Skip to content
GitLab
Commit 49b9dc91 authored by Benedikt Zoennchen's avatar Benedikt Zoennchen
Browse files

extend CLDistMesh from CLOperation.

parent 27da0fa3
Pipeline #264552 failed with stages
in 1 minute and 50 seconds
Hide whitespace changes
Inline Side-by-side
VadereMeshing/src/org/vadere/meshing/mesh/triangulation/improver/eikmesh/opencl/CLEikMesh.java
View file @ 49b9dc91
......@@ -196,7 +196,12 @@ public class CLEikMesh implements IMeshImprover<AVertex, AHalfEdge, AFace>, ITri
}
numberOfIterations++;
return clDistMesh.step(flipAll);
try {
return clDistMesh.step(flipAll);
} catch (OpenCLException e) {
e.printStackTrace();
return false;
}
/*log.info("#illegalMovementTests: " + numberOfIllegalMovementTests);
log.info("#retriangulations: " + numberOfRetriangulations);
log.info("#steps: " + numberOfIterations);
......@@ -204,7 +209,11 @@ public class CLEikMesh implements IMeshImprover<AVertex, AHalfEdge, AFace>, ITri
}
public void finish() {
clDistMesh.finish();
try {
clDistMesh.finish();
} catch (OpenCLException e) {
e.printStackTrace();
}
}
public void refresh() {
......
VadereMeshing/src/org/vadere/meshing/mesh/triangulation/plots/VisualTestGPUEdgeBased.java
View file @ 49b9dc91
......@@ -56,21 +56,23 @@ public class VisualTestGPUEdgeBased {
overAllTime.resume();
meshGenerator.improve();
overAllTime.suspend();
meshGenerator.refresh();
synchronized (meshGenerator.getMesh()) {
meshGenerator.refresh();
}
try {
Thread.sleep(100);
Thread.sleep(10);
} catch (InterruptedException e) {
e.printStackTrace();
}
distmeshPanel.repaint();
}
overAllTime.stop();
log.info("#vertices: " + meshGenerator.getMesh().getVertices().size());
log.info("#edges: " + meshGenerator.getMesh().getEdges().size());
log.info("#faces: " + meshGenerator.getMesh().getFaces().size());
log.info("quality: " + meshGenerator.getQuality());
log.info("overall time: " + overAllTime.getTime() + "[ms]");
meshGenerator.finish();
System.out.println("#vertices: " + meshGenerator.getMesh().getVertices().size());
System.out.println("#edges: " + meshGenerator.getMesh().getEdges().size());
System.out.println("#faces: " + meshGenerator.getMesh().getFaces().size());
System.out.println("quality: " + meshGenerator.getQuality());
System.out.println("overall time: " + overAllTime.getTime() + "[ms]");
}
......
VadereMeshing/src/org/vadere/meshing/mesh/triangulation/plots/VisualTestGPUVertexBased.java
View file @ 49b9dc91
......@@ -66,11 +66,11 @@ public class VisualTestGPUVertexBased {
overAllTime.stop();
meshGenerator.finish();
log.info("#vertices: " + meshGenerator.getMesh().getVertices().size());
log.info("#edges: " + meshGenerator.getMesh().getEdges().size());
log.info("#faces: " + meshGenerator.getMesh().getFaces().size());
log.info("quality: " + meshGenerator.getQuality());
log.info("overall time: " + overAllTime.getTime() + "[ms]");
System.out.println("#vertices: " + meshGenerator.getMesh().getVertices().size());
System.out.println("#edges: " + meshGenerator.getMesh().getEdges().size());
System.out.println("#faces: " + meshGenerator.getMesh().getFaces().size());
System.out.println("quality: " + meshGenerator.getQuality());
System.out.println("overall time: " + overAllTime.getTime() + "[ms]");
}
public static void main(String[] args) throws OpenCLException {
......
VadereMeshing/src/org/vadere/meshing/opencl/CLDistMesh.java
View file @ 49b9dc91
......@@ -2,6 +2,7 @@ package org.vadere.meshing.opencl;
import org.apache.commons.lang3.time.StopWatch;
import org.lwjgl.system.Configuration;
import org.vadere.util.logging.Logger;
import org.jetbrains.annotations.NotNull;
import org.lwjgl.PointerBuffer;
......@@ -16,6 +17,7 @@ import org.vadere.meshing.mesh.gen.AVertex;
import org.vadere.meshing.mesh.gen.CLGatherer;
import org.vadere.util.geometry.shapes.IPoint;
import org.vadere.util.opencl.CLInfo;
import org.vadere.util.opencl.CLOperation;
import org.vadere.util.opencl.CLUtils;
import org.vadere.util.opencl.OpenCLException;
......@@ -28,28 +30,19 @@ import java.util.*;
import java.util.stream.Collectors;
import static org.lwjgl.opencl.CL10.*;
import static org.lwjgl.opencl.CL10.clEnqueueNDRangeKernel;
import static org.lwjgl.opencl.CL11.CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE;
import static org.lwjgl.system.MemoryStack.stackPush;
import static org.lwjgl.system.MemoryUtil.NULL;
import static org.lwjgl.system.MemoryUtil.memUTF8;
/**
* @author Benedikt Zoennchen
*
* DistMesh GPU implementation.
*/
public class CLDistMesh {
public class CLDistMesh extends CLOperation {
private static Logger log = Logger.getLogger(CLDistMesh.class);
// CL ids
private long clPlatform;
private long clDevice;
private long clContext;
private long clQueue;
private long clProgram;
// CL kernel ids
private long clKernelForces;
private long clKernelMove;
......@@ -70,10 +63,6 @@ public class CLDistMesh {
//private long clKernelRemoveTriangles;
//private long clKernelCheckTriangles;
// CL callbacks
private CLContextCallback contextCB;
private CLProgramCallback programCB;
// data on the host
private DoubleBuffer vD;
......@@ -139,19 +128,20 @@ public class CLDistMesh {
private AMesh mesh;
private boolean doublePrecision = true;
private boolean profiling = true;
private boolean doublePrecision = false;
private List<IPoint> result;
private boolean hasToRead = false;
public CLDistMesh(@NotNull AMesh mesh) {
/*if(profiling) {
public CLDistMesh(@NotNull final AMesh mesh) {
super(CL_DEVICE_TYPE_CPU);
profiling = true;
if(profiling) {
Configuration.DEBUG.set(true);
Configuration.DEBUG_MEMORY_ALLOCATOR.set(true);
Configuration.DEBUG_STACK.set(true);
Configuration.DEBUG_STACK.set(true);
}*/
}
this.mesh = mesh;
this.mesh.garbageCollection();
......@@ -188,126 +178,21 @@ public class CLDistMesh {
this.result = mesh.streamPoints().collect(Collectors.toList());
}
private void initCallbacks() {
contextCB = CLContextCallback.create((errinfo, private_info, cb, user_data) ->
{
log.warn("[LWJGL] cl_context_callback");
log.warn("\tInfo: " + memUTF8(errinfo));
});
programCB = CLProgramCallback.create((program, user_data) ->
{
try {
log.info("The cl_program [0x"+program+"] was built " + (CLInfo.getProgramBuildInfoInt(program, clDevice, CL_PROGRAM_BUILD_STATUS) == CL_SUCCESS ? "successfully" : "unsuccessfully"));
String message = CLInfo.getProgramBuildInfoStringASCII(program, clDevice, CL_PROGRAM_BUILD_LOG);
if (!message.isEmpty()) {
log.info("BUILD LOG:\n----\n"+message+"\n-----");
}
}
catch (OpenCLException e) {
log.error(e.getMessage());
}
});
}
private void initCL() throws OpenCLException {
try (MemoryStack stack = stackPush()) {
// helper for the memory allocation in java
//stack = MemoryStack.stackPush();
IntBuffer errcode_ret = stack.callocInt(1);
IntBuffer numberOfPlatforms = stack.mallocInt(1);
clGetPlatformIDs(null, numberOfPlatforms);
PointerBuffer platformIDs = stack.mallocPointer(numberOfPlatforms.get(0));
clGetPlatformIDs(platformIDs, numberOfPlatforms);
clPlatform = platformIDs.get(0);
IntBuffer numberOfDevices = stack.mallocInt(1);
clGetDeviceIDs(clPlatform, CL_DEVICE_TYPE_GPU, null, numberOfDevices);
PointerBuffer deviceIDs = stack.mallocPointer(numberOfDevices.get(0));
clGetDeviceIDs(clPlatform, CL_DEVICE_TYPE_GPU, deviceIDs, numberOfDevices);
clDevice = deviceIDs.get(0);
printDeviceInfo(clDevice, "CL_DEVICE_NAME", CL_DEVICE_NAME);
PointerBuffer ctxProps = stack.mallocPointer(3);
ctxProps.put(CL_CONTEXT_PLATFORM)
.put(clPlatform)
.put(NULL)
.flip();
clContext = clCreateContext(ctxProps, clDevice, contextCB, NULL, errcode_ret);
CLInfo.checkCLError(errcode_ret);
if(profiling) {
clQueue = clCreateCommandQueue(clContext, clDevice, CL_QUEUE_PROFILING_ENABLE, errcode_ret);
}
else {
clQueue = clCreateCommandQueue(clContext, clDevice, 0, errcode_ret);
}
CLInfo.checkCLError(errcode_ret);
PointerBuffer pp = stack.mallocPointer(1);
clGetDeviceInfo(clDevice, CL_DEVICE_MAX_WORK_GROUP_SIZE, pp, null);
maxGroupSize = pp.get(0);
clGetDeviceInfo(clDevice, CL_DEVICE_MAX_COMPUTE_UNITS, pp, null);
maxComputeUnits = pp.get(0);
log.info("MAX_GRP_SIZE = " + maxGroupSize);
log.info("MAX_COMPUTE_UNITS = " + maxComputeUnits);
}
}
private void buildProgram() throws OpenCLException {
try (MemoryStack stack = stackPush()) {
// helper for the memory allocation in java
//stack = MemoryStack.stackPush();
IntBuffer errcode_ret = stack.mallocInt(1);
IntBuffer numberOfPlatforms = stack.mallocInt(1);
clGetPlatformIDs(null, numberOfPlatforms);
PointerBuffer platformIDs = stack.mallocPointer(numberOfPlatforms.get(0));
clGetPlatformIDs(platformIDs, numberOfPlatforms);
clPlatform = platformIDs.get(0);
IntBuffer numberOfDevices = stack.mallocInt(1);
clGetDeviceIDs(clPlatform, CL_DEVICE_TYPE_GPU, null, numberOfDevices);
PointerBuffer deviceIDs = stack.mallocPointer(numberOfDevices.get(0));
clGetDeviceIDs(clPlatform, CL_DEVICE_TYPE_GPU, deviceIDs, numberOfDevices);
clDevice = deviceIDs.get(0);
printDeviceInfo(clDevice, "CL_DEVICE_NAME", CL_DEVICE_NAME);
PointerBuffer ctxProps = stack.mallocPointer(3);
ctxProps.put(CL_CONTEXT_PLATFORM)
.put(clPlatform)
.put(NULL)
.flip();
clContext = clCreateContext(ctxProps, clDevice, contextCB, NULL, errcode_ret);
CLInfo.checkCLError(errcode_ret);
if (profiling) {
clQueue = clCreateCommandQueue(clContext, clDevice, CL_QUEUE_PROFILING_ENABLE, errcode_ret);
} else {
clQueue = clCreateCommandQueue(clContext, clDevice, 0, errcode_ret);
}
CLInfo.checkCLError(errcode_ret);
PointerBuffer pp = stack.mallocPointer(1);
clGetDeviceInfo(clDevice, CL_DEVICE_MAX_WORK_GROUP_SIZE, pp, null);
maxGroupSize = pp.get(0);
clGetDeviceInfo(clDevice, CL_DEVICE_MAX_COMPUTE_UNITS, pp, null);
maxComputeUnits = pp.get(0);
log.info("MAX_GRP_SIZE = " + maxGroupSize);
log.info("MAX_COMPUTE_UNITS = " + maxComputeUnits);
PointerBuffer clProgramStrings = stack.mallocPointer(1);
PointerBuffer clProgramLengths = stack.mallocPointer(1);
PointerBuffer pp = stack.mallocPointer(1);
clGetDeviceInfo(clDevice, CL_DEVICE_MAX_WORK_GROUP_SIZE, pp, null);
maxGroupSize = pp.get(0);
clGetDeviceInfo(clDevice, CL_DEVICE_MAX_COMPUTE_UNITS, pp, null);
maxComputeUnits = pp.get(0);
log.info("MAX_GRP_SIZE = " + maxGroupSize);
log.info("MAX_COMPUTE_UNITS = " + maxComputeUnits);
PointerBuffer clProgramStrings = stack.mallocPointer(1);
PointerBuffer clProgramLengths = stack.mallocPointer(1);
try {
if (doublePrecision) {
......@@ -319,7 +204,6 @@ public class CLDistMesh {
throw new RuntimeException(e);
}
clProgramStrings.put(0, source);
clProgramLengths.put(0, source.remaining());
......@@ -554,12 +438,12 @@ public class CLDistMesh {
}
}*/
public boolean step() {
public boolean step() throws OpenCLException {
return step(true);
}
// TODO: think about the use of only 1 work-group!!! It might be bad! solution: use global barrier? force computation? flip hangs after some time?
public boolean step(final boolean flipAll) {
public boolean step(final boolean flipAll) throws OpenCLException {
try (MemoryStack stack = stackPush()) {
/*
* DistMesh-Loop
......@@ -646,25 +530,6 @@ public class CLDistMesh {
}
}
private long enqueueNDRangeKernel(final String name, long command_queue, long kernel, int work_dim, PointerBuffer global_work_offset, PointerBuffer global_work_size, PointerBuffer local_work_size, PointerBuffer event_wait_list, PointerBuffer event) {
if(profiling) {
long result = clEnqueueNDRangeKernel(command_queue, kernel, work_dim, global_work_offset, global_work_size, local_work_size, event_wait_list, clEvent);
clWaitForEvents(clEvent);
long eventAddr = clEvent.get();
clGetEventProfilingInfo(eventAddr, CL_PROFILING_COMMAND_START, startTime, retSize);
clGetEventProfilingInfo(eventAddr, CL_PROFILING_COMMAND_END, endTime, retSize);
clEvent.clear();
// in nanaSec
log.info(name + " event time " + "0x"+eventAddr + ": " + (endTime.getLong() - startTime.getLong()) + " ns");
endTime.clear();
startTime.clear();
return result;
}
else {
return clEnqueueNDRangeKernel(command_queue, kernel, work_dim, global_work_offset, global_work_size, local_work_size, event_wait_list, event);
}
}
private void checkTriLocks() {
for(int i = 0; i < numberOfFaces; i++) {
int lock = triLocks.get(i);
......@@ -861,7 +726,8 @@ public class CLDistMesh {
return tmp;
}
private void clearCL() {
@Override
protected void clearCL() throws OpenCLException {
clReleaseMemObject(clVertices);
clReleaseMemObject(clEdges);
clReleaseMemObject(clTwins);
......@@ -888,13 +754,7 @@ public class CLDistMesh {
clReleaseKernel(clKernelRepair);
clReleaseKernel(clKernelLabelEdges);
clReleaseKernel(clKernelLabelEdgesUpdate);
clReleaseProgram(clProgram);
clReleaseCommandQueue(clQueue);
clReleaseContext(clContext);
contextCB.close();
programCB.close();
super.clearCL();
}
private void clearHost() {
......@@ -955,11 +815,10 @@ public class CLDistMesh {
//printResult();
}
public void finish() {
public void finish() throws OpenCLException {
refresh();
clearHost();
clearCL();
}
private void printTri() {
......
VadereMeshing/src/org/vadere/meshing/opencl/CLDistMeshHE.java
View file @ 49b9dc91
......@@ -124,7 +124,7 @@ public class CLDistMeshHE {
private AMesh mesh;
private boolean doublePrecision = true;
private boolean doublePrecision = false;
private boolean profiling = false;
private boolean hasToRead = false;
......@@ -189,9 +189,9 @@ public class CLDistMeshHE {
clGetPlatformIDs(platformIDs, numberOfPlatforms);
clPlatform = platformIDs.get(0);
clGetDeviceIDs(clPlatform, CL_DEVICE_TYPE_GPU, null, numberOfDevices);
clGetDeviceIDs(clPlatform, CL_DEVICE_TYPE_CPU, null, numberOfDevices);
PointerBuffer deviceIDs = stack.mallocPointer(numberOfDevices.get(0));
clGetDeviceIDs(clPlatform, CL_DEVICE_TYPE_GPU, deviceIDs, numberOfDevices);
clGetDeviceIDs(clPlatform, CL_DEVICE_TYPE_CPU, deviceIDs, numberOfDevices);
clDevice = deviceIDs.get(0);
printDeviceInfo(clDevice, "CL_DEVICE_NAME", CL_DEVICE_NAME);
......
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment

LRZ Homepage | Datenschutz | Dokumentation und Betriebsbedingungen | Impressum