embeded/ARM2015. 1. 28. 08:39

neon을 쓰기 위해서는 kernel이나 uboot 등에서

FPU/NEON을 활성화 해야 하는 것으로 보인다.


The NEON/VFP unit comes up disabled on power on reset. To enable Neon requires some co-processor commands. Below is the assembly code required to enable NEON/VFP. It is in the gcc type syntax. ARM code tools use a slightly different syntax.


Neon and VFP both support floating point, which should I use?

The VFPv3 is fully compliant with IEEE 754

Neon is not fully compliant with IEEE 754, so it is mainly targeted for multimedia applications


[링크 : http://processors.wiki.ti.com/index.php/Cortex-A8]


A sequence for initializing the VFPU can be found in u-boot source.


.macro init_vfpu

  ldr r0, =(0xF << 20)

  mcr p15, 0, r0, c1, c0, 2

  mov r3, #0x40000000

  .long 0xeee83a10

  /* vmsr FPEXC, r3 */

.endm /* init_vfpu */


[링크 : http://stackoverflow.com/questions/19231197/enable-neon-on-arm-cortex-a-series]


Using the Advanced SIMD and VFP in Secure state and Non-secure state other than Hyp mode

To use the Advanced SIMD and VFP in Secure state and Non-secure state other than Hyp mode, you must first define the NSACR, then define the CPACR and FPEXC registers. See Non-Secure Access Control RegisterCoprocessor Access Control Register, and Floating-Point Exception Register.

  1. Enable Non-secure access to CP10 and CP11 and clear the NSASEDIS bit in the NSACR:

    MRC p15, 0, r0, c1, c1, 2
    
    ORR r0, r0, #(3<<10)	    ; Enable Non-secure access to CP10 and CP11
    
    BIC r0, r0, #(3<<14)	    ; Clear NSASEDIS bit
    
    MCR p15, 0, r0, c1, c1, 2
    
    ISB
    
  2. Enable access to CP10 and CP11 and clear the ASEDIS bit in the CPACR:

    MOV r0, 0x00F00000
    
    MCR p15, 0, r0, c1, c0, 2
    
    ISB
    
  3. Set the FPEXC.EN bit to enable Advanced SIMD and VFP:

    MOV r3, #0x40000000
    
    VMSR FPEXC, r3


Kernel mode NEON

================


Introduction

------------

It is possible to use NEON instructions (and in some cases, VFP instructions) in code that runs in kernel mode. However, for performance reasons, the NEON/VFP register file is not preserved and restored at every context switch or taken exception like the normal register file is, so some manual intervention is required. Furthermore, special care is required for code that may sleep [i.e.,may call schedule()], as NEON or VFP instructions will be executed in a non-preemptible section for reasons outlined below.


[링크 : https://www.kernel.org/doc/Documentation/arm/kernel_mode_neon.txt] 



아무튼.. 커널 패치하고 make ARCH=arm menuconfig 로 살펴보니 Floatting point emulation에

(x86일 기본 make menuconfig에서는 보이지 않음)


이렇게 뚜둥! VFP 하위에 Advanced SIMD(NEON)이라고 똭!


머. 도움말은 별거 없다.

[링크 : http://cateee.net/lkddb/web-lkddb/NEON.html]

'embeded > ARM' 카테고리의 다른 글

i.mx515 SDMA  (0) 2015.02.02
armv7 errata / kernel  (0) 2015.01.28
armel / armhf  (0) 2015.01.26
ltib on ubuntu  (0) 2015.01.12
i.mx283 부팅모드  (0) 2015.01.09
Posted by 구차니
embeded/ARM2015. 1. 26. 17:06

리눅스에서 arm 아키텍쳐를 지칭하는 말인데

armhf는 arm Hard Float의 약자인건 확인이 되는데 반해

armel은 약자가 확인이 되지 않는다.


개인적인 추측으로는..

arm emulated 이 아닐까 싶지만..

이렇게 약자를 만드는건 극히 드물테니.. 멀려나?


The armel architecture supports the ARMv4 instruction set.

The armhf architecture supports ARMv7 platform, and more, it adds direct hardware floating-point support.

[링크 : http://www.xappsoftware.com/wordpress/2013/01/31/armhf-versus-armel/]


In Debian Linux and derivatives armhf (ARM hard float) refers to the ARMv7 architecture including the additional VFP3-D16 floating-point hardware extension (and Thumb-2) above.

Software packages and cross-compiler tools use the armhf vs. arm/armel suffixes to differentiate.

[링크 : http://en.wikipedia.org/wiki/ARM_architecture#cite_ref-60]

[링크 :https://blogs.oracle.com/jtc/entry/is_it_armhf_or_armel]



두둥! arm endian little / big 인 듯?

---

Name of the port

The table below recaps which port names Debian/dpkg saw so far.

name

endianess

status

arm

little-endian

Original Debian arm port using original ABI ('OABI'), last release in Debian lenny; being retired in favor of armel

armel

little-endian

introduced in Debian lenny; EABI, actively maintained; targets armv4t; doesn't require an FPU

armeb

big-endian

unofficial OABI port; inactive and dead

[링크 : https://wiki.debian.org/ArmHardFloatPort]

'embeded > ARM' 카테고리의 다른 글

armv7 errata / kernel  (0) 2015.01.28
NEON enabled?  (0) 2015.01.28
ltib on ubuntu  (0) 2015.01.12
i.mx283 부팅모드  (0) 2015.01.09
freescale imx283 boot error code  (0) 2015.01.07
Posted by 구차니
embeded/ARM2015. 1. 12. 10:42

플랫폼

Ubuntu 14.04 LTS

ltib 13.2.1 (cvs)


Package                Minimum ver   Installed info

-------                -----------   ---------------

glibc-devel            0             not installed

zlib                   0             not installed

zlib-devel             0             not installed

ncurses-devel          0             not installed

m4                     0             not installed

bison                  0             not installed


$ sudo apt-get install libncurses-dev bison zlibc zlib1g-dev

[링크 : http://ubuntuforums.org/showthread.php?t=1632027]



32비트의 경우 아래의 내용으로

$ vi ltib/bin/Ltibutils.pm

 559 $app_checks = {

 560     binutils         => 'ar --version 2>/dev/null',

 561     'gcc-c++'        => 'g++ --version 2>/dev/null',

 562     glibc            => 'ldd --version 2>/dev/null',

 563     'glibc-devel'    => sub { -f '/usr/lib/i386-linux-gnu/libm.so' || -f '/usr/lib/i386-linux-gnu/libz.so'},

 564     'glibc-headers'  => sub { -f '/usr/include/stdio.h' },

 565     'libstdc++' => sub {

 566             return system_nb(<<TXT) == 0;

 567 echo '#include <iostream>

 568 int main() { std::cout << "test"; }' | g++ -x c++ - -o /dev/null

 569 TXT

 570                        },

 571     lkc              => 'mconf -h 2>/dev/null',

 572     ncurses          => 'tic -V 2>/dev/null',

 573     'ncurses-devel'  => sub { -f '/usr/include/ncurses.h' },

 574     'rpm-build'      => sub { `rpmbuild --version 2>/dev/null` },

 575     sudo             => 'sudo -V 2>/dev/null',

 576     tcl              => "echo 'puts \$tcl_patchLevel' | tclsh",

 577     texinfo          => sub {

 578                        warn  "WARNING: you may also need to install:"

 579                             ." tetex-fonts, dialog and textex\n"

 580                          unless  -f '/usr/share/texmf/tex/texinfo/texinfo.tex';

 581                         `makeinfo --version 2>/dev/null`;

 582                     },

 583     zlib         => sub { my @f = (glob('/usr/lib/libz.so*'),

 584                                    glob('/lib/libz.so*'),

 585                                    glob('/lib/i386-linux-gnu/libz.so*')  ); @f > 1 ? 1 : 0 },

 586     'zlib-devel' => sub { -f '/usr/include/zlib.h' },

 587 }; 


[링크 : http://markdingst.blogspot.kr/2011/06/ltib-1105-miss-zlib-under-ubuntu-1104.html]


64비트는 링크의 내용으로

[링크 : https://community.freescale.com/thread/311978]




으아악 막혔다!!! ㅠㅠ

rpmdb: --force-debian: unknown option
Died at ./ltib line 2340.
traceback:
 main::setup_rpmdb:2340
  main::check_rpm_setup:2403
   main::host_checks:1423
    main:542


Started: Mon Jan 12 00:37:29 2015
Ended:   Mon Jan 12 00:37:30 2015
Elapsed: 1 seconds


Build Failed



$ vi ltib

2341 set -e

2342 $cf->{sudo} $cf->{rpm} --root $cf->{rpmroot} --dbpath $cf->{rpmdb} --initdb ||

2343 $cf->{sudo}       rpm   --root $cf->{rpmroot} --dbpath $cf->{rpmdb} --initdb

2344 TXT

2345     } elsif($do_rebuild) {

2346         system_nb(<<TXT) == 0 or die;

2347 set -e

2348 $cf->{sudo} $cf->{rpm} --root $cf->{rpmroot} --dbpath $cf->{rpmdb} --define '_tmppath $cf->{tmppath}' --rebuilddb

2349 $cf->{sudo}        rpm --root $cf->{rpmroot} --dbpath $cf->{rpmdb} --define '_tmppath $cf->{tmppath}' --rebuilddb

2350 TXT 


[링크 : http://cvs.savannah.gnu.org/viewvc/ltib/ltib?root=ltib&r1=1.76&r2=1.77]

 [링크 : https://community.freescale.com/thread/263568]


Processing: lkc

=================

Build path taken because: directory build, no prebuilt rpm, 


Cowardly refusing to clobber existing directory:

 /opt/freescale/ltib/usr/src/rpm/BUILD/lkc-1.4

Remove this by hand if you really want to rebuild this package from scratch


Died at ./ltib line 1380.

traceback:

 main::build_host_rpms:1380

  main::host_checks:1435

   main:542


[링크 : https://community.freescale.com/message/271839#271839]




----


$ sudo apt-get install libncurses-dev bison perl libwww-perl alien cvs

$ cvs -z3 -d:pserver:anonymous@cvs.savannah.nongnu.org:/sources/ltib co -P ltib

$ cd ltib

$ wget https://community.freescale.com/servlet/JiveServlet/downloadBody/93455-102-3-2835/patch-ltib-ubuntu12.04.sh

$ chmod +x patch-ltib-ubuntu12.04.sh

$ ./patch-ltib-ubuntu12.04.sh


$ sudo visudo

interm ALL = NOPASSWD: /usr/bin/rpm, /opt/ltib/usr/bin/rpm


$ ./ltib 


wget-ssl에서 실패 -_-

[링크 : https://community.freescale.com/thread/324608]

[링크 : https://lists.gnu.org/archive/html/bug-wget/2011-08/msg00041.html]

--without-ssl  삭제


[링크 : http://ltib.org/resources-download#ref_2]

[링크 : https://community.freescale.com/docs/DOC-93455]



우분투 하위버전은 지원하긴 하는데.. 왜케 안되는거야 -_-

x86 Linux

    Redhat: 7.3, 8.0, 9.0, Enterprise release 4

    Fedora Core: 1, 2, 3, 4, 5, 8

    Debian: 3.1r0 (stable), unstable, 4.0

    Suse: 8.2, 9.1, 9.2, 10.0, 10.2, 10.3

    Ubuntu 6.10, 7.04, 7.10, 8.04


[링크 : http://ltib.org/documentation-LtibFaq]


-- 2015.02.10 추가

[링크 : https://importgeek.wordpress.com/2014/05/13/i-mx28-ltib-on-ubuntu-13-04/]


패치

[링크 : https://turbosree.wordpress.com/tag/freescale-ltib-on-ubuntu-12-04/]

    [링크 : https://community.freescale.com/docs/DOC-93455]

    [링크 : https://community.freescale.com/.../93454-102-3-2834/ubuntu-ltib-patch.tgz]

[링크: https://community.freescale.com/thread/308138]

    [링크 : https://community.freescale.com/.../download/271839-220893/438-CreateUbuntu1110ltib110901.docx]


---

2015.04.13

$cf->{rpm} = "rpm" 을 $cf->{rpm} = "rpm --force-debian"$cf->{rpm} = "rpm" 을 $cf->{rpm} = "rpm --force-debian"

[링크 : http://poongbek.blogspot.kr/2011/06/ltib.html]

'embeded > ARM' 카테고리의 다른 글

NEON enabled?  (0) 2015.01.28
armel / armhf  (0) 2015.01.26
i.mx283 부팅모드  (0) 2015.01.09
freescale imx283 boot error code  (0) 2015.01.07
Unknown HZ value! (91) Assume 100.  (0) 2015.01.05
Posted by 구차니
embeded/ARM2015. 1. 9. 23:33

USB Recovery mode로 인해

SD나 MMC 부팅 실패시 USB 부팅모드로 넘어가게 된다. (기본값)

설정에 따라서는 USB 복구 모드를 끌수는 있으나 아무튼 기본값 -_-


이런 중요(!)한 내용이 왜.. cpu 관련 내용에는 없고

EVK 관련 내용에 있는걸까 -_-


12.13.4 USB Recovery Mode

USB boot mode is provided as a fail-safe mechanism for writing system firmware to the boot media. The boot mode is not usually entered by the normal methods of setting the boot pins or OTP, the other methods of entering USB boot mode are referred to generally as recovery mode.

An end user can manually start the recovery mode by holding the recovery switch for several seconds while plugging in USB. Holding the recovery switch is defined as reading the i.MX28 PSWITCH input as a 0x3. There are several switch circuits that will produce this input. The loader also automatically starts recovery mode if a nonrecoverableerror is detected from any boot mode other than USBdisabled completely. Attempts to enter USB boot mode through boot pins, OTP, or recovery methods will result in a chip power-down.


[링크 : http://cache.freescale.com/...=pdf&WT_ASSET=Documentation&fileExt=.pdf]


'embeded > ARM' 카테고리의 다른 글

armel / armhf  (0) 2015.01.26
ltib on ubuntu  (0) 2015.01.12
freescale imx283 boot error code  (0) 2015.01.07
Unknown HZ value! (91) Assume 100.  (0) 2015.01.05
크로스 컴파일 옵션 configure --host  (0) 2014.10.13
Posted by 구차니
embeded/ARM2015. 1. 7. 11:15

SD 메모리 미 삽입

#define ERROR_DDI_SD_MMC_DEVICE_NOT_SUPPORTED (ERROR_DDI_GROUP | ERROR_DDI_SD_DRIVER_GROUP | 0x14)

0x8020A014 ?


빈 SD 메모리

#define ERROR_DDI_SD_MBR_NOT_FOUND (ERROR_DDI_GROUP | ERROR_DDI_SD_DRIVER_GROUP | 0x1D)

0x8020A01D ?


부팅 안될경우 일정 시간뒤 출력

#define ERROR_ROM_USB_CONNECT_TIMEOUT (ERROR_ROM_GROUP | ERROR_ROM_USB_DRIVER_GROUP | 0x8)

0x80502008

'embeded > ARM' 카테고리의 다른 글

ltib on ubuntu  (0) 2015.01.12
i.mx283 부팅모드  (0) 2015.01.09
Unknown HZ value! (91) Assume 100.  (0) 2015.01.05
크로스 컴파일 옵션 configure --host  (0) 2014.10.13
iWMMX / iWMMXt - Intel XScale SIMD instructions  (0) 2014.10.08
Posted by 구차니
embeded/ARM2015. 1. 5. 15:10



sysinfo.c

 /***********************************************************************

* Some values in /proc are expressed in units of 1/HZ seconds, where HZ * is the kernel clock tick rate. One of these units is called a jiffy. * The HZ value used in the kernel may vary according to hacker desire. * According to Linus Torvalds, this is not true. He considers the values * in /proc as being in architecture-dependant units that have no relation * to the kernel clock tick rate. Examination of the kernel source code * reveals that opinion as wishful thinking. * * In any case, we need the HZ constant as used in /proc. (the real HZ value * may differ, but we don't care) There are several ways we could get HZ: * * 1. Include the kernel header file. If it changes, recompile this library. * 2. Use the sysconf() function. When HZ changes, recompile the C library! * 3. Ask the kernel. This is obviously correct... * * Linus Torvalds won't let us ask the kernel, because he thinks we should * not know the HZ value. Oh well, we don't have to listen to him. * Someone smuggled out the HZ value. :-) * * This code should work fine, even if Linus fixes the kernel to match his * stated behavior. The code only fails in case of a partial conversion. * * Recent update: on some architectures, the 2.4 kernel provides an * ELF note to indicate HZ. This may be for ARM or user-mode Linux * support. This ought to be investigated. Note that sysconf() is still * unreliable, because it doesn't return an error code when it is * used with a kernel that doesn't support the ELF note. On some other * architectures there may be a system call or sysctl() that will work. */ unsigned long long Hertz; static void old_Hertz_hack(void){ unsigned long long user_j, nice_j, sys_j, other_j; /* jiffies (clock ticks) */ double up_1, up_2, seconds; unsigned long long jiffies; unsigned h; char *restrict savelocale; savelocale = setlocale(LC_NUMERIC, NULL); setlocale(LC_NUMERIC, "C"); do{ FILE_TO_BUF(UPTIME_FILE,uptime_fd); sscanf(buf, "%lf", &up_1); /* uptime(&up_1, NULL); */ FILE_TO_BUF(STAT_FILE,stat_fd); sscanf(buf, "cpu %Lu %Lu %Lu %Lu", &user_j, &nice_j, &sys_j, &other_j); FILE_TO_BUF(UPTIME_FILE,uptime_fd); sscanf(buf, "%lf", &up_2); /* uptime(&up_2, NULL); */ } while((long long)( (up_2-up_1)*1000.0/up_1 )); /* want under 0.1% error */ setlocale(LC_NUMERIC, savelocale); jiffies = user_j + nice_j + sys_j + other_j; seconds = (up_1 + up_2) / 2; h = (unsigned)( (double)jiffies/seconds/smp_num_cpus ); /* actual values used by 2.4 kernels: 32 64 100 128 1000 1024 1200 */ switch(h){ case 9 ... 11 : Hertz = 10; break; /* S/390 (sometimes) */ case 18 ... 22 : Hertz = 20; break; /* user-mode Linux */ case 30 ... 34 : Hertz = 32; break; /* ia64 emulator */ case 48 ... 52 : Hertz = 50; break; case 58 ... 61 : Hertz = 60; break; case 62 ... 65 : Hertz = 64; break; /* StrongARM /Shark */ case 95 ... 105 : Hertz = 100; break; /* normal Linux */ case 124 ... 132 : Hertz = 128; break; /* MIPS, ARM */ case 195 ... 204 : Hertz = 200; break; /* normal << 1 */ case 253 ... 260 : Hertz = 256; break; case 393 ... 408 : Hertz = 400; break; /* normal << 2 */ case 790 ... 808 : Hertz = 800; break; /* normal << 3 */ case 990 ... 1010 : Hertz = 1000; break; /* ARM */ case 1015 ... 1035 : Hertz = 1024; break; /* Alpha, ia64 */ case 1180 ... 1220 : Hertz = 1200; break; /* Alpha */ default: #ifdef HZ Hertz = (unsigned long long)HZ; /* <asm/param.h> */ #else /* If 32-bit or big-endian (not Alpha or ia64), assume HZ is 100. */ Hertz = (sizeof(long)==sizeof(int) || htons(999)==999) ? 100UL : 1024UL; #endif fprintf(stderr, "Unknown HZ value! (%d) Assume %Ld.\n", h, Hertz); } }





[링크 : http://lkml.iu.edu/hypermail/linux/kernel/0401.2/0065.html]

[링크 : http://forum.buffalo.nas-central.org/viewtopic.php?f=18&t=13451]

[링크 : http://procps.sourceforge.net/]

[링크 : https://bugs.launchpad.net/ubuntu/+source/procps/+bug/364656]

[링크 : https://launchpadlibrarian.net/29934773/procps_3.2.7-11ubuntu3.debdiff.gz]

[링크 : https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=460331]

    [링크 : https://lkml.org/lkml/2002/2/18/187]


'embeded > ARM' 카테고리의 다른 글

i.mx283 부팅모드  (0) 2015.01.09
freescale imx283 boot error code  (0) 2015.01.07
크로스 컴파일 옵션 configure --host  (0) 2014.10.13
iWMMX / iWMMXt - Intel XScale SIMD instructions  (0) 2014.10.08
thumb의 장단점?  (0) 2014.09.29
Posted by 구차니
embeded/ARM2014. 10. 13. 12:10
target만 지정하면 되는줄 알았는데 음..
target으로 지정할 경우
libtool에서 인식을 하지 못해 shared object등을 생성하지 못하는 문제가 있기에
host로 지정을 하라고 한다.

--build=BUILD BUILD 상에서의 빌드를 위한 설정 (i686?)
--host=HOST HOST 상에서 실행되는 프로그램을 빌드하기 위한 크로스 컴파일러 (arm-none-linux-gnueabi)
--target=TARGET TARGET을 위한 빌딩 컴파일러를 위한 설정

System types:
  --build=BUILD     configure for building on BUILD [guessed]
  --host=HOST       cross-compile to build programs to run on HOST [BUILD]
  --target=TARGET   configure for building compilers for TARGET [HOST]

[링크 : http://stackoverflow.com/.../libtool-claims-it-does-not-support-shared-libraries-during-cross-compilation

'embeded > ARM' 카테고리의 다른 글

freescale imx283 boot error code  (0) 2015.01.07
Unknown HZ value! (91) Assume 100.  (0) 2015.01.05
iWMMX / iWMMXt - Intel XScale SIMD instructions  (0) 2014.10.08
thumb의 장단점?  (0) 2014.09.29
dsp & simd / neon  (0) 2014.09.22
Posted by 구차니
embeded/ARM2014. 10. 8. 13:52
췟 -_-
우연히 MPlayer 컴파일 하다가 발견했는데
집에 굴러 다니는 녀석은 PXA255.. -_-
해당사항이 없네? ㅠㅠ
 Intel's and Marvell's XScale microprocessor core starting with PXA270 include an SIMD instruction set extension to the ARM core called iwMMXt whose functions are similar to those of the IA-32 MMX extension. iwMMXt stands for "Intel Wireless MMX Technology". It provides arithmetic and logic operations on 64-bit integer numbers (the software may choose to instead perform two 32-bit, four 16-bit or eight 8-bit operations in a single instruction). The extension contains 16 data registers of 64-bits and eight control registers of 32-bits. All registers are accessed through standardARM architecture coprocessor mapping mechanism. iwMMXt occupies coprocessors 0 and 1 space, and some of itsopcodes clash with the opcodes of the earlier floating-point extension, FPA.

Later versions of Marvell's ARM processors supports both WMMX (Wireless MMX) and WMMX2 (Wireless MMX2) support.


[링크 : http://en.wikipedia.org/wiki/MMX_(instruction_set)#MMX_in_embedded_applications
[링크 : https://gcc.gnu.org/onlinedocs/gcc-4.3.4/gcc/ARM-iWMMXt-Built_002din-Functions.html

'embeded > ARM' 카테고리의 다른 글

Unknown HZ value! (91) Assume 100.  (0) 2015.01.05
크로스 컴파일 옵션 configure --host  (0) 2014.10.13
thumb의 장단점?  (0) 2014.09.29
dsp & simd / neon  (0) 2014.09.22
Freescale Multimedia framwork  (0) 2014.09.17
Posted by 구차니
embeded/ARM2014. 9. 29. 17:37
ffmpeg 컴파일 옵션중에
  --enable-thumb           compile for Thumb instruction set

thumb 활성화가 있고 cpu도 지원을 해서
Processor       : ARM926EJ-S rev 5 (v5l)
BogoMIPS        : 226.09
Features        : swp half thumb fastmult edsp java 

 --enable-armv5te 랑 같이 썼더니.... 안되잖아!!!! ㅠㅠ
Assembler messages:
Error: instruction not supported in Thumb16 mode -- `adds r2,r5,r4,lsr#31'
Error: selected processor does not support `itet ne'
Error: Thumb does not support conditional execution

armv5te의 DSP enhancement 명령어들이 16bit(thumb mode) 가 아닌 32bit라 안되는게 아닐까 라고 추측되는데..
아무튼..  thumb의 장점으로는 ARM에서 코드 밀도가 올라간다(코드 사이즈가 65%까지 준다고..) 는 것 외에는
성능 향상적인 측면은 크게 없으니 굳이 무리하게 thumb 옵션을 쓰지 않아도 될 듯.
 
단점들을 나열해 볼게요. 
 
1. 분기 명령어를 제외하고는 조건부 실행이 안됩니다. 
2. 레지스터 사용이 R0~R7으로 제안 됩니다.
3. Immediate 상수 값의 사용 범위가 제한적입니다.
4. Inline barrel shifter의 사용이 제안적입니다.
5. Exception 처리를 할 수 없습니다.
[링크 : http://recipes.egloos.com/viewer/5651064

8.1 Thumb Instruction 특징
(1) 16-bit length instruction set
(2) ARM 명령어보다 코드의 집적도가 높습니다.( about 65% of ARM instruction )
(3) 일반적으로는 32bit ARM명령어 보다는 속도가 느리지만 16bit memory 시스템에서는 그렇지 않을 수도 있습니다.

8.2 Thumb Instruction 제약 사항

- Limited Access to Registers : R0-R7 registers are accessible.
- Narrow Range of Immediate Value
- Not Flexible for Exception Mode
- Exception Handler should be executed in ARM mode. : Exception이 발생하면 항상 ARM 모드로 전환이 됩니다.
- Limited conditional instruction.
- Branch instructions can be executed conditionally.
- Inline Barrel Shifter is not used. 
 
[링크 : http://www.jkelec.co.kr/img/lecture/arm_arch/arm_arch_4.html#8]  

'embeded > ARM' 카테고리의 다른 글

크로스 컴파일 옵션 configure --host  (0) 2014.10.13
iWMMX / iWMMXt - Intel XScale SIMD instructions  (0) 2014.10.08
dsp & simd / neon  (0) 2014.09.22
Freescale Multimedia framwork  (0) 2014.09.17
freescale i.mx283 / i.mx515  (0) 2014.08.19
Posted by 구차니
embeded/ARM2014. 9. 22. 14:38
i.mx283 뒤지다가 영 스펙이 제대로 안나와서 헤매느중..
아무튼 검색을 하다보니 DSP & SIMD로
ARMv5TE 계열에 지원하는 DSP Enchancement 로 몇가지 명령어를 지원하는 것으로 보이지만...
컴파일 옵션에 DSP multiply 정도로 밖에 안나오는것 봐서는
ARMv6 계열의 SIMD에 비하면 정말 미미한 수준의 DSP/멀티미디어 확장일 것으로 보인다.

ARMv5TE 계열인 ARM946의 DSP enhancement instruction
느낌으로는.. 32bit 짜리로 8bit 씩 4개의 데이터에 대한 확장 명령이 존재할 것으로 보여진다.

[링크 : http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.dvi0022a/ar01s02s14.html]

QADD, QSUB, QDADD, and QDSUB
Signed Add, Subtract, Double and Add, Double and Subtract, saturating the result to the signed range -2^31 ≤ x ≤ 2^31-1.

Syntax
op{cond} {Rd}, Rm, Rn

where:
op        is one of QADD, QSUB, QDADD, or QDSUB.
cond     is an optional condition code.
Rd         is the destination register.
Rm, Rn  are the registers holding the operands.
[링크 : http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.dui0489e/Cihidceh.html

[링크 : http://www.arm.com/products/processors/technologies/dsp-simd.php]

5TEJ ARMv5 with Thumb, interworking, DSP multiply, double-word instructions, and Jazelle® extensions ARM926EJ-S, ARM1026EJ-S, SC200 
[링크 : http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.dui0491g/CIHGEBDH.html]

DSP enhancement instructions
To improve the ARM architecture for digital signal processing and multimedia applications, DSP instructions were added to the set. These are signified by an "E" in the name of the ARMv5TE and ARMv5TEJ architectures. E-variants also imply T, D, M and I.

The new instructions are common in digital signal processor architectures. They include variations on signed multiply–accumulate, saturated add and subtract, and count leading zeros. 

'embeded > ARM' 카테고리의 다른 글

iWMMX / iWMMXt - Intel XScale SIMD instructions  (0) 2014.10.08
thumb의 장단점?  (0) 2014.09.29
Freescale Multimedia framwork  (0) 2014.09.17
freescale i.mx283 / i.mx515  (0) 2014.08.19
ARM thumb과 Jazelle  (0) 2014.04.19
Posted by 구차니