search for: _mm_cvtsi128_si32

...g/no-spilling code put on conditional compile: #if __SSE4_1__ != 0 #include <smmintrin.h> #else #include <emmintrin.h> #endif #include <stdint.h> #include <assert.h> #if SPILLING_ENSUES == 1 static int32_t geti(const __m128i v, const size_t i) { switch (i) { case 0: return _mm_cvtsi128_si32(v); case 1: return _mm_cvtsi128_si32(_mm_shuffle_epi32(v, 0xe5)); case 2: return _mm_cvtsi128_si32(_mm_shuffle_epi32(v, 0xe6)); case 3: return _mm_cvtsi128_si32(_mm_shuffle_epi32(v, 0xe7)); } assert(0); return -1; } #else static int32_t geti(const __m128i v, const size_t i) { switch (i) { case 0:...

...h> + #include <time.h> ]], [[ - static __m128i mtest; - mtest = _mm_setzero_si128(); + __m128i mtest; + mtest = _mm_set1_epi32((int)time(NULL)); + mtest = _mm_mul_epu32(mtest, mtest); + return _mm_cvtsi128_si32(mtest); ]] ) AS_IF([test x"$OPUS_X86_MAY_HAVE_SSE2" = x"1" && test x"$OPUS_X86_PRESUME_SSE2" != x"1"], @@ -557,11 +564,13 @@ AS_IF([test x"$enable_intrinsics" = x"yes"],[ [OPUS_X86_MAY_HAVE_SSE4_1]...

...at somebody else will find what's wrong and will create a bugreport... Well, here is the link: https://connect.microsoft.com/VisualStudio/feedback/details/2659191/incorrect-code-generation-for-x86-64 It seems that MSVC miscompiles abs_residual_partition_sums[partition] = (FLAC__uint32)_mm_cvtsi128_si32(mm_sum); into movq QWORD PTR [rsi], xmm2 So it incorrectly copies 8 bytes from mm_sum to abs_residual_partition_sums[partition] (it should copy 4 lower bytes and zero out 4 upper bytes). It should be something like movd eax, xmm2 movq QWORD PTR [rsi], rax

It turns out that int64 shift is quite slow... This patch changes the code from: (FLAC__int32)(xmm.m128i_i64[0] >> lp_quantization) into: _mm_cvtsi128_si32(_mm_srli_epi64(xmm, lp_quantization)); Encoding of 24-bit .wav files with 32-bit FLAC became noticeably faster. The new code works only if quantization <= 32, but its max value is 15 so the code always work. (max_shiftlimit == (1 << (FLAC__SUBFRAME_LPC_QLP_SHIFT_LEN-1)) - 1 == 15)

Here's a new version of a patch that fixes a problem with MSVC2105 update2, but it doesn't disable any optimization, so the resulting encoding performance should be almost unaffected by this workaround. MSVC compiles abs_residual_partition_sums[partition] = (FLAC__uint32)_mm_cvtsi128_si32(mm_sum); into this: movq QWORD PTR [rsi], xmm2 while it should be movd eax, xmm2 mov QWORD PTR [rsi], rax With this patch, MSVC emits movq QWORD PTR [rsi], xmm2 mov DWORD PTR [rsi+4], r9d so the price of this workaround is 1 extra write instruction p...

lvqcl wrote: > It turns out that int64 shift is quite slow... > > This patch changes the code from: > (FLAC__int32)(xmm.m128i_i64[0] >> lp_quantization) > into: > _mm_cvtsi128_si32(_mm_srli_epi64(xmm, lp_quantization)); > > Encoding of 24-bit .wav files with 32-bit FLAC became noticeably faster. > > > The new code works only if quantization <= 32, but its max value is 15 so the code always work. > (max_shiftlimit == (1 << (FLAC__SUBFRAME_LPC_QLP_...

Erik de Castro Lopo wrote: >> As I wrote earlier, MSVC 2015 U2 incorrectly compiles >> stream_encoder_intrin_*.c files for x86-64 platform. >> As a result, flac works, but compression ratio is close to 1. >> This patch disables some compiler optimizations, and >> compression ratio reverts back to normal values. > > Rather than having the same chunk of code in

...i + 0]); - - inVec1_3210 = _mm_mullo_epi32(inVec1_3210, inVec2_3210); - - acc1 = _mm_add_epi32(acc1, inVec1_3210); - i += 4; - } - - acc1 = _mm_add_epi32(acc1, _mm_unpackhi_epi64(acc1, acc1)); - acc1 = _mm_add_epi32(acc1, _mm_shufflelo_epi16(acc1, 0x0E)); - - sum += _mm_cvtsi128_si32(acc1); - - for (;i<N;i++) - { - sum = silk_SMLABB(sum, x[i], y[i]); - } +#include <xmmintrin.h> +#include "arch.h" - return sum; +void xcorr_kernel_sse(const opus_val16 *x, const opus_val16 *y, opus_val32 sum[4], int len) +{ + int j; + __m128 xsum1, xsum2...

...i + 0]); - - inVec1_3210 = _mm_mullo_epi32(inVec1_3210, inVec2_3210); - - acc1 = _mm_add_epi32(acc1, inVec1_3210); - i += 4; - } - - acc1 = _mm_add_epi32(acc1, _mm_unpackhi_epi64(acc1, acc1)); - acc1 = _mm_add_epi32(acc1, _mm_shufflelo_epi16(acc1, 0x0E)); - - sum += _mm_cvtsi128_si32(acc1); - - for (;i<N;i++) - { - sum = silk_SMLABB(sum, x[i], y[i]); - } +#include <xmmintrin.h> +#include "arch.h" - return sum; +void xcorr_kernel_sse(const opus_val16 *x, const opus_val16 *y, opus_val32 sum[4], int len) +{ + int j; + __m128 xsum1, xsum2...

The attached patch cleans up Opus's x86 intrinsics configury. It: * Makes ?enable-intrinsics work with clang and other non-GCC compilers * Enables RTCD for the floating-point-mode SSE code in Celt. * Disables use of RTCD in cases where the compiler targets an instruction set by default. * Enables the SSE4.1 Silk optimizations that apply to the common parts of Silk when Opus is built in

Hi All, Since I continue to base my work on top of Jonathan's patch, and my previous Ne10 fft/ifft/mdct_forward/backward patches, I thought it would be better to just post all new patches as a patch series. Please let me know if anyone disagrees with this approach. You can see wip branch of all latest patches at https://git.linaro.org/people/viswanath.puttagunta/opus.git Branch:

Hi Timothy, As I mentioned earlier [1], I now fixed compile issues with fixed point and resubmitting the patch. I also have new patch that does intrinsics optimizations for celt_pitch_xcorr targetting aarch64. You can find my latest work-in-progress branch at [2] For reference, you can use the Ne10 pre-built libraries at [3] Note that I am working with Phil at ARM to get my patch at [4]

Hi All, As per Timothy's suggestion, disabling mdct_forward for fixed point. Only effects armv7,armv8: Extend fixed fft NE10 optimizations to mdct Rest of patches are same as in [1] For reference, latest wip code for opus is at [2] Still working with NE10 team at ARM to get corner cases of mdct_forward. Will update with another patch when issue in NE10 gets fixed. Regards, Vish [1]:

Hi All, Changes from RFC v2 [1] armv7,armv8: Extend fixed fft NE10 optimizations to mdct - Overflow issue fixed by Phil at ARM. Ne10 wip at [2]. Should be upstream soon. - So, re-enabled using fixed fft for mdct_forward which was disabled in RFCv2 armv7,armv8: Optimize fixed point fft using NE10 library - Thanks to Jonathan Lennox, fixed some build fixes on iOS and some copy-paste errors Rest

Hello Timothy / Jean-Marc / opus-dev, This patch series is follow up on work I posted on [1]. In addition to what was posted on [1], this patch series mainly integrates Fixed point FFT implementations in NE10 library into opus. You can view my opus wip code at [2]. Note that while I found some issues both with the NE10 library(fixed fft) and with Linaro toolchain (armv8 intrinsics), the work