diff --git a/Makefile b/Makefile index 5ac4ae52d..ef98391ef 100644 --- a/Makefile +++ b/Makefile @@ -7,9 +7,9 @@ PY_INCPATH := $(shell $(PY_BIN) -c "import os.path,sys;print os.path.join(sys.ex PY_LDFLAGS := $(shell $(PY_BIN) -c "import distutils.sysconfig;print distutils.sysconfig.get_config_var('LINKFORSHARED')") PYCOMMAND := $(PY_BIN) getheader.py -CFLAGS := -w -std=c99 -D_POSIX_C_SOURCE=200112L -DLUACONSOLE -Iincludes/ +CFLAGS := -w -std=c99 -D_POSIX_C_SOURCE=200112L -DLUACONSOLE -DGRAVFFT -Iincludes/ OFLAGS := -O3 -ffast-math -ftree-vectorize -funsafe-math-optimizations -LFLAGS := -lpthread -lSDL -lm -lbz2 -lX11 -llua5.1 #-lpython$(PY_VERSION) -L$(PY_LIBPATH) -I$(PY_INCPATH) $(PY_LDFLAGS) +LFLAGS := -lpthread -lSDL -lfftw3f -lm -lbz2 -lX11 -llua5.1 #-lpython$(PY_VERSION) -L$(PY_LIBPATH) -I$(PY_INCPATH) $(PY_LDFLAGS) LFLAGS_X := -lm -lbz2 -lSDLmain -I/Library/Frameworks/Python.framework/Versions/$(PY_VERSION)/include/python$(PY_VERSION) MFLAGS_SSE3 := -march=native -DX86 -DX86_SSE3 -msse3 MFLAGS_SSE2 := -march=native -DX86 -DX86_SSE2 -msse2 diff --git a/includes/air.h b/includes/air.h index b1dc819bc..0f3731054 100644 --- a/includes/air.h +++ b/includes/air.h @@ -41,6 +41,11 @@ void update_airh(void); void update_grav(void); +#ifdef GRAVFFT +void grav_fft_init(); +void grav_fft_cleanup(); +#endif + void update_air(void); #endif diff --git a/includes/defines.h b/includes/defines.h index 3f3760e9c..7c0072209 100644 --- a/includes/defines.h +++ b/includes/defines.h @@ -92,6 +92,7 @@ extern unsigned char ZSIZE; #define BRUSH_NUM 3 +//#define GRAVFFT //#define LUACONSOLE //#define PYCONSOLE //#define PYEXT diff --git a/includes/powder.h b/includes/powder.h index 5299dc06e..6abb47ba1 100644 --- a/includes/powder.h +++ b/includes/powder.h @@ -786,30 +786,30 @@ typedef struct gol_menu gol_menu; static gol_menu gmenu[NGOL] = { - {"GOL", PIXPACK(0x0CAC00), 0, "GOL"}, - {"HLIF", PIXPACK(0xFF0000), 1, "GOL"}, - {"ASIM", PIXPACK(0x0000FF), 2, "GOL"}, - {"2x2", PIXPACK(0xFFFF00), 3, "GOL"}, - {"DANI", PIXPACK(0x00FFFF), 4, "GOL"}, - {"AMOE", PIXPACK(0xFF00FF), 5, "GOL"}, - {"MOVE", PIXPACK(0xFFFFFF), 6, "GOL"}, - {"PGOL", PIXPACK(0xE05010), 7, "GOL"}, - {"DMOE", PIXPACK(0x500000), 8, "GOL"}, - {"34", PIXPACK(0x500050), 9, "GOL"}, - {"LLIF", PIXPACK(0x505050), 10, "GOL"}, - {"STAN", PIXPACK(0x5000FF), 11, "GOL"}, - {"SEED", PIXPACK(0xFBEC7D), 12, "GOL"}, - {"MAZE", PIXPACK(0xA8E4A0), 13, "GOL"}, - {"COAG", PIXPACK(0x9ACD32), 14, "GOL"}, - {"WALL", PIXPACK(0x0047AB), 15, "GOL"}, - {"GNAR", PIXPACK(0xE5B73B), 16, "GOL"}, - {"REPL", PIXPACK(0x259588), 17, "GOL"}, - {"MYST", PIXPACK(0x0C3C00), 18, "GOL"}, - {"LOTE", PIXPACK(0xFF0000), 19, "GOL"}, - {"FRG2", PIXPACK(0x00FF00), 20, "GOL"}, - {"STAR", PIXPACK(0x0000FF), 21, "GOL"}, - {"FROG", PIXPACK(0x00AA00), 22, "GOL"}, - {"BRAN", PIXPACK(0xCCCC00), 23, "GOL"} + {"GOL", PIXPACK(0x0CAC00), 0, "Game Of Life B3/S23"}, + {"HLIF", PIXPACK(0xFF0000), 1, "High Life B36/S23"}, + {"ASIM", PIXPACK(0x0000FF), 2, "Assimilation B345/S4567"}, + {"2x2", PIXPACK(0xFFFF00), 3, "2x2 B36/S125"}, + {"DANI", PIXPACK(0x00FFFF), 4, "Day and Night B3678/S34678"}, + {"AMOE", PIXPACK(0xFF00FF), 5, "Amoeba B357/S1358"}, + {"MOVE", PIXPACK(0xFFFFFF), 6, "'Move' particles. Does not move things.. it is a life type B368/S245"}, + {"PGOL", PIXPACK(0xE05010), 7, "Pseudo Life B357/S238"}, + {"DMOE", PIXPACK(0x500000), 8, "Diamoeba B35678/S5678"}, + {"34", PIXPACK(0x500050), 9, "34 B34/S34"}, + {"LLIF", PIXPACK(0x505050), 10, "Long Life B345/S5"}, + {"STAN", PIXPACK(0x5000FF), 11, "Stains B3678/S235678"}, + {"SEED", PIXPACK(0xFBEC7D), 12, "B2/S"}, + {"MAZE", PIXPACK(0xA8E4A0), 13, "B3/S12345"}, + {"COAG", PIXPACK(0x9ACD32), 14, "B378/S235678"}, + {"WALL", PIXPACK(0x0047AB), 15, "B45678/S2345"}, + {"GNAR", PIXPACK(0xE5B73B), 16, "B1/S1"}, + {"REPL", PIXPACK(0x259588), 17, "B1357/S1357"}, + {"MYST", PIXPACK(0x0C3C00), 18, "B3458/S05678"}, + {"LOTE", PIXPACK(0xFF0000), 19, "Behaves kinda like Living on the Edge S3458/B37/4"}, + {"FRG2", PIXPACK(0x00FF00), 20, "Like Frogs rule S124/B3/3"}, + {"STAR", PIXPACK(0x0000FF), 21, "Like Star Wars rule S3456/B278/6"}, + {"FROG", PIXPACK(0x00AA00), 22, "Frogs S12/B34/3"}, + {"BRAN", PIXPACK(0xCCCC00), 23, "Brian 6 S6/B246/3"} }; static int grule[NGOL+1][10] = diff --git a/src/air.c b/src/air.c index 1b83665b3..45d000a17 100644 --- a/src/air.c +++ b/src/air.c @@ -2,6 +2,11 @@ #include #include #include + +#ifdef GRAVFFT +#include +#endif + float kernel[9]; float gravmap[YRES/CELL][XRES/CELL]; //Maps to be used by the main thread @@ -154,6 +159,159 @@ void bilinear_interpolation(float *src, float *dst, int sw, int sh, int rw, int } } +#ifdef GRAVFFT +int grav_fft_status = 0; +float *th_ptgravx, *th_ptgravy, *th_gravmapbig, *th_gravxbig, *th_gravybig; +fftwf_complex *th_ptgravxt, *th_ptgravyt, *th_gravmapbigt, *th_gravxbigt, *th_gravybigt; +fftwf_plan plan_gravmap, plan_gravx_inverse, plan_gravy_inverse; + +void grav_fft_init() +{ + int xblock2 = XRES/CELL*2; + int yblock2 = YRES/CELL*2; + int x, y, fft_tsize = (xblock2/2+1)*yblock2; + float distance, scaleFactor; + fftwf_plan plan_ptgravx, plan_ptgravy; + if (grav_fft_status) return; + + //use fftw malloc function to ensure arrays are aligned, to get better performance + th_ptgravx = fftwf_malloc(xblock2*yblock2*sizeof(float)); + th_ptgravy = fftwf_malloc(xblock2*yblock2*sizeof(float)); + th_ptgravxt = fftwf_malloc(fft_tsize*sizeof(fftwf_complex)); + th_ptgravyt = fftwf_malloc(fft_tsize*sizeof(fftwf_complex)); + th_gravmapbig = fftwf_malloc(xblock2*yblock2*sizeof(float)); + th_gravmapbigt = fftwf_malloc(fft_tsize*sizeof(fftwf_complex)); + th_gravxbig = fftwf_malloc(xblock2*yblock2*sizeof(float)); + th_gravybig = fftwf_malloc(xblock2*yblock2*sizeof(float)); + th_gravxbigt = fftwf_malloc(fft_tsize*sizeof(fftwf_complex)); + th_gravybigt = fftwf_malloc(fft_tsize*sizeof(fftwf_complex)); + + //select best algorithm, could use FFTW_PATIENT or FFTW_EXHAUSTIVE but that increases the time taken to plan, and I don't see much increase in execution speed + plan_ptgravx = fftwf_plan_dft_r2c_2d(yblock2, xblock2, th_ptgravx, th_ptgravxt, FFTW_MEASURE); + plan_ptgravy = fftwf_plan_dft_r2c_2d(yblock2, xblock2, th_ptgravy, th_ptgravyt, FFTW_MEASURE); + plan_gravmap = fftwf_plan_dft_r2c_2d(yblock2, xblock2, th_gravmapbig, th_gravmapbigt, FFTW_MEASURE); + plan_gravx_inverse = fftwf_plan_dft_c2r_2d(yblock2, xblock2, th_gravxbigt, th_gravxbig, FFTW_MEASURE); + plan_gravy_inverse = fftwf_plan_dft_c2r_2d(yblock2, xblock2, th_gravybigt, th_gravybig, FFTW_MEASURE); + + //(XRES/CELL)*(YRES/CELL)*4 is size of data array, scaling needed because FFTW calculates an unnormalized DFT + scaleFactor = -M_GRAV/((XRES/CELL)*(YRES/CELL)*4); + //calculate velocity map caused by a point mass + for (y=0; y 0) + menuid = luaL_optint(l, 1, -1); + if (menuid < SC_TOTAL && menuid >= 0) active_menu = menuid; else return luaL_error(l, "Invalid menu"); diff --git a/src/main.c b/src/main.c index 83c03d290..3a0203045 100644 --- a/src/main.c +++ b/src/main.c @@ -1450,6 +1450,9 @@ void* update_grav_async(void* unused) memset(th_gravmap, 0, sizeof(th_gravmap)); memset(th_gravy, 0, sizeof(th_gravy)); memset(th_gravx, 0, sizeof(th_gravx)); +#ifdef GRAVFFT + grav_fft_init(); +#endif while(!thread_done){ if(!done){ update_grav(); @@ -1476,8 +1479,6 @@ void* update_grav_async(void* unused) void start_grav_async() { if(!ngrav_enable){ - /*pthread_mutexattr_t gma; //I do not know why this is here - pthread_mutexattr_init(&gma);*/ gravthread_done = 0; pthread_mutex_init (&gravmutex, NULL); pthread_cond_init(&gravcv, NULL); @@ -3617,6 +3618,9 @@ int main(int argc, char *argv[]) } SDL_CloseAudio(); http_done(); +#ifdef GRAVFFT + grav_fft_cleanup(); +#endif #ifdef LUACONSOLE luacon_close(); #endif diff --git a/src/powder.c b/src/powder.c index dbf9c45ac..a1fa55ea1 100644 --- a/src/powder.c +++ b/src/powder.c @@ -1031,6 +1031,7 @@ static void create_gain_photon(int pp)//photons from PHOT going through GLOW parts[i].vy = parts[pp].vy; parts[i].temp = parts[pmap[ny][nx] >> 8].temp; parts[i].tmp = 0; + parts[i].pavg[0] = parts[i].pavg[1] = 0.0f; photons[ny][nx] = PT_PHOT|(i<<8); temp_bin = (int)((parts[i].temp-273.0f)*0.25f); @@ -1067,6 +1068,7 @@ static void create_cherenkov_photon(int pp)//photons from NEUT going through GLA parts[i].y = parts[pp].y; parts[i].temp = parts[pmap[ny][nx] >> 8].temp; parts[i].tmp = 0; + parts[i].pavg[0] = parts[i].pavg[1] = 0.0f; photons[ny][nx] = PT_PHOT|(i<<8); if (lr) {