digitalmars.com
Last update Sun Mar 4 11:58:06 2018
Compiler & Tools Guide
Tools
BCC
CHMOD
CL
COFF2OMF
COFFIMPLIB
DMC
DIFF
DIFFDIR
DMAN
DUMP
DUMPOBJ
DUMPEXE
EXE2BIN
FLPYIMG
GREP
HC
HTOD
IMPLIB
LIB
LIBUNRES
MAKE
MAKEDEP
ME
OBJ2ASM
PATCHOBJ
RC
RCC
SC
SHELL
SMAKE
TOUCH
UNMANGLE
WHEREIS
Compiling
Compiling Code
C Implementation
C++ Implementation
C++0x Implementation
Language Extensions
Mixing Languages
Assembly Language
Inline Assembler
Optimizing Code
Numerics Programming
Regular Expressions
Acrtused
Pragmas
Precompiled Headers
Predefined Macros
Warning Messages
Error Messages
Runtime Messages
Linking
Optlink
Switches
Module Definition Files
Operation and Design
Error Messages
Win32 Programming
Win32 Programming
DOS and Win16 Programming
Memory Models
16 Bit Pointer Types and Type Modifiers
Handle Pointers
DOS
DOS 32 (DOSX)
Win16
Win16 DLLs
Win16 Prolog/Epilog
Virtual Memory For 640Kb DOS
C/C++ Extensions
Contract Programming
__debug statement
__debug declaration
Dynamic Profiling
Embedding C in HTML
Porting to DMC++
Switching to DMC++
from Microsoft
from Borland
Porting Guide
<h1><a name="ctgInlineAsm">Inline Assembler</a> </h1> The compiler includes a powerful inline assembler. With it, assembly language instructions can be used directly in C and C++ source programs without requiring a separate assembler program. Assembly language enables optimizing critical functions, interfacing to the BIOS, operating system and special hardware, and access capabilities of the processor that are not available from C++. <p> It supports both 16-bit and 32-bit code generation in all memory models. <h2>What's in This Chapter</h2> <ul> <li> Basic features of the inline assembler. <li> The ASM statement and the ASM block. <li> Using ASM registers. <li> Calling C and C++ from assembly language. <li> Registers and opcodes. </ul> <h2>Advantages of writing inline assembly language functions</h2> Use assembly language functions to: <ul> <li> Create a subroutine that executes as quickly as possible <li> Provide an interface to functions compiled with another compiler <li> Acess capabilities of the CPU and the native instruction set that are not available from C++ <li> Interface to specialized hardware <li> Write code where specific instruction selection and ordering is critical </ul> <h3>The asm Statement</h3> The asm statement invokes the assembler. Use this statement wherever a C or C++ statement is legal. You can use asm in any of three ways. <p> The first example shows asm followed simply by an assembly instruction: <pre>asm mov AH,2 asm mov DL,7 asm int 21H </pre> <p>The second example shows asm followed by a set of assembly instructions enclosed by braces. An empty set of braces may follow the directive. </p> <pre>asm { mov AH,2 mov DL,7 int 21H } </pre> <p>Because the asm statement is a statement separator, assembly instructions can appear on the same line: </p> <pre>asm mov AH,2 asm mov DL,7 asm int 21H </pre> <p>The three previous examples generate identical code. But enclosing assembly language in braces, as in the second example, has the advantage of setting the assembly language off from the surrounding C++ code and avoids repeating the asm statement. </p> <p>No assembler instruction can continue onto a second line. Use the form of the last example primarily for writing macros, which must be one line long after expansion. </p> <h4>Note</h4> The Digital Mars C++ asm statement emulates the Borland asm statement. The _asm and __asm statements emulate the Microsoft _asm and __asm statements. <h3>The ASM Block</h3> A series of assembler instructions enclosed by braces following the asm keyword are called an "ASM block." Unlike C++ blocks, ASM blocks do not affect the scope of variables. <h4>Restrictions on using C and C++ in an ASM block</h4> An ASM block can use the following C and C++ language elements: <ul> <li> Symbols, including labels, variables, and function names. <li> Constants, including symbolic constants and enum members. <li> Macros and preprocessor directives. <li> Comments delimited by /**/ symbols or defined by // symbols. In Microsoft-compatible mode, semicolons (;) can also be used to delimit comments. <li> Type names (where a MASM type would be legal). <li> Type names, including declarations of structs and pointers. <li> C-style type casts. </ul> <h4>Note</h4> Microsoft and Borland inline assemblers do not support type casts. <p> Inline assembly instructions within C or C++ statements can refer to C or C++ variables by name. <h3>MASM-Style hexadecimal constants</h3> Support for MASM-style hexadecimal constants provides easy conversion of MASM-style source code. The constants take the form: <pre>digit {hex_digit} ('H'| 'h') </pre> <p>You cannot use hexadecimal constants if the -A (for ANSI compatibility) option is used. </p> <h3>C and C++ operators in an ASM block</h3> An ASM block cannot use operators specific to C and C++, such as the left-shift (<<) operator. <p> You can use operators common to C, C++, and MASM within an ASM block but interpret them as assembly language operators. C and C++ interpret brackets ([]) as enclosing array subscripts and scale them to the size of an array element. But within an asm statement, C and C++ interpret brackets as the MASM index operator, which adds an unscaled byte offset to any adjacent operand. <p> In Microsoft-compatible mode, the semicolon delimits comments, as in MASM. <h3>Assembly language in an asm statement</h3> In common with other assemblers, the inline assembler accepts any instruction that is legal in MASM. The following are some of the assembly language features of the asm statement: <ul> <li> Expressions. Inline assembly code can use any MASM expression. A MASM expression is any combination of operands and operators that evaluates to a single value or address. <li> Data directives and operators. An asm statement can define data objects in the code segment with the MASM directives DB, DW, and DQ. The MASM directives DT, DF, STRUC, and RECORD, and the operators DUP, THIS, WIDTH, and MASK are not accepted. <li> Macros. An asm statement can use C preprocessor macros even though the inline assembler is not a macro assembler and does not support MASM macro directives. </ul> <h3>Other restrictions on C and C++ symbols</h3> An asm statement can reference any C or C++ variable name, function, or label in scope, provided those names are not symbolic constants. However, you cannot call a C++ member function from within an asm statement. <p> Prototype the functions referenced in an asm statement before using them in programs. This lets the compiler distinguish them from names and labels. <p> Each assembly language instruction can contain a single C or C++ symbol. <p> C or C++ symbols within an ASM block must not have the same spelling as an asm reserved word. <p> The inline assembler allows structure or union tags in asm statements, but only as qualifiers of references to members of the structure or union. <h3>Accessing C or C++ data in an asm statement</h3> In general, instructions in an asm statement can reference any symbol in scope where the statement appears. The following statement loads the AX register with the value of var, a C variable in scope: <pre>asm mov AX,var </pre> <p>An asm statement can reference a uniquely named member of a class, structure, or union without specifying the variable name or type before the period operator. But if the member name is not unique, you must specify a variable or type name before the period operator. If two structure types have a member name in common, as in this example: </p> <pre>struct first_type { char *mold; int common_name; }; struct second__type { char *mildew; long common_name; int unique_name; }; </pre> <p>then qualify the reference to <em>common_name</em> with the tag name:</p> <pre>asm mov [bx]first_type.common_name,10 </pre> <p>You need not qualify a reference to a unique member name. In the following example, unique_name is an anonymous structure member because it is a member of first_type. </p> <pre>asm mov [bx].unique_name,10 </pre> <p>This statement generates the same instruction whether or not a qualifying name or type is present. For more information, see the section "Making anonymous references to structure members" later in this chapter. </p> <h3>Functions in inline assembly language</h3> Because ASM blocks do not require separate source file assembly steps, writing a function using ASM blocks is easier than using a separate assembler. In addition, the compiler generates function prolog and epilog code. <p> The expon2 function is an example of a function written in inline assembly language: <pre>int expon2(int num, int power) { asm { mov AX,num // get first argument mov CX,power // get second argument shl AX,CL // AX = AX * (2 to the power of CL) } } </pre> <p>An inline function refers to its arguments by name and may appear in the same source file as the callers of the function. </p> <p>Refer to <a href="/index.cgi/contrast/https://digitalmars.com/ctg/ctgAsm.html">Using Assembly Language Functions</a> for a description of the register stacks used by inline assembly instructions. </p> <h3>Making anonymous references to structure members</h3> You can make anonymous references to members of a given structure, as in the following: <pre>struct x { int i; int j; int k; } foo; </pre> <p>You can refer to these members i, j, and k anonymously, for example, with the assembly instruction: </p> <pre>asm { mov BX,4 mov AX,foo[BX]; Refers to member j of foo } </pre> <h3>Using register variables</h3> <p>Digital Mars C++ supports register variables. Register variables are useful with inline assembly. If asm statements place results in registers, you can use register variables to access those values. </p> <p>For more information see "Using Register Variables" in <a href="/index.cgi/contrast/https://digitalmars.com/ctg/ctgAsm.html">Using Assembly Language Functions</a>. </p> <h3>Using the __LOCAL_SIZE symbol</h3> When using the inline assembler, the special symbol __LOCAL_SIZE expands to the number of bytes used by all local symbols. __LOCAL_SIZE is useful in combination with __declspec(naked), as __LOCAL_SIZE is the amount of space to reserve on the stack. <p> For example: <pre>__declspec(naked) int test() { int x, y, z; _asm { push BP mov BP,SP sub SP,__LOCAL_SIZE mov BX,__LOCAL_SIZE[BP] mov BX,__LOCAL_SIZE+2[BP] mov AX,__LOCAL_SIZE mov AX,__LOCAL_SIZE+2 } _AX = x + y + z; _asm { mov SP,BP pop BP ret } } </pre> <h3>Using ASM Registers</h3> The asm statement alters the registers outside the programmer's explicit assembly language instructions. Registers contain whatever values the normal control flow leaves in them at the point of the asm statement. <h3>For 16-bit memory models</h3> You do not need to preserve the following registers when writing inline assembly language: AX, BX, CX, DX, SI, DI, ES, and flags (other than DF). <p> C and C++ do not expect these registers to be maintained between statements, but they do preserve the following registers: CS, DS, SS, SP and BP. <p> <h4>Note</h4> The compiler does not use registers to hold register variables for functions containing inline assembly code. <h3>For 32-bit memory models</h3> Functions can change the values in the EAX, ECX, EDX, ESI, EDI registers. <p> Functions must preserve the values in the EBX, ESI, EDI, EBP, ESP, SS, CS, DS registers (plus ES and GS for the NT memory model). <p> Always set the direction flag to forward. <p> To maximize speed on 32-bit buses, make sure data aligns along 32-Function return values <ul> <li> For 16-bit models. If the return value for a function is short (a char, int, or near pointer) store it in the AX register, as in the previous example, expon2. If the return value is long, store the high word in the DX register and the low word in AX. To return a longer value, store the value in memory and return a pointer to the value. <li> For 32-bit models. Return near pointers, ints, unsigned ints, chars, shorts, longs and unsigned longs in EAX. 32-bit models return far pointers in EDX, EAX, where EDX contains the segment and EAX contains the offset. <li> When C linkage is in effect. Floats are returned in EAX and doubles in EDX, EAX, where EDX contains the most significant 32 bits and EAX the least significant. <li> When C++ linkage is in effect. The compiler creates a temporary copy on the stack and returns a pointer to it. </ul> <h3>Function return values</h3> <ul> <li> For 16-bit memory models. If the return value for a function is short (a char, int, or near pointer) store it in the AX register, as in the previous example, expon2. If the return value is long, store the high word in the DX register and the low word in AX. To return a longer value, store the value in memory and return a pointer to the value. <li> For 32-bit models. Return near pointers, ints, unsigned ints, chars, shorts, longs, and unsigned longs in EAX. 32-bit models return far pointers in EDX, EAX, where EDX contains the segment and EAX contains the offset. <li> When C linkage is in effect. Floats are returned in EAX and doubles in EDX, EAX, where EDX contains the most significant 32 bits and EAX the least significant. <li> When C++ linkage is in effect. The compiler creates a temporary copy on the stack and returns a pointer to it. </ul> <h3>Interfacing to a member function</h3> The easiest way to interface an assembly language routine to a class member function is to provide a C wrapper function that can be called from the assembly language routine. <p> An alternative is to write the member function in C++ and compile it with normal out-of-line member functions. <h3>Calling C Functions from an ASM Block</h3> C functions, including C library functions, can be called from within the asm block, as in the following example: <pre>#include <stdio.h> char format [] = "% s %s %s \n"; char alas[] = "Alas,"; char poor[] = "poor"; char Yorick[] = "Yorick!"; void main(void) { asm { mov AX, offset Yorick push AX mov AX, offset poor push AX mov AX, offset Alas push AX mov AX, offset format push AX call printf } } </pre> <p>Simply push the needed arguments from right to left before calling the function, since function arguments are passed on the stack. To print the message, the example pushes pointers to the three strings, formats them, and then calls printf. </p> <h3>Calling C++ functions</h3> An ASM block can call only global C++ functions that are not overloaded because the types of the arguments are unknown. The compiler issues an error if an ASM block calls an overloaded global C++ function or a C member function. <p> You can also call a function declared with extern "C" linkage from an asm statement within a C++ program, because all the standard header files declare the library functions to have extern "C" linkage. <h3>Defining ASM blocks as C macros</h3> A C++ macro is a convenient way to insert assembly language into source code. But, because a macro expands into a single, logical line, take care when writing them. <p> If the macro expands into multiple instructions, enclose the instructions in an ASM block. The asm statement must precede each instruction. Also separate comments from code with /**/ characters rather than //. Unless you take these precautions, the compiler can be confused by C or C++ statements to the left or right of the assembly code or interpret instructions as comments when the macro becomes a single line. Without the closing brace, the compiler cannot tell where the assembly language ends. <p> <strong>Warning</strong>: Do not use double-slash (//) characters within a macro. The compiler terminates the macro when it sees a double-slash. <p> An ASM block written as a macro can accept arguments but, unlike a C macro, it cannot return values. But some MASM macros can be written as macros for C. The following MASM macro sets a video page to the value specified in the argument page: <pre>findpage MACRO page mov AH, 5 MOV AL,page int 10h ENDM </pre> <p>The following C macro does the same thing: </p> <pre>#define findpage(page) asm \ { \ asm mov AH,5 \ asm mov AL,page \ asm int 10h \ } </pre> <h2>Registers</h2> The following registers are supported. Register names are in upper or lower case. <dl><dl> <dt><strong>AL</strong>, <strong>AH</strong>, <strong>AX</strong>, <strong>EAX</strong> <dt><strong>BL</strong>, <strong>BH</strong>, <strong>BX</strong>, <strong>EBX</strong> <dt><strong>CL</strong>, <strong>CH</strong>, <strong>CX</strong>, <strong>ECX</strong> <dt><strong>DL</strong>, <strong>DH</strong>, <strong>DX</strong>, <strong>EDX</strong> <dt><strong>BP</strong>, <strong>EBP</strong> <dt><strong>SP</strong>, <strong>ESP</strong> <dt><strong>DI</strong>, <strong>EDI</strong> <dt><strong>SI</strong>, <strong>ESI</strong> <dt><strong>ES</strong>, <strong>CS</strong>, <strong>SS</strong>, <strong>DS</strong>, <strong>GS</strong>, <strong>FS</strong> <dt><strong>CR0</strong>, <strong>CR2</strong>, <strong>CR3</strong>, <strong>CR4</strong> <dt><strong>DR0</strong>, <strong>DR1</strong>, <strong>DR2</strong>, <strong>DR3</strong>, <strong>DR6</strong>, <strong>DR7</strong> <dt><strong>TR3</strong>, <strong>TR4</strong>, <strong>TR5</strong>, <strong>TR6</strong>, <strong>TR7</strong> <dt><strong>ST</strong> <dt><strong>ST(0)</strong>, <strong>ST(1)</strong>, <strong>ST(2)</strong>, <strong>ST(3)</strong>, <strong>ST(4)</strong>, <strong>ST(5)</strong>, <strong>ST(6)</strong>, <strong>ST(7)</strong> <dt><strong>MM0</strong>, <strong>MM1</strong>, <strong>MM2</strong>, <strong>MM3</strong>, <strong>MM4</strong>, <strong>MM5</strong>, <strong>MM6</strong>, <strong>MM7</strong> <dt><strong>XMM0</strong>, <strong>XMM1</strong>, <strong>XMM2</strong>, <strong>XMM3</strong>, <strong>XMM4</strong>, <strong>XMM5</strong>, <strong>XMM6</strong>, <strong>XMM7</strong> </dl></dl> <h2>Opcodes</h2> The following instructions are supported. Opcode names are in upper or lower case. <div class="naked_tr"> aaa aad aam aas adc </div><div class="naked_tr"> add addpd addps addsd addss </div><div class="naked_tr"> and andnpd andnps andpd andps </div><div class="naked_tr"> arpl bound bsf bsr bswap </div><div class="naked_tr"> bt btc btr bts call </div><div class="naked_tr"> cbw cdq clc cld clflush </div><div class="naked_tr"> cli clts cmc cmova cmovae </div><div class="naked_tr"> cmovb cmovbe cmovc cmove cmovg </div><div class="naked_tr"> cmovge cmovl cmovle cmovna cmovnae </div><div class="naked_tr"> cmovnb cmovnbe cmovnc cmovne cmovng </div><div class="naked_tr"> cmovnge cmovnl cmovnle cmovno cmovnp </div><div class="naked_tr"> cmovns cmovnz cmovo cmovp cmovpe </div><div class="naked_tr"> cmovpo cmovs cmovz cmp cmppd </div><div class="naked_tr"> cmpps cmps cmpsb cmpsd cmpss </div><div class="naked_tr"> cmpsw cmpxch8b cmpxchg comisd comiss </div><div class="naked_tr"> cpuid cvtdq2pd cvtdq2ps cvtpd2dq cvtpd2pi </div><div class="naked_tr"> cvtpd2ps cvtpi2pd cvtpi2ps cvtps2dq cvtps2pd </div><div class="naked_tr"> cvtps2pi cvtsd2si cvtsd2ss cvtsi2sd cvtsi2ss </div><div class="naked_tr"> cvtss2sd cvtss2si cvttpd2dq cvttpd2pi cvttps2dq </div><div class="naked_tr"> cvttps2pi cvttsd2si cvttss2si cwd cwde </div><div class="naked_tr"> da daa das db dd </div><div class="naked_tr"> de dec df di div </div><div class="naked_tr"> divpd divps divsd divss dl </div><div class="naked_tr"> dq ds dt dw emms </div><div class="naked_tr"> enter f2xm1 fabs fadd faddp </div><div class="naked_tr"> fbld fbstp fchs fclex fcmovb </div><div class="naked_tr"> fcmovbe fcmove fcmovnb fcmovnbe fcmovne </div><div class="naked_tr"> fcmovnu fcmovu fcom fcomi fcomip </div><div class="naked_tr"> fcomp fcompp fcos fdecstp fdisi </div><div class="naked_tr"> fdiv fdivp fdivr fdivrp feni </div><div class="naked_tr"> ffree fiadd ficom ficomp fidiv </div><div class="naked_tr"> fidivr fild fimul fincstp finit </div><div class="naked_tr"> fist fistp fisub fisubr fld </div><div class="naked_tr"> fld1 fldcw fldenv fldl2e fldl2t </div><div class="naked_tr"> fldlg2 fldln2 fldpi fldz fmul </div><div class="naked_tr"> fmulp fnclex fndisi fneni fninit </div><div class="naked_tr"> fnop fnsave fnstcw fnstenv fnstsw </div><div class="naked_tr"> fpatan fprem fprem1 fptan frndint </div><div class="naked_tr"> frstor fsave fscale fsetpm fsin </div><div class="naked_tr"> fsincos fsqrt fst fstcw fstenv </div><div class="naked_tr"> fstp fstsw fsub fsubp fsubr </div><div class="naked_tr"> fsubrp ftst fucom fucomi fucomip </div><div class="naked_tr"> fucomp fucompp fwait fxam fxch </div><div class="naked_tr"> fxrstor fxsave fxtract fyl2x fyl2xp1 </div><div class="naked_tr"> hlt idiv imul in inc </div><div class="naked_tr"> ins insb insd insw int </div><div class="naked_tr"> into invd invlpg iret iretd </div><div class="naked_tr"> ja jae jb jbe jc </div><div class="naked_tr"> jcxz je jecxz jg jge </div><div class="naked_tr"> jl jle jmp jna jnae </div><div class="naked_tr"> jnb jnbe jnc jne jng </div><div class="naked_tr"> jnge jnl jnle jno jnp </div><div class="naked_tr"> jns jnz jo jp jpe </div><div class="naked_tr"> jpo js jz lahf lar </div><div class="naked_tr"> ldmxcsr lds lea leave les </div><div class="naked_tr"> lfence lfs lgdt lgs lidt </div><div class="naked_tr"> lldt lmsw lock lods lodsb </div><div class="naked_tr"> lodsd lodsw loop loope loopne </div><div class="naked_tr"> loopnz loopz lsl lss ltr </div><div class="naked_tr"> maskmovdqu maskmovq maxpd maxps maxsd </div><div class="naked_tr"> maxss mfence minpd minps minsd </div><div class="naked_tr"> minss mov movapd movaps movd </div><div class="naked_tr"> movdq2q movdqa movdqu movhlps movhpd </div><div class="naked_tr"> movhps movlhps movlpd movlps movmskpd </div><div class="naked_tr"> movmskps movntdq movnti movntpd movntps </div><div class="naked_tr"> movntq movq movq2dq movs movsb </div><div class="naked_tr"> movsd movss movsw movsx movupd </div><div class="naked_tr"> movups movzx mul mulpd mulps </div><div class="naked_tr"> mulsd mulss neg nop not </div><div class="naked_tr"> or orpd orps out outs </div><div class="naked_tr"> outsb outsd outsw packssdw packsswb </div><div class="naked_tr"> packuswb paddb paddd paddq paddsb </div><div class="naked_tr"> paddsw paddusb paddusw paddw pand </div><div class="naked_tr"> pandn pavgb pavgw pcmpeqb pcmpeqd </div><div class="naked_tr"> pcmpeqw pcmpgtb pcmpgtd pcmpgtw pextrw </div><div class="naked_tr"> pinsrw pmaddwd pmaxsw pmaxub pminsw </div><div class="naked_tr"> pminub pmovmskb pmulhuw pmulhw pmullw </div><div class="naked_tr"> pmuludq pop popa popad popf </div><div class="naked_tr"> popfd por prefetchnta prefetcht0 prefetcht1 </div><div class="naked_tr"> prefetcht2 psadbw pshufd pshufhw pshuflw </div><div class="naked_tr"> pshufw pslld pslldq psllq psllw </div><div class="naked_tr"> psrad psraw psrld psrldq psrlq </div><div class="naked_tr"> psrlw psubb psubd psubq psubsb </div><div class="naked_tr"> psubsw psubusb psubusw psubw punpckhbw </div><div class="naked_tr"> punpckhdq punpckhqdq punpckhwd punpcklbw punpckldq </div><div class="naked_tr"> punpcklqdq punpcklwd push pusha pushad </div><div class="naked_tr"> pushf pushfd pxor rcl rcpps </div><div class="naked_tr"> rcpss rcr rdmsr rdpmc rdtsc </div><div class="naked_tr"> rep repe repne repnz repz </div><div class="naked_tr"> ret retf rol ror rsm </div><div class="naked_tr"> rsqrtps rsqrtss sahf sal sar </div><div class="naked_tr"> sbb scas scasb scasd scasw </div><div class="naked_tr"> seta setae setb setbe setc </div><div class="naked_tr"> sete setg setge setl setle </div><div class="naked_tr"> setna setnae setnb setnbe setnc </div><div class="naked_tr"> setne setng setnge setnl setnle </div><div class="naked_tr"> setno setnp setns setnz seto </div><div class="naked_tr"> setp setpe setpo sets setz </div><div class="naked_tr"> sfence sgdt shl shld shr </div><div class="naked_tr"> shrd shufpd shufps sidt sldt </div><div class="naked_tr"> smsw sqrtpd sqrtps sqrtsd sqrtss </div><div class="naked_tr"> stc std sti stmxcsr stos </div><div class="naked_tr"> stosb stosd stosw str sub </div><div class="naked_tr"> subpd subps subsd subss sysenter </div><div class="naked_tr"> sysexit test ucomisd ucomiss ud2 </div><div class="naked_tr"> unpckhpd unpckhps unpcklpd unpcklps verr </div><div class="naked_tr"> verw wait wbinvd wrmsr xadd </div><div class="naked_tr"> xchg xlat xlatb xor xorpd </div><div class="naked_tr"> xorps </div> <h3>Pentium 4 (Prescott) Opcodes Supported</h3> <div class="naked_tr"> addsubpd addsubps fisttp haddpd haddps </div><div class="naked_tr"> hsubpd hsubps lddqu monitor movddup </div><div class="naked_tr"> movshdup movsldup mwait </div> <h3>AMD Opcodes</h3> <div class="naked_tr"> pavgusb pf2id pfacc pfadd pfcmpeq </div><div class="naked_tr"> pfcmpge pfcmpgt pfmax pfmin pfmul </div><div class="naked_tr"> pfnacc pfpnacc pfrcp pfrcpit1 pfrcpit2 </div><div class="naked_tr"> pfrsqit1 pfrsqrt pfsub pfsubr pi2fd </div><div class="naked_tr"> pmulhrw pswapd </div> </div> <div id="footernav"> <a href="https://www.digitalmars.com/" title="digitalmars.com">Home</a> | <a href="/index.cgi/contrast/https://digitalmars.com/ctg/../rtl/rtl.html" title="Runtime Library Reference">Runtime Library</a> | <a href="/index.cgi/contrast/https://digitalmars.com/ctg/../ugr/ugr.html" title="IDDE Reference">IDDE Reference</a> | <a href="/index.cgi/contrast/https://digitalmars.com/ctg/../stl/index.html" title="Standard Template Library">STL</a> | <a href="https://www.digitalmars.com/advancedsearch.html" title="Search Digital Mars web site">Search</a> | <a href="https://www.digitalmars.com//download/freecompiler.html" title="download C++">Download</a> | <a href="https://www.digitalmars.com/NewsGroup.html" title="Forums">Forums</a> </div> <div id="copyright"> Copyright © 1999-2018 by Digital Mars ®, All Rights Reserved | Page generated by <a href="https://dlang.org/spec/ddoc.html">Ddoc</a>. </div> </div><div class="naked_ctrl"> <form action="/index.cgi/contrast" method="get" name="gate"> <p><a href="http://altstyle.alfasado.net">AltStyle</a> によって変換されたページ <a href="https://digitalmars.com/ctg/ctgInlineAsm.html">(->オリジナル)</a> / <label>アドレス: <input type="text" name="naked_post_url" value="https://digitalmars.com/ctg/ctgInlineAsm.html" size="22" /></label> <label>モード: <select name="naked_post_mode"> <option value="default">デフォルト</option> <option value="speech">音声ブラウザ</option> <option value="ruby">ルビ付き</option> <option value="contrast" selected="selected">配色反転</option> <option value="larger-text">文字拡大</option> <option value="mobile">モバイル</option> </select> <input type="submit" value="表示" /> </p> </form> </div> </body> </html>