Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

MxM example measured in ns to obtain a better reference on fast GPUs #423

Merged
merged 4 commits into from
May 15, 2024
Merged

MxM example measured in ns to obtain a better reference on fast GPUs #423

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
41de30e
MxM example measured in ns to obtain a better reference on fast GPUs
jjfumero May 15, 2024
d8ab896
Remove default scheduler for MxM example
jjfumero May 15, 2024
27db842
Update tornado-examples/src/main/java/uk/ac/manchester/tornado/exampl…
jjfumero May 15, 2024
9a55c3e
Update tornado-examples/src/main/java/uk/ac/manchester/tornado/exampl…
jjfumero May 15, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
31 changes: 16 additions & 15 deletions ...mples/src/main/java/uk/ac/manchester/tornado/examples/compute/MatrixMultiplication2D.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,6 +36,7 @@
* $ tornado --threadInfo --jvm="-Ds0.t0.device=0:0" -m tornado.examples/uk.ac.manchester.tornado.examples.compute.MatrixMultiplication2D
* </code>
*/

public class MatrixMultiplication2D {
// CHECKSTYLE:OFF
private static final int WARMING_UP_ITERATIONS = 15;
Expand Down Expand Up @@ -105,9 +106,9 @@ public static void main(String[] args) {
}

// 2. Run parallel on the GPU with Tornado
long start = System.currentTimeMillis();
long start = System.nanoTime();
executor.execute();
long end = System.currentTimeMillis();
long end = System.nanoTime();

// Run sequential
// 1. Warm up sequential
Expand All @@ -116,39 +117,39 @@ public static void main(String[] args) {
}

// 2. Run the sequential code
long startSequential = System.currentTimeMillis();
long startSequential = System.nanoTime();
matrixMultiplication(matrixA, matrixB, resultSeq, size);
long endSequential = System.currentTimeMillis();
long endSequential = System.nanoTime();

// Run multithread
for (int i = 0; i < WARMING_UP_ITERATIONS; i++) {
parallelStreamsMxM(matrixA, matrixB, resultSeq, size);
}

// 2. Run the sequential code
long startStream = System.currentTimeMillis();
long startStream = System.nanoTime();
parallelStreamsMxM(matrixA, matrixB, resultSeq, size);
long endStream = System.currentTimeMillis();
long endStream = System.nanoTime();

// Compute Gigaflops and performance
long msecGPUElapsedTime = (end - start);
long msecCPUElaptedTime = (endSequential - startSequential);
long msecStreamElaptedTime = (endStream - startStream);
long nanoSecGPUElapsedTime = (end - start);
long nanoSecCPUElaptedTime = (endSequential - startSequential);
long nanoSecStreamElaptedTime = (endStream - startStream);

double flops = 2 * Math.pow(size, 3);
double gpuGigaFlops = (1.0E-9 * flops) / (msecGPUElapsedTime / 1000.0f);
double cpuGigaFlops = (1.0E-9 * flops) / (msecCPUElaptedTime / 1000.0f);
double streamGigaFlops = (1.0E-9 * flops) / (msecStreamElaptedTime / 1000.0f);
double gpuGigaFlops = (1.0E-9 * flops) / (nanoSecGPUElapsedTime / 1000000000.0f);
double cpuGigaFlops = (1.0E-9 * flops) / (nanoSecCPUElaptedTime / 1000000000.0f);
double streamGigaFlops = (1.0E-9 * flops) / (nanoSecStreamElaptedTime / 1000000000.0f);
double speedup = (double) (endSequential - startSequential) / (double) (end - start);

String formatGPUFGlops = String.format("%.2f", gpuGigaFlops);
String formatCPUFGlops = String.format("%.2f", cpuGigaFlops);
String formatStreamFGlops = String.format("%.2f", streamGigaFlops);

TornadoDeviceType deviceType = executor.getDevice(0).getDeviceType();
System.out.println("\tSingle Threaded CPU Execution: " + formatCPUFGlops + " GFlops, Total time = " + (endSequential - startSequential) + " ms");
System.out.println("\tStreams Execution: " + formatStreamFGlops + " GFlops, Total time = " + (msecStreamElaptedTime) + " ms");
System.out.println("\tTornadoVM Execution on " + deviceType + " (Accelerated): " + formatGPUFGlops + " GFlops, Total Time = " + (end - start) + " ms");
System.out.println("\tSingle Threaded CPU Execution: " + formatCPUFGlops + " GFlops, Total time = " + (endSequential - startSequential) + " ns");
System.out.println("\tStreams Execution: " + formatStreamFGlops + " GFlops, Total time = " + (nanoSecStreamElaptedTime) + " ns");
System.out.println("\tTornadoVM Execution on " + deviceType + " (Accelerated): " + formatGPUFGlops + " GFlops, Total Time = " + (end - start) + " ns");
System.out.println("\tSpeedup: " + speedup + "x");
System.out.println("\tVerification " + verify(matrixC, resultSeq, size));
}
Expand Down