package com.mbien.opencl.demos.hellojocl; import com.mbien.opencl.CLBuffer; import com.mbien.opencl.CLCommandQueue; import com.mbien.opencl.CLContext; import com.mbien.opencl.CLKernel; import com.mbien.opencl.CLProgram; import java.io.IOException; import java.nio.FloatBuffer; import java.util.Random; import static java.lang.System.*; import static com.mbien.opencl.CLBuffer.Mem.*; /** * Hello Java OpenCL example. Adds all elements of buffer A to buffer B * and stores the result in buffer C.
* Sample was inspired by the Nvidia VectorAdd example written in C/C++ * which is bundled in the Nvidia OpenCL SDK. * @author Michael Bien */ public class HelloJOCL { public static void main(String[] args) throws IOException { int elementCount = 11444777; // Length of arrays to process int localWorkSize = 256; // Local work size dimensions int globalWorkSize = roundUp(localWorkSize, elementCount); // rounded up to the nearest multiple of the localWorkSize // set up CLContext context = CLContext.create(); CLProgram program = context.createProgram(HelloJOCL.class.getResourceAsStream("VectorAdd.cl")).build(); CLBuffer clBufferA = context.createFloatBuffer(globalWorkSize, READ_ONLY); CLBuffer clBufferB = context.createFloatBuffer(globalWorkSize, READ_ONLY); CLBuffer clBufferC = context.createFloatBuffer(globalWorkSize, WRITE_ONLY); out.println("used device memory: " + (clBufferA.buffer.capacity()+clBufferB.buffer.capacity()+clBufferC.buffer.capacity())*4/1000000 +"MB"); // fill read buffers with random numbers (just to have test data; seed is fixed -> results will not change between runs). fillBuffer(clBufferA.buffer, 12345); fillBuffer(clBufferB.buffer, 67890); // get a reference to the kernel functon with the name 'VectorAdd' and map the buffers to its input parameters. CLKernel kernel = program.getCLKernels().get("VectorAdd"); kernel.setArg(0, clBufferA) .setArg(1, clBufferB) .setArg(2, clBufferC) .setArg(3, elementCount); // create command queue on fastest device. CLCommandQueue queue = context.getMaxFlopsDevice().createCommandQueue(); // asynchronous write of data to GPU device, blocking read later to get the computed results back. long time = nanoTime(); queue.putWriteBuffer(clBufferA, false) .putWriteBuffer(clBufferB, false) .put1DRangeKernel(kernel, 0, globalWorkSize, localWorkSize) .putReadBuffer(clBufferC, true); time = nanoTime() - time; // cleanup all resources associated with this context. context.release(); // print first few elements of the resulting buffer to the console. out.println("a+b=c results snapshot: "); for(int i = 0; i < 10; i++) out.print(clBufferC.buffer.get() + ", "); out.println("...; " + clBufferC.buffer.remaining() + " more"); out.println("computation took: "+(time/1000000)+"ms"); } private static final void fillBuffer(FloatBuffer buffer, int seed) { Random rnd = new Random(seed); while(buffer.remaining() != 0) buffer.put(rnd.nextFloat()*100); buffer.rewind(); } private static final int roundUp(int groupSize, int globalSize) { int r = globalSize % groupSize; if (r == 0) { return globalSize; } else { return globalSize + groupSize - r; } } }1'>21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428


#include "config.h"

#include "nfcfilter.h"

#include "alu.h"


/* Near-field control filters are the basis for handling the near-field effect.
 * The near-field effect is a bass-boost present in the directional components
 * of a recorded signal, created as a result of the wavefront curvature (itself
 * a function of sound distance). Proper reproduction dictates this be
 * compensated for using a bass-cut given the playback speaker distance, to
 * avoid excessive bass in the playback.
 *
 * For real-time rendered audio, emulating the near-field effect based on the
 * sound source's distance, and subsequently compensating for it at output
 * based on the speaker distances, can create a more realistic perception of
 * sound distance beyond a simple 1/r attenuation.
 *
 * These filters do just that. Each one applies a low-shelf filter, created as
 * the combination of a bass-boost for a given sound source distance (near-
 * field emulation) along with a bass-cut for a given control/speaker distance
 * (near-field compensation).
 *
 * Note that it is necessary to apply a cut along with the boost, since the
 * boost alone is unstable in higher-order ambisonics as it causes an infinite
 * DC gain (even first-order ambisonics requires there to be no DC offset for
 * the boost to work). Consequently, ambisonics requires a control parameter to
 * be used to avoid an unstable boost-only filter. NFC-HOA defines this control
 * as a reference delay, calculated with:
 *
 * reference_delay = control_distance / speed_of_sound
 *
 * This means w0 (for input) or w1 (for output) should be set to:
 *
 * wN = 1 / (reference_delay * sample_rate)
 *
 * when dealing with NFC-HOA content. For FOA input content, which does not
 * specify a reference_delay variable, w0 should be set to 0 to apply only
 * near-field compensation for output. It's important that w1 be a finite,
 * positive, non-0 value or else the bass-boost will become unstable again.
 * Also, w0 should not be too large compared to w1, to avoid excessively loud
 * low frequencies.
 */

static const float B[4][3] = {
    {    0.0f                             },
    {    1.0f                             },
    {    3.0f,     3.0f                   },
    { 3.6778f,  6.4595f, 2.3222f          },
  /*{ 4.2076f, 11.4877f, 5.7924f, 9.1401f }*/
};

static void NfcFilterCreate1(struct NfcFilter1 *nfc, const float w0, const float w1)
{
    float b_00, g_0;
    float r;

    nfc->g = 1.0f;
    nfc->coeffs[0] = 1.0f;

    /* Calculate bass-boost coefficients. */
    r = 0.5f * w0;
    b_00 = B[1][0] * r;
    g_0 = 1.0f + b_00;

    nfc->coeffs[0] *= g_0;
    nfc->coeffs[1] = (2.0f * b_00) / g_0;

    /* Calculate bass-cut coefficients. */
    r = 0.5f * w1;
    b_00 = B[1][0] * r;
    g_0 = 1.0f + b_00;

    nfc->g /= g_0;
    nfc->coeffs[0] /= g_0;
    nfc->coeffs[1+1] = (2.0f * b_00) / g_0;
}

static void NfcFilterAdjust1(struct NfcFilter1 *nfc, const float w0)
{
    float b_00, g_0;
    float r;

    r = 0.5f * w0;
    b_00 = B[1][0] * r;
    g_0 = 1.0f + b_00;

    nfc->coeffs[0] = nfc->g * g_0;
    nfc->coeffs[1] = (2.0f * b_00) / g_0;
}


static void NfcFilterCreate2(struct NfcFilter2 *nfc, const float w0, const float w1)
{
    float b_10, b_11, g_1;
    float r;

    nfc->g = 1.0f;
    nfc->coeffs[0] = 1.0f;

    /* Calculate bass-boost coefficients. */
    r = 0.5f * w0;
    b_10 = B[2][0] * r;
    b_11 = B[2][1] * r * r;
    g_1 = 1.0f + b_10 + b_11;

    nfc->coeffs[0] *= g_1;
    nfc->coeffs[1] = ((2.0f * b_10) + (4.0f * b_11)) / g_1;
    nfc->coeffs[2] = (4.0f * b_11) / g_1;

    /* Calculate bass-cut coefficients. */
    r = 0.5f * w1;
    b_10 = B[2][0] * r;
    b_11 = B[2][1] * r * r;
    g_1 = 1.0f + b_10 + b_11;

    nfc->g /= g_1;
    nfc->coeffs[0] /= g_1;
    nfc->coeffs[2+1] = ((2.0f * b_10) + (4.0f * b_11)) / g_1;
    nfc->coeffs[2+2] = (4.0f * b_11) / g_1;
}

static void NfcFilterAdjust2(struct NfcFilter2 *nfc, const float w0)
{
    float b_10, b_11, g_1;
    float r;

    r = 0.5f * w0;
    b_10 = B[2][0] * r;
    b_11 = B[2][1] * r * r;
    g_1 = 1.0f + b_10 + b_11;

    nfc->coeffs[0] = nfc->g * g_1;
    nfc->coeffs[1] = ((2.0f * b_10) + (4.0f * b_11)) / g_1;
    nfc->coeffs[2] = (4.0f * b_11) / g_1;
}


static void NfcFilterCreate3(struct NfcFilter3 *nfc, const float w0, const float w1)
{
    float b_10, b_11, g_1;
    float b_00, g_0;
    float r;

    nfc->g = 1.0f;
    nfc->coeffs[0] = 1.0f;

    /* Calculate bass-boost coefficients. */
    r = 0.5f * w0;
    b_10 = B[3][0] * r;
    b_11 = B[3][1] * r * r;
    g_1 = 1.0f + b_10 + b_11;

    nfc->coeffs[0] *= g_1;
    nfc->coeffs[1] = ((2.0f * b_10) + (4.0f * b_11)) / g_1;
    nfc->coeffs[2] = (4.0f * b_11) / g_1;

    b_00 = B[3][2] * r;
    g_0 = 1.0f + b_00;

    nfc->coeffs[0] *= g_0;
    nfc->coeffs[2+1] = (2.0f * b_00) / g_0;

    /* Calculate bass-cut coefficients. */
    r = 0.5f * w1;
    b_10 = B[3][0] * r;
    b_11 = B[3][1] * r * r;
    g_1 = 1.0f + b_10 + b_11;

    nfc->g /= g_1;
    nfc->coeffs[0] /= g_1;
    nfc->coeffs[3+1] = ((2.0f * b_10) + (4.0f * b_11)) / g_1;
    nfc->coeffs[3+2] = (4.0f * b_11) / g_1;
    
    b_00 = B[3][2] * r;
    g_0 = 1.0f + b_00;

    nfc->g /= g_0;
    nfc->coeffs[0] /= g_0;
    nfc->coeffs[3+2+1] = (2.0f * b_00) / g_0;
}

static void NfcFilterAdjust3(struct NfcFilter3 *nfc, const float w0)
{
    float b_10, b_11, g_1;
    float b_00, g_0;
    float r;

    r = 0.5f * w0;
    b_10 = B[3][0] * r;
    b_11 = B[3][1] * r * r;
    g_1 = 1.0f + b_10 + b_11;

    nfc->coeffs[0] = nfc->g * g_1;
    nfc->coeffs[1] = ((2.0f * b_10) + (4.0f * b_11)) / g_1;
    nfc->coeffs[2] = (4.0f * b_11) / g_1;

    b_00 = B[3][2] * r;
    g_0 = 1.0f + b_00;

    nfc->coeffs[0] *= g_0;