ADDITIONAL SYSTEM INFORMATION :
Windows 10

A DESCRIPTION OF THE PROBLEM :
I just recognized that
java.io.File::exists is factor ~ 1.4 slower then the eclipe.org native equivalent org.eclipse.core.internal.filesystem.local.LocalFileNatives.internalGetFileInfoW(char[], IFileInfo).exists()
Both provide the same information but the internal systemcalls to windows differ: While JDK opens the file itself eclipse does open the directory, (as reported by sysinternals ProcessMonitor).
I dont know why that is faster (antivirus detection may be?)
Maybe that may be an improvement to the JDK.

STEPS TO FOLLOW TO REPRODUCE THE PROBLEM :
benchmark
org.eclipse.core.internal.filesystem.local.LocalFileNativesManager
.fetchFileInfo(FILENAME).exists()
vs
new File(FILENAME).exists()

EXPECTED VERSUS ACTUAL BEHAVIOR :
EXPECTED -
JDK not slower then eclipse
ACTUAL -
JDK slower then eclipse

---------- BEGIN SOURCE ----------
package org.eclipse.core.tests.filesystem;

import java.io.File;
import org.eclipse.core.filesystem.provider.FileInfo;


public class Test extends FileSystemTest {
private static final String FILENAME = "C:\\Users\\JKubitz\\eclipseDev\\jdt-master2\\git\\eclipse.jdt.core_copy\\org.eclipse.jdt.core\\dom\\org\\eclipse\\jdt\\core\\dom\\Annotation.java";

public void test() throws InterruptedException {
time(this::extracted);
time(this::extracted2);
time(this::extracted);
time(this::extracted2);
}

private void extracted2() {
File file = new File(FILENAME);
file.exists();
}

private void extracted() {
FileInfo info = org.eclipse.core.internal.filesystem.local.LocalFileNativesManager
.fetchFileInfo(FILENAME);
info.exists();
}

void time(Runnable run) {
long n0 = System.nanoTime();
for (long i = 1; i < 10000; i++) {
run.run();
}
long n1 = System.nanoTime();
long diff = n1 - n0;
System.out.println("Took: " + (diff / 1000_000) + "ms =" + (diff / 1000_000_000) + "s");
}
}

---------- END SOURCE ----------

FREQUENCY : always