u00a0\u00a0\u00a0 - enables intermediates in source precision
\n\u4f7f\u6e90\u4e2d\u95f4\u4f53\u7cbe\u5ea6
\nstrict\u00a0\u00a0\u00a0\u00a0\u00a0 - enables -fp-model precise -fp-model except, disables
\ncontractions and enables pragma stdc fenv_access
\ndouble\u00a0\u00a0\u00a0\u00a0\u00a0 - rounds intermediates in 53-bit (double) precision
\nextended\u00a0\u00a0\u00a0 - rounds intermediates in 64-bit (extended) precision
\n-fp-speculation=<mode>
\nenable floating point speculations with the following <mode>
\nconditions:
\nfast\u00a0\u00a0 - speculate floating point operations (DEFAULT)
\nsafe\u00a0\u00a0 - speculate only when safe
\nstrict - same as off
\noff\u00a0\u00a0\u00a0 - disables speculation of floating-point operations
\n-pc32\u00a0\u00a0\u00a0\u00a0 set internal FPU precision to 24 bit significand
\n\u8bbe\u7f6e\u5185\u90e8\u6d6e\u70b9\u8fd0\u7b97\u5355\u5143\u7cbe\u786e\u523024\u4f4d\u5c3e\u6570<\/p>\n

-pc64\u00a0\u00a0\u00a0\u00a0 set internal FPU precision to 53 bit significand (DEFAULT)
\n\u8bbe\u7f6e\u5185\u90e8\u6d6e\u70b9\u8fd0\u7b97\u5355\u5143\u7cbe\u786e\u523053\u4f4d\u5c3e\u6570<\/p>\n

-pc80\u00a0\u00a0\u00a0\u00a0 set internal FPU precision to 64 bit significand
\n\u8bbe\u7f6e\u5185\u90e8\u6d6e\u70b9\u8fd0\u7b97\u5355\u5143\u7cbe\u786e\u523064\u4f4d\u5c3e\u6570
\n-mp1\u00a0\u00a0\u00a0\u00a0\u00a0 improve floating-point precision (speed impact less than -mp)
\n\u6539\u8fdb\u6d6e\u70b9\u7cbe\u5ea6<\/p>\n

-mieee-fp
\nsame as -mp, can be disabled with -mno-ieee-fp
\n-[no-]prec-sqrt
\ndetermine if certain square root optimizations are enabled
\n-[no-]prec-div
\nimprove precision of FP divides (some speed impact)
\n-[no-]fast-transcendentals
\ngenerate a faster version of the transcendental functions
\n-[no-]fp-port
\nround fp results at assignments and casts (some speed impact)
\n-fp-stack-check
\nenable fp stack checking after every function\/procedure call
\n-rcd\u00a0\u00a0\u00a0\u00a0\u00a0 rounding mode to enable fast float-to-int conversions
\n-[no-]ftz
\nenable\/disable flush denormal results to zero<\/p>\n

Inlining
\n\u5185\u8054
\n--------<\/p>\n

-inline-level=<n>
\ncontrol inline expansion:
\nn=0\u00a0 disable inlining
\nn=1\u00a0 inline functions declared with __inline, and perform C++
\ninlining
\nn=2\u00a0 inline any function, at the compiler's discretion
\n-f[no-]inline
\ninline functions declared with __inline, and perform C++ inlining
\n-f[no-]inline-functions
\ninline any function at the compiler's discretion
\n-finline-limit=<n>
\nset maximum number of statements a function can have and still be
\nconsidered for inlining
\n-inline-min-size=<n>
\nset size limit for inlining small routines
\n-no-inline-min-size
\nno size limit for inlining small routines
\n-inline-max-size=<n>
\nset size limit for inlining large routines
\n-no-inline-max-size
\nno size limit for inlining large routines
\n-inline-max-total-size=<n>
\nmaximum increase in size for inline function expansion
\n-no-inline-max-total-size
\nno size limit for inline function expansion
\n-inline-max-per-routine=<n>
\nmaximum number of inline instances in any function
\n-no-inline-max-per-routine
\nno maximum number of inline instances in any function
\n-inline-max-per-compile=<n>
\nmaximum number of inline instances in the current compilation
\n-no-inline-max-per-compile
\nno maximum number of inline instances in the current compilation
\n-inline-factor=<n>
\nset inlining upper limits by n percentage
\n-no-inline-factor
\ndo not set set inlining upper limits
\n-inline-forceinline
\ntreat inline routines as forceinline
\n-inline-calloc
\ndirects the compiler to inline calloc() calls as malloc()\/memset()<\/p>\n

Output, Debug, PCH
\n\u8f93\u51fa\uff0c\u8c03\u8bd5\uff0c\u9884\u7f16\u8bd1\u5934\u6587\u4ef6
\n------------------<\/p>\n

-c\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 compile to object (.o) only, do not link
\n-S\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 compile to assembly (.s) only, do not link
\n-fsource-asm
\nproduce assembly file with optional source annotations (requires -S)
\n-f[no-]verbose-asm
\nproduce assembly file with compiler comments (DEFAULT) (requires -S)
\n-fcode-asm
\nproduce assembly file with optional code annotations (requires -S)
\n-use-msasm
\nsupport Microsoft* style assembly language insertion using MASM style
\nsyntax
\n-o <file>
\nname output file
\n-g\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 produce symbolic debug information in object file (implies -O0 when
\nanother optimization option is not explicitly set)
\n-debug [keyword]
\nenable debug information and control output of enhanced debug
\ninformation
\n\u542f\u7528\u8c03\u8bd5\u4fe1\u606f\u548c\u63a7\u5236\u8f93\u51fa \u589e\u5f3a\u8c03\u8bd5\u8f93\u51fa\u4fe1\u606f
\nkeywords:\u00a0 all, full, minimal, none, [no]inline-debug-info
\n[no]variable-locations, [no]semantic-stepping,
\nextended
\nparallel
\n[no]expr-source-pos
\n-g0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 disable generation of symbolic debug information
\n-gdwarf-2
\nenable generation of debug information using the DWARF2 format
\n-ftrapuv\u00a0 trap uninitialized variables
\n-map-opts
\nenable option mapping tool
\n-print-multi-lib
\nprint information about libraries being used
\n-pch-create <file>
\ncreate precompiled header file
\n-pch-use <file>
\nuse precompiled header file
\n-pch\u00a0\u00a0\u00a0\u00a0\u00a0 enable automatic precompiled header file creation\/usage
\n-pch-dir <dir>
\nname precompiled header directory<\/p>\n

Preprocessor
\n\u9884\u5904\u7406
\n------------<\/p>\n

-A<name>[<val>]
\ncreate an assertion <name> having value <val>
\n-C\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 do not strip comments
\n-D<name>[=<text>]
\ndefine macro
\n-U<name>\u00a0 remove predefined macro
\n-E\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 preprocess to stdout
\n-EP\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 preprocess to stdout, omitting #line directives
\n-P\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 preprocess to file, omitting #line directives
\n-I<dir>\u00a0\u00a0 add directory to include file search path
\n-idirafter<dir>
\nadd directory to the second include file search path (after -I)
\n-isystem<dir>
\nadd directory to the start of the system include path
\n-X, -nostdinc
\nremove standard directories from include file search path
\n-nostdinc++
\nremove standard C++ directories from include file search path
\n-iprefix <prefix>
\nuse <prefix> with -iwithprefix as a prefix
\n-iwithprefix <dir>
\nappend <dir> to the prefix passed in by -iprefix and put it on the
\ninclude search path at the end of the include directories
\n-iwithprefixbefore <dir>
\nsimilar to -iwithprefix except the include directory is placed in the
\nsame place as -I command line include directories
\n-iquote <dir>
\nadd directory to the front of the include file search path for files
\nincluded with quotes, but not brackets
\n-imacros <file>
\ntreat <file> as an #include file, but throw away all preprocessing
\nwhile macros defined remain defined
\n-H\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 print include file order
\n-B<prefix>
\nfind libraries, headers and executables in <prefix>
\n-M\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 generate makefile dependency information
\n-MM\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 similar to -M, but do not include system header files
\n-MG\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 similar to -M, but treat missing header files as generated files
\n-MD\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 preprocess and compile, generating output file containing dependency
\ninformation ending with extension .d
\n-MMD\u00a0\u00a0\u00a0\u00a0\u00a0 similar to -MD, but do not include system header files
\n-MF<file>
\ngenerate makefile dependency information in file (must specify -M
\nor -MM)
\n-MP\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 add a phony target for each dependency
\n-MT<target>
\nchange the default target rule for dependency generation
\n-MQ<target>
\nsame as -MT, but quotes special Make characters
\n-dM\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 output macro definitions in effect after preprocessing (use with -E)
\n-dD\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 same as -dM, but output #define directives in preprocessed source
\n-dN\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 same as -dD, but #define directives contain only macro names
\n-gcc\u00a0\u00a0\u00a0\u00a0\u00a0 Predefine the \"__GNUC__\", \"__GNUC_MINOR__\", and
\n\"__GNUC_PATCHLEVEL__\" macros (DEFAULT)
\n-no-gcc\u00a0\u00a0 Do not predefine GNUC macros listed in -gcc mode. Warning: can
\nprevent correct system header compilation, see -gcc-sys
\n-gcc-sys\u00a0 same as -no-gcc, except that the GNU macros are defined only while
\npreprocessing the system include headers
\n-no-icc\u00a0\u00a0 do not predefine the \"__ICC\" and \"__INTEL_COMPILER\" macros.
\nWarning: can prevent correct Intel header compilation
\n-pragma-optimization-level=[Intel|GCC]
\nprocess #pragma optimize using Intel (DEFAULT) or GCC syntax<\/p>\n

Component Control
\n\u63a7\u5236\u7ec4\u4ef6
\n-----------------<\/p>\n

-Qoption,<tool>,<opts>
\npass options <opts> to tool specified by <tool>
\n-Qlocation,<tool>,<dir>
\nset <dir> as the location of tool specified by <tool>
\n-Qinstall <dir>
\nset <dir> as root of compiler installation<\/p>\n

Language
\n\u8bed\u8a00
\n--------<\/p>\n

-std=<std>
\nenable language support for <std>, as described below
\nc99\u00a0\u00a0 enable C99 support for C programs
\nc++0x enable preliminary support for some C++0x features
\n-x <type>
\nall source files found subsequent to -x <type> will be recognized
\nas one of the following types:
\nc\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - C source file
\nc++\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - C++ source file
\nc-header\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - C header file
\ncpp-output\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - C pre-processed file
\nc++-cpp-output\u00a0\u00a0\u00a0\u00a0 - C++ pre-processed file
\nassembler\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - assembly file
\nassembler-with-cpp - assembly file that needs to be preprocessed
\nnone\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - revert to original file extension
\n-ansi\u00a0\u00a0\u00a0\u00a0 equivalent to GNU -ansi
\n-strict-ansi
\nstrict ANSI conformance dialect
\n-Zp[n]\u00a0\u00a0\u00a0 specify alignment constraint for structures where
\nn=1,2,4,8,16. 16 is the default
\n-fsyntax-only
\nperform syntax and semantic checking only (no object file produced)
\n-fpermissive
\nallows extensions for some non-conformant code
\n-funsigned-char
\nchange default char type to unsigned
\n-fshort-enums
\nallocate as many bytes as needed for enumerated types
\n-f[no-]unsigned-bitfields
\nchange default bitfield type to unsigned
\n-fno-rtti
\ndisable RTTI support
\n-fno-implicit-templates
\nnever emit code for non-inline templates which are instantiated
\nimplicitly; only emit code for explicit instantiations
\n-fno-implicit-inline-templates
\ndo not emit code for implicit instantiations of inline templates
\n-ftemplate-depth-[n]
\ncontrol the depth in which recursive templates are expanded
\n-[no]align
\nanalyze and reorder memory layout for variables and arrays
\n-[no-]restrict
\nenable\/disable the 'restrict' keyword for disambiguating pointers
\n-[no-]early-template-check
\nenable\/disable (DEFAULT) semantic checking of function template
\nprototypes (before instantiation). Requires -gcc-version=340 or
\nlater
\n-[no-]check-uninit
\ncheck for uninitialized variables
\n-export\u00a0\u00a0 enable the export template feature
\n-export-dir <dir>
\nadd directory to export template search path
\n-help-pragma
\noutput supported pragmas and their usage syntax<\/p>\n

Compiler Diagnostics
\n\u7f16\u8bd1\u8bca\u65ad
\n--------------------<\/p>\n

-w\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 disable all warnings
\n-w<n>\u00a0\u00a0\u00a0\u00a0 control diagnostics
\nn = 0\u00a0\u00a0\u00a0 display errors (same as -w)
\nn = 1\u00a0\u00a0\u00a0 display warnings and errors (DEFAULT)
\nn = 2\u00a0\u00a0\u00a0 display remarks, warnings and errors
\n-Wbrief\u00a0\u00a0 print brief one-line diagnostics
\n-Werror\u00a0\u00a0 force warnings to be reported as errors
\n-Werror-all
\nforce warnings and remarks to be reported as errors
\n-Wall\u00a0\u00a0\u00a0\u00a0 enable all warnings
\n-Wcheck\u00a0\u00a0 enable more strict diagnostics
\n-Weffc++\u00a0 enable effective C++ diagnostic warnings
\n-wo<L1>[,<L2>,...]
\nissue diagnostics L1 through LN only once
\n-W[no-]fatal-errors
\nends compilation after the first error encountered
\n-W[no-]format
\nenable argument checking for calls to printf, scanf, etc
\n-W[no-]format-security
\nwarn for uses of format strings and specifiers in printf,
\nscanf, etc that may introduce security problems
\n-W[no-]missing-declarations
\nwarn for global functions and variables without prior declaration
\n-W[no-]missing-prototypes
\nwarn for missing prototypes
\n-W[no-]strict-aliasing
\nwarn for code that might violate the optimizer's strict aliasing
\nrules. Warnings are issued only when using -fstrict-aliasing or
\n-ansi-alias.
\n-W[no-]strict-prototypes
\nwarn for functions declared or defined without specified argument
\ntypes
\n-W[no-]pointer-arith
\nwarn for questionable pointer arithmetic
\n-W[no-]uninitialized
\nwarn if a variable is used before being initialized
\n-Winline\u00a0 enable inline diagnostics
\n-W[no-]deprecated
\nprint warnings related to deprecated features
\n-W[no-]abi
\nwarn if generated code is not C++ ABI compliant
\n-Wcontext-limit=<n>
\nset maximum number of template instantiation contexts shown in
\ndiagnostic
\n-Wcast-qual
\nwarn if cast is used to override pointer type qualifier
\n-W[no-]unused-function
\nwarn if declared function is not used
\n-W[no-]unknown-pragmas
\nwarn if an unknown #pragma directive is used (DEFAULT)
\n-W[no-]main
\nwarn if return type of main is not expected
\n-W[no-]comment[s]
\nwarn when\u00a0\u00a0 comment
\n-W[no-]conversion
\nwarn for implicit conversions that may alter a value
\n-W[no-]return-type
\nwarn when a function uses the default int return type and warn when
\na return statement is used in a void function
\n-W[no-]extra-tokens
\nwarn about extra tokens after prepreprocessor directives
\n-W[no-]pragma-once
\nwarn about the use of #pragma once
\n-W[no-]shadow
\nwarn when a variable declaration hides a previous declaration
\n-Woverloaded-virtual
\nwarn when a function declaration hides virtual functions from a base
\nclass
\n-W[no-]trigraphs
\nwarn about the recognition and conversion of trigraphs
\n-W[no-]multichar
\nwarn if a multicharacter constant ('ABC') is used
\n-W[no-]overflow
\nwarn for operations that could result in integer overflow
\n-Wwrite-strings
\nIssues a diagnostic message if const char * is converted
\nto (non-const) char *.
\n-Wp64\u00a0\u00a0\u00a0\u00a0 print diagnostics for 64-bit porting
\n-W[no-]shorten-64-to-32
\nwarn for values implicitly converted from a 64-bit to a 32-bit type.
\nSimilar to -Wp64
\n-[no]traceback
\nspecify whether the compiler generates data to allow for source file
\ntraceback information at runtime (only to be used when linking with
\nFortran programs)
\n-noerror-limit
\nset no maximum number on error-level or fatal-level error messages
\n-diag-enable <v1>[,<v2>,...]
\nenable the specified diagnostics or diagnostic groups
\n-diag-disable <v1>[,<v2>,...]
\ndisable the specified diagnostics or diagnostic groups
\nwhere <vN> may be individual diagnostic numbers or group names.
\nwhere group names include:
\nsc[n]\u00a0\u00a0\u00a0\u00a0\u00a0 - perform source code analysis: n=1 for critical
\nerrors, n=2 for all errors and n=3 for all errors
\nand warnings
\nsc-include - perform source code analysis on include files
\nsc-parallel[n] - perform analysis of parallelization in
\nsource code: n=1 for critical errors,
\nn=2 for errors, n=3 for all errors and
\nwarnings
\nthread\u00a0\u00a0\u00a0\u00a0 - enable diagnostics to aid in thread-enabling source
\nport-win\u00a0\u00a0 - enable diagnostics for GNU extensions that may
\ncause errors when porting to Windows
\n-diag-error <v1>[,<v2>,...]
\noutput the specified diagnostics or diagnostic groups as errors
\n-diag-warning <v1>[,<v2>,...]
\noutput the specified diagnostics or diagnostic groups as warnings
\n-diag-remark <v1>[,<v2>,...]
\noutput the the specified diagnostics or diagnostic groups as remarks
\n-diag-dump
\ndisplay the currently enabled diagnostic messages to stdout or to a
\nspecified diagnostic output file.
\n-diag-file[=<file>]
\n<file> where diagnostics are emitted to.\u00a0 Not specifying this causes
\nmessages to be output to stderr
\n-diag-file-append[=<file>]
\n<file> where diagnostics are emitted to. When <file> already exists,
\noutput is appended to the file
\n-[no-]diag-id-numbers
\nenable(DEFAULT)\/disable the diagnostic specifiers to be output in
\nnumeric form
\n-diag-error-limit <num>
\nspecify the maximum number of errors emitted
\n-diag-once <v1>[,<v2>,...]
\nissue diagnostics v1 through vN only once<\/p>\n

Miscellaneous
\n\u6df7\u6742
\n-------------<\/p>\n

-V\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 display compiler version information
\n\u663e\u793a\u7f16\u8bd1\u5668\u7248\u672c\u4fe1\u606f
\n-dumpversion
\ndisplay the compiler version number only
\n\u663e\u793a\u7f16\u8bd1\u5668\u7248\u672c\u53f7
\n-dumpmachine
\ndisplay the target machine only
\n\u663e\u793a\u76ee\u6807\u673a\u5668
\n--version
\ndisplay GCC style version information
\n\u663e\u793a\u7248\u672c\u4fe1\u606f(GCC \u6837\u5f0f)
\n-[no-]sox
\nenable\/disable(DEFAULT) saving of compiler options and version in
\nthe executable
\n\u542f\u7528\/\u7981\u7528 \u7f16\u8bd1\u5668\u53c2\u6570\u548c\u7248\u672c\u4fdd\u5b58\u529f\u80fd
\n-save-temps
\nstore the intermediate files in current directory and name them
\nbased on the source file.\u00a0 Only saves files that are generated by
\ndefault
\n-dryrun\u00a0\u00a0 show driver tool commands but do not execute tools
\n-v\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 show driver tool commands and execute tools
\n-[no-]multibyte-chars
\nprovide support for multi-byte characters
\n-multiple-processes[=<n>]
\ncreate multiple processes that can be used to compile large numbers
\nof source files at the same time<\/p>\n

Data
\n\u6570\u636e
\n----<\/p>\n

-Zp[n]\u00a0\u00a0\u00a0 specify alignment constraint for structures (n=1,2,4,8,16
\n-Zp16 DEFAULT)
\n-[no]align
\nanalyze and reorder memory layout for variables and arrays
\n-fminshared
\nCompilation is for the main executable. Absolute addressing can be
\nused and non-position independent code generated for symbols that
\nare at least protected
\n-fcommon\u00a0 Enables the compiler to treat common variables as if they were
\ndefined.\u00a0 That in turn allows the use of gprel addressing of common
\ndata variables.\u00a0 -fno-common disables
\n-freg-struct-return
\nreturn struct and union values in registers when possible
\n-fstack-security-check
\nenable overflow security checks.
\n-f[no-]stack-security-check disables (DEFAULT)
\n-fstack-protector
\nenable stack overflow security checks.
\n-f[no-]stack-protector disables (DEFAULT)
\n-fpic, -fPIC
\ngenerate position independent code (-fno-pic\/-fno-PIC is DEFAULT)
\n-fpie, -fPIE
\ngenerate position independent code that will be linked into an
\nexecutable (-fno-pie\/-fno-PIE is DEFAULT)
\n-[no-]global-hoist
\nenable(DEFAULT)\/disable external globals are load safe
\n-f[no-]keep-static-consts
\nenable\/disable(DEFAULT) emission of static const variables even
\nwhen not referenced
\n-fpack-struct
\npack structure members together
\n-f[no-]math-errno
\nset ERRNO after calling standard math library functions
\n-no-bss-init
\ndisable placement of zero-initialized variables in BSS (use DATA)
\n-falign-functions=[2|16]
\nalign the start of functions on a 2 (DEFAULT) or 16 byte boundary
\n-falign-functions
\nalign the start of functions to an optimal machine-dependent value.
\n-fno-align-functions (DEFAULT) aligns on a 2-byte boundary
\n-falign-stack=<mode>
\nstack alignment method to use on entry to routines
\ndefault\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - use default heuristics
\nmaintain-16-byte - do not assume any specific alignment, but
\nmaintain alignment in case the stack is already
\naligned
\nassume-16-byte\u00a0\u00a0 - assume the stack is aligned on 16-byte
\nboundaries
\n-fvisibility=[extern|default|protected|hidden|internal]
\nGlobal symbols (data and functions) will get the visibility
\nattribute given by default. Symbol visibility attributes explicitly
\nset in the source code or using the symbol visibility attribute
\nfile options will override the -fvisibility setting
\n-fvisibility-extern=<file>
\nSpace separated symbols listed in the <file> argument will geti
\nvisibility set to extern
\n-fvisibility-default=<file>
\nSpace separated symbols listed in the <file> argument will get
\nvisibility set to default
\n-fvisibility-protected=<file>
\nSpace separated symbols listed in the <file> argument will get
\nvisibility set to protected
\n-fvisibility-hidden=<file>
\nSpace separated symbols listed in the <file> argument will get
\nvisibility set to hidden
\n-fvisibility-internal=<file>
\nSpace separated symbols listed in the <file> argument will get
\nvisibility set to internal
\n-fvisibility-inlines-hidden
\nmark inline member functions as hidden<\/p>\n

Compatibility
\n\u517c\u5bb9\u6027
\n-------------<\/p>\n

-gcc-name=<name>
\nname and location of gcc if not where expected
\n-gxx-name=<name>
\nname and location of g++ if not where expected
\n-gcc-version=<version>
\nspecify the <version> of gcc compatibility.\u00a0 Default value matches
\ngcc version installed.\u00a0 Major\/Minor versions listed but patch
\nlevels (i.e. 345) are permissible
\n320 - gcc 3.2.x compatibility
\n330 - gcc 3.3.x compatibility
\n340 - gcc 3.4.x compatibility
\n400 - gcc 4.0.x compatibility
\n410 - gcc 4.1.x compatibility
\n420 - gcc 4.2.x compatibility
\n-fabi-version=<val>
\ndirects the compiler to select a specific ABI implementation
\n0 - most recent ABI implementation
\n1 - g++ 3.2 compatible ABI implementation
\n2 - most conformant ABI implementation<\/p>\n

Linking\/Linker
\n\u8fde\u63a5\uff0c\u8fde\u63a5\u5668
\n--------------<\/p>\n

-L<dir>\u00a0\u00a0 instruct linker to search <dir> for libraries
\n-l<string>
\ninstruct the linker to link in the -l<string> library
\n-shared-intel
\nlink Intel provided libraries dynamically
\n-static-intel
\nlink Intel provided libraries statically
\nIntel \u63d0\u4f9b\u7684\u9759\u6001\u94fe\u63a5\u5e93<\/p>\n

-shared-libgcc
\nlibgcc\u7684\u52a8\u6001\u94fe\u63a5
\n-static-libgcc
\nlibgcc\u7684\u9759\u6001\u94fe\u63a5
\n-dynamic-linker<file>
\nselect dynamic linker other than the default
\n-no-cxxlib
\ndo not link in C++ runtime libraries
\n-cxxlib[=dir]
\nlink using C++ run-time libraries provided with gcc dir is an
\noptional top-level location for the gcc binaries and libraries
\n-nodefaultlibs
\ndo not use standard libraries when linking
\n-nostartfiles
\ndo not use standard startup files when linking
\n-nostdlib
\ndo not use standard libraries and startup files when linking
\n-static\u00a0\u00a0 prevents linking with shared libraries
\n-shared\u00a0\u00a0 produce a shared object
\n-Bstatic\u00a0 specify following libraries are linked statically
\n-Bdynamic
\nspecify following libraries are linked dynamically
\n-pthread\u00a0 use POSIX* thread library for multithreading support
\n-cxxlib-<mode>
\ntell the compiler which C++ run-time libraries to use
\nnostd - do not link in standard C++ library
\n-u <symbol>
\npretend the <symbol> is undefined
\n-T <file>
\ndirect linker to read link commands from <file>
\n-Xlinker <val>
\npass <val> directly to the linker for processing
\n-Wa,<o1>[,<o2>,...]
\npass options o1, o2, etc. to the assembler
\n-Wl,<o1>[,<o2>,...]
\npass options o1, o2, etc. to the linker for processing
\n-Wp,<o1>[,<o2>,...]
\npass options o1, o2, etc. to the preprocessor<\/p>\n

Linker Specific Options
\n\u8fde\u63a5\u5668\u6307\u5b9a\u53c2\u6570
\n-----------------------<\/p>\n

These options are specific to the linker.\u00a0 Details can be found in the linker
\ndocumentation and man page
\n-L<dir>
\n-T<arg>
\n-h<arg>
\n-u<arg>
\n-z<arg>
\n-i
\n-r
\n-s
\n-N
\n-Bsymbolic
\n-Bdynamic
\n-Bstatic<\/p>\n

Deprecated Options
\n\u5df2\u65e0\u6548\u7684\u53c2\u6570\uff08\u76f8\u5bf9\u65e9\u671f\u7248\u672c\uff09
\n------------------<\/p>\n

-Ob\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -inline-level=<n>
\n-cxxlib-gcc[=dir]\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -cxxlib[=dir]
\n-i-dynamic\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -shared-intel
\n-i-static\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -static-intel
\n-inline-debug-info\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -debug inline-debug-info
\n-mp\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -fp-model <arg>
\n-use-asm\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 no replacement
\n-A-\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -U<arg>
\n-prefetch\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -opt-prefetch
\n-prof-genx\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -prof-gen=srcpos
\n-openmp-lib=legacy\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -openmp-lib=compat
\n-xB\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -xSSE2 (i32 only)
\n-xK\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -xSSE (i32 only)
\n-axK\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 No replacement
\n-xW\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -msse2
\n-axW\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -msse2
\n-xN\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -xSSE2
\n-axN\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -axSSE2
\n-xP\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -xSSE3
\n-axP\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -axSSE3
\n-xT\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -xSSSE3
\n-axT\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -axSSSE3
\n-xS\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -xSSE4.1
\n-axS\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -axSSE4.1
\n-xO\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -msse3
\n-func-groups\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -prof-func-groups
\n-diag-enable sv<n>\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -diag-enable sc<n>
\n-diag-enable sv-include\u00a0 use -diag-enable sc-include
\n-diag-disable sv\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -diag-disable sc
\n-diag-sv\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -diag-enable sc
\n-diag-sv-error\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -diag-disable warning
\n-diag-sv-include\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -diag-enable sc-include
\n-diag-sv-level\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 No replacement
\n-diag-sv-sup\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -diag-disable <v1>[,<v2>,...]
\n-alias-args\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -fargument-alias
\n-c99\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -std=c99
\n-create-pch\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -pch-create
\n-fwritable-strings\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 no replacement
\n-Kc++\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -x c++
\n-no-alias-args\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -fargument-noalias
\n-use-pch\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -pch-use
\n-wn\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -diag-error-limit
\n-wd\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -diag-disable
\n-we\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -diag-error
\n-ww\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -diag-warning
\n-wr\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -diag-remark
\n-wo\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 use -diag-once<\/p>\n

-help [category]\u00a0\u00a0 print full or category help message<\/p>\n

Valid categories include
\nadvanced\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - Advanced Optimizations
\ncodegen\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - Code Generation
\ncompatibility\u00a0\u00a0 - Compatibility
\ncomponent\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - Component Control
\ndata\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - Data
\ndeprecated\u00a0\u00a0\u00a0\u00a0\u00a0 - Deprecated Options
\ndiagnostics\u00a0\u00a0\u00a0\u00a0 - Compiler Diagnostics
\nfloat\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - Floating Point
\nhelp\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - Help
\ninline\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - Inlining
\nipo\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - Interprocedural Optimization (IPO)
\nlanguage\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - Language
\nlink\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - Linking\/Linker
\nmisc\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - Miscellaneous
\nopt\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - Optimization
\noutput\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - Output
\npgo\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - Profile Guided Optimization (PGO)
\npreproc\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - Preprocessor
\nreports\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - Optimization Reports
\nopenmp\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 - OpenMP and Parallel Processing<\/p>\n

Copyright (C) 1985-2009, Intel Corporation.\u00a0 All rights reserved.
\n* Other brands and names are the property of their respective owners.<\/p>\n","protected":false},"excerpt":{"rendered":"

\u65e9\u671f\u5728\u5c1d\u8bd5icc \u7f16\u8bd1\u5668\u6d4b\u8bd5\u65f6\u987a\u4fbf\u5c06\u82f1\u6587\u6587\u6863\u505a\u4e86\u90e8\u5206\u4e2d\u6587\u7ffb\u8bd1\uff0c\u672f\u8bed\u592a\u5f3a\uff0c\u90e8\u5206\u7ffb\u8bd1 …<\/p>\n

\u7ee7\u7eed\u9605\u8bfb »<\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[3],"tags":[],"class_list":["post-59","post","type-post","status-publish","format-standard","hentry","category-linux"],"_links":{"self":[{"href":"https:\/\/www.sulabs.net\/index.php?rest_route=\/wp\/v2\/posts\/59","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.sulabs.net\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.sulabs.net\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.sulabs.net\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.sulabs.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=59"}],"version-history":[{"count":2,"href":"https:\/\/www.sulabs.net\/index.php?rest_route=\/wp\/v2\/posts\/59\/revisions"}],"predecessor-version":[{"id":742,"href":"https:\/\/www.sulabs.net\/index.php?rest_route=\/wp\/v2\/posts\/59\/revisions\/742"}],"wp:attachment":[{"href":"https:\/\/www.sulabs.net\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=59"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.sulabs.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=59"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.sulabs.net\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=59"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}