summaryrefslogtreecommitdiff
path: root/m4
diff options
context:
space:
mode:
authorTimothy B. Terriberry <tterribe@xiph.org>2013-05-19 17:11:17 -0700
committerTimothy B. Terriberry <tterribe@xiph.org>2013-05-19 19:12:51 -0700
commit972a34ec2c79d241318af24389b8ee042d10556a (patch)
tree18894d8e576d351923ed57aacbdec125919d3ba8 /m4
parentb7bd4c20acfd951ba46647e07411285997d952f4 (diff)
downloadopus-972a34ec2c79d241318af24389b8ee042d10556a.tar.gz
Add ARMv4/ARMv5E macros.
Original patch by Aurélien Zanelli <aurelien.zanelli@parrot.com>: http://lists.xiph.org/pipermail/opus/2013-May/002078.html Revised version: - Add autconf detection (ported from libtheora). - Rename ARM5E to ARMv5E (an ARM5 is not the same thing as ARMv5!). - Use actual macros so they can still be selectively overridden. - Split out ARMv4 parts and add a few more ARMv4 macros. - Label blocks to make them easy to find in generated assembly. - Fix MULT16_32_Q15() so we can pass make check. The MDCT test passes in values larger than 2**30 for b. The new version should be just as fast (or faster, since it's easier to merge the shift with following instructions), and there's no appreciable impact on accuracy (FFT/MDCT SNR actually goes up in most cases). - Fix register constraints. We were using early-clobber flags in a bunch of places that didn't need them, and commutative-pair flags in a bunch of places that weren't actually commutative. This was Jean-Marc's fault (the original code came from Speex). - Simplify silk_CLZ16(). - Port over iFFT C_MULC asm by Andree Buschmann <AndreeBuschmann@t-online.de> from Rockbox. - Speed up the C_MULC asm by using LDRD, allowing more flexible addressing, re-ordering instructions to avoid some stalls, allowing more flexible register allocation, and getting things out of the inline asm block so the compiler can schedule them better. - Add C_MUL and C_MUL4 asm for the FFT to the encoder based, on the new C_MULC. In total, this patch gives a 22.3% speed-up on test_opus_encoder on a 600 MHz Cortex A8 using gcc 4.2.1, When restricted to ARMv4 optimizations, it gives a 9.6% speed-up on the same processor/compiler. On the conformance test vectors: Average mono quality is 97.0583 % Average stereo quality is 97.775 %
Diffstat (limited to 'm4')
-rw-r--r--m4/as-gcc-inline-assembly.m4106
1 files changed, 106 insertions, 0 deletions
diff --git a/m4/as-gcc-inline-assembly.m4 b/m4/as-gcc-inline-assembly.m4
new file mode 100644
index 00000000..4437a9d0
--- /dev/null
+++ b/m4/as-gcc-inline-assembly.m4
@@ -0,0 +1,106 @@
+dnl as-gcc-inline-assembly.m4 0.1.0
+
+dnl autostars m4 macro for detection of gcc inline assembly
+
+dnl David Schleef <ds@schleef.org>
+
+dnl $Id$
+
+dnl AS_COMPILER_FLAG(ACTION-IF-ACCEPTED, [ACTION-IF-NOT-ACCEPTED])
+dnl Tries to compile with the given CFLAGS.
+dnl Runs ACTION-IF-ACCEPTED if the compiler can compile with the flags,
+dnl and ACTION-IF-NOT-ACCEPTED otherwise.
+
+AC_DEFUN([AS_GCC_INLINE_ASSEMBLY],
+[
+ AC_MSG_CHECKING([if compiler supports gcc-style inline assembly])
+
+ AC_TRY_COMPILE([], [
+#ifdef __GNUC_MINOR__
+#if (__GNUC__ * 1000 + __GNUC_MINOR__) < 3004
+#error GCC before 3.4 has critical bugs compiling inline assembly
+#endif
+#endif
+__asm__ (""::) ], [flag_ok=yes], [flag_ok=no])
+
+ if test "X$flag_ok" = Xyes ; then
+ $1
+ true
+ else
+ $2
+ true
+ fi
+ AC_MSG_RESULT([$flag_ok])
+])
+
+AC_DEFUN([AC_TRY_ASSEMBLE],
+[ac_c_ext=$ac_ext
+ ac_ext=${ac_s_ext-s}
+ cat > conftest.$ac_ext <<EOF
+ .file "configure"
+[$1]
+EOF
+if AC_TRY_EVAL(ac_compile); then
+ ac_ext=$ac_c_ext
+ ifelse([$2], , :, [ $2
+ rm -rf conftest*])
+else
+ echo "configure: failed program was:" >&AC_FD_CC
+ cat conftest.$ac_ext >&AC_FD_CC
+ ac_ext=$ac_c_ext
+ifelse([$3], , , [ rm -rf conftest*
+ $3
+])dnl
+fi
+rm -rf conftest*])
+
+
+AC_DEFUN([AS_ASM_ARM_NEON],
+[
+ AC_MSG_CHECKING([if assembler supports NEON instructions on ARM])
+
+ AC_TRY_ASSEMBLE([vorr d0,d0,d0], [flag_ok=yes], [flag_ok=no])
+
+ if test "X$flag_ok" = Xyes ; then
+ $1
+ true
+ else
+ $2
+ true
+ fi
+ AC_MSG_RESULT([$flag_ok])
+])
+
+
+AC_DEFUN([AS_ASM_ARM_MEDIA],
+[
+ AC_MSG_CHECKING([if assembler supports ARMv6 media instructions on ARM])
+
+ AC_TRY_ASSEMBLE([shadd8 r3,r3,r3], [flag_ok=yes], [flag_ok=no])
+
+ if test "X$flag_ok" = Xyes ; then
+ $1
+ true
+ else
+ $2
+ true
+ fi
+ AC_MSG_RESULT([$flag_ok])
+])
+
+
+AC_DEFUN([AS_ASM_ARM_EDSP],
+[
+ AC_MSG_CHECKING([if assembler supports EDSP instructions on ARM])
+
+ AC_TRY_ASSEMBLE([qadd r3,r3,r3], [flag_ok=yes], [flag_ok=no])
+
+ if test "X$flag_ok" = Xyes ; then
+ $1
+ true
+ else
+ $2
+ true
+ fi
+ AC_MSG_RESULT([$flag_ok])
+])