search for: flac__real

The following series adds initial vector support for PowerPC64. On POWER9, flac --best is about 3.3x faster. Amitay Isaacs (2): Add m4 macro to check for C __attribute__ features Check if compiler supports target attribute on ppc64 Anton Blanchard (5): configure.ac: Remove SPE detection code configure.ac: Add VSX enable/disable configure.ac: Fix FLAC__CPU_PPC on little endian, and add

...aid I'm not familiar with assembler so please check it. --- a\src\libFLAC\include\private\lpc.h 2013-08-13 13:30:24.000000000 +0400 +++ b\src\libFLAC\include\private\lpc.h 2013-08-22 20:11:17.524302900 +0400 @@ -75,6 +75,7 @@ void FLAC__lpc_compute_autocorrelation_asm_ia32_sse_lag_4(const FLAC__real data[], unsigned data_len, unsigned lag, FLAC__real autoc[]); void FLAC__lpc_compute_autocorrelation_asm_ia32_sse_lag_8(const FLAC__real data[], unsigned data_len, unsigned lag, FLAC__real autoc[]); void FLAC__lpc_compute_autocorrelation_asm_ia32_sse_lag_12(const FLAC__real data[], unsigned dat...

Hi, I was doing some speed and compression comparisons with various apodization/windowing functions, and found out that the definitions for the bartlett and bartlett_hann window in the FLAC codebase have been wrong since their introduction. The attached patch fixes that. Furthermore, I found some peculiar behaviour of the gauss apodization that seems to expose bug. Using different windows

...ime spend in the > function FLAC__lpc_compute_autocorrelation_intrin_sse_lag_16 > increases 20-fold while the number of calls stays the same. > Could this be a bug? <http://en.wikipedia.org/wiki/Denormal_number#Performance_issues> ? Try something like this: void FLAC__window_gauss(FLAC__real *window, const FLAC__int32 L, const FLAC__real stddev) { static const double anti_denormal = 0.88817841970012523233890533447266e-15; /* 2e-50 */ const FLAC__int32 N = L - 1; const double N2 = (double)N / 2.; FLAC__int32 n; for (n = 0; n <= N; n++) { const double k = ((double)n - N2) / (s...

...jne .not_order_1 - mov ebp, 1 - jmp short .got_order -.not_order_1: - cmp eax, edx - jne .not_order_2 - mov ebp, 2 - jmp short .got_order -.not_order_2: - cmp eax, esi - jne .not_order_3 - mov ebp, 3 - jmp short .got_order -.not_order_3: - mov ebp, 4 -.got_order: ; residual_bits_per_sample[0] = (FLAC__real)((data_len > 0 && total_error_0 > 0) ? log(M_LN2 * (double)total_error_0 / (double)data_len) / M_LN2 : 0.0); ; residual_bits_per_sample[1] = (FLAC__real)((data_len > 0 && total_error_1 > 0) ? log(M_LN2 * (double)total_error_1 / (double)data_len) / M_LN2 : 0.0); ; r...

...f defined(FLAC__CPU_IA32) && defined(FLAC__HAS_NASM) I'm not sure if there are portability issues with the above, though. --- orig/src/libFLAC/include/private/lpc.h +++ mod/src/libFLAC/include/private/lpc.h @@ -61,6 +61,7 @@ void FLAC__lpc_compute_autocorrelation_asm_ia32_3dnow(const FLAC__real data[], unsigned data_len, unsigned lag, FLAC__real autoc[]); #endif #endif +#endif /* * FLAC__lpc_compute_lp_coefficients() @@ -146,17 +147,16 @@ void FLAC__lpc_restore_signal(const FLAC__int32 residual[], unsigned data_len, const FLAC__int32 qlp_coeff[], unsigned order, int lp_quantizatio...

...ined(__GNUC__) +#define copysign __builtin_copysign +#endif +static inline long int lround(double x) { + return (long)(x + copysign (0.5, x)); +} +//If this fails, we are in the precence of a mid 90's compiler..move along... +#endif void FLAC__lpc_window_data(const FLAC__int32 in[], const FLAC__real window[], FLAC__real out[], unsigned data_len) { @@ -199,14 +212,8 @@ int FLAC__lpc_quantize_coefficients(const FLAC__real lp_coeff[], unsigned order, FLAC__int32 q; for(i = 0; i < order; i++) { error += lp_coeff[i] * (1 << *shift); -#if 1 /* unfortunately lround() is C99 */ -...

...cros.h" -#if defined DEBUG || defined FLAC__OVERFLOW_DETECT || defined FLAC__OVERFLOW_DETECT_VERBOSE +#if !defined(NDEBUG) || defined FLAC__OVERFLOW_DETECT || defined FLAC__OVERFLOW_DETECT_VERBOSE #include <stdio.h> #endif @@ -234,7 +234,7 @@ int FLAC__lpc_quantize_coefficients(const FLAC__real lp_coeff[], unsigned order, const int nshift = -(*shift); double error = 0.0; FLAC__int32 q; -#ifdef DEBUG +#ifndef NDEBUG fprintf(stderr,"FLAC__lpc_quantize_coefficients: negative shift=%d order=%u cmax=%f\n", *shift, order, cmax); #endif for(i = 0; i < order; i++) { d...

...cros.h" -#if defined DEBUG || defined FLAC__OVERFLOW_DETECT || defined FLAC__OVERFLOW_DETECT_VERBOSE +#if !defined(NDEBUG) || defined FLAC__OVERFLOW_DETECT || defined FLAC__OVERFLOW_DETECT_VERBOSE #include <stdio.h> #endif @@ -234,7 +234,7 @@ int FLAC__lpc_quantize_coefficients(const FLAC__real lp_coeff[], uint32_t order, const int nshift = -(*shift); double error = 0.0; FLAC__int32 q; -#ifdef DEBUG +#ifndef NDEBUG fprintf(stderr,"FLAC__lpc_quantize_coefficients: negative shift=%d order=%u cmax=%f\n", *shift, order, cmax); #endif for(i = 0; i < order; i++) { d...

Hi Erik, I've found a middleground for the problem of setting default CFLAGS. I've gone back to setting them if {C,CXX,CPP,LD}FLAGS are unset at the onset of the configure script (i.e., the user hasn't specified anything) and then proceed to set them to the defaults as before. This has been suggested before: https://lists.gnu.org/archive/html/autoconf/2006-04/msg00022.html In

Dear FLAC devs, I would like to get some of our patches merged into master. Most of them deal with adhering to GNU conventions, respectively not overriding flags/variables that are up to the user to set. For instance, honoring $(htmldir) is important, as we have installation paths for the documentation that differ from what is currently coded in the various Makefile.am's. Regards David