Date: 2018.9.13
1、参考
https://blog.csdn.net/SoaringLee_fighting/article/details/81906495
https://blog.csdn.net/u011514906/article/details/38142177
https://blog.csdn.net/listener51/article/details/82530464
2、前言
本文是arm架构64位(AArch64执行状态) neon优化的总结文档,主要包括arm架构64位优化的基础知识,特殊用法,打印调试和常用指令使用注意事项以及资料来源等相关知识。前文已有arm架构32位汇编优化总结对arm架构32位neon优化进行了全面总结,并且讲述了arm汇编语法,下面主要以gnu arm汇编语法为例讲述。
3、arm架构64位优化基础知识
https://blog.csdn.net/SoaringLee_fighting/article/details/81906495
该博客已经分析了arm架构64位汇编优化的入门基础知识,主要包括架构分析,寄存器,调用规则,指令集和程序打印调试相关知识,可以作为入门arm64位汇编优化的基础知识。
4、ARMv8/AArch64 neon指令格式
In the AArch64 execution state, the syntax of NEON instruction has changed. It can be described as follows:
{<prefix>}<op>{<suffix>} Vd.<T>, Vn.<T>, Vm.<T>
Where:
< prefix> - prefix, such as using S/U/F/P to represent signed/unsigned/float/bool data type.
< op> – operation, such as ADD, AND etc.
< suffix> - suffixP: “pairwise” operations, such as ADDP,LDP,STP.
V: the new reduction (across-all-lanes) operations, such as ADDV,SMAXV,FMAXV.
2:new widening/narrowing “second part” instructions, such as ADDHN2, SADDL2,SMULL2.< T> - data type, 8B/16B/4H/8H/2S/4S/2D. B represents byte (8-bit). H represents half-word (16-bit). S represents word (32-bit). D represents a double-word (64-bit).
For example:
UADDLP V0.8H, V0.16B
FADD V0.4S, V0.4S, V0.4S
5、ARM相关编译参数
嵌入式设备(即arm架构的板子)在编译时,最好加上 -fsigned-char 因为嵌入式设备默认类型为unsigned char类型,非char 类型。此外在编译arm汇编优化代码时,编译选项需要加上-c 。-c都表示编译或汇编源文件,但是不进行链接。
ARM相关或者硬件相关编译参数一般以-m开头,常用ARM平台编译选项包括:
-mcpu = cortex-a7
-mabi = atpcs
-march = armv7
-mtune = cortex-a53
-mfpu = neon, neon-vfpv4
-mfloat-api = soft, softfp, hard
更多详细内容可以参考:https://gcc.gnu.org/onlinedocs/gcc-5.2.0/gcc.pdf 3.17.1小节AArch64 option和3.17.4小节 ARM options.
6、查看状态标记位NZCV的方法
mrs x15, nzcv
mov w0, w15
bl print
7、A64指令集特有的指令及其用法
1. shl和ushr指令
shl <V>.<d>, <V>.<n>, #<shift>
ushr <V>.<d>, <V>.<n>, #<shift>
ushr d2, d2, #8
使用注意事项:这两条指令只能操作64位数据,即只能对D寄存器进行处理。
ushr最多只能进行64位数据的右移,并且右移时会影响V2寄存器的高64位数据(清零),因此高64位数据需要在右移前保存,否则相关数据会被修改。
2. INS指令
用法与MOV指令基本一样,可以实现neon标量与neon标量之间的传送,以及ARM寄存器与neon标量之间的传送。
INS <Vd>.<Ts>[index1], <Vn>.<Ts>[index2]
INS <Vd>.<Ts>[index1], Rn
3. SUQADD、USQADD指令
既有标量用法,也有矢量用法。
SUQADD <V><d>, <V><d> // signed saturating accumulate of unsigned value
SUQADD <Vd>.<T>, <Vn>.<T>
USQADD <V><d>, <V><d> // unsigned saturating accumulate of signed value
USQADD <Vd>.<T>, <Vn>.<T>
4. RBIT、REV指令
RBIT <Wd>, <Wn> //reverse bits
REV <Wd>, <Wn> //reverse bytes
5. ADDV,SADDLV,SMAXV,SMINV (Vector Reduce(across lanes))
ADDV <V><d>, <Vn><T> // Integer sum element to scalar(vector)
SADDLV <V><d>, <Vn><T> // Signed Interger sum elements to long scalar(vector)
SMAXV <V><d>, <Vn><T> // Signed Interger maximum elements to scalar(vector)
SMINV <V><d>, <Vn><T> // Signed Interger minimum elements to scalar(vector)
8、资料文档查阅
THE END!