Weird Code result in different file size and time

[raesoo80]

Hello, I'm totally new to programming.
I have a issue with hard disk performance i/o with my windows 2003 servers
one of my vendor sent me the code below, they said it's testing write speed of hard disk but when I run this code on few of my windows 2003 servers, the actual file size are different.
about 60% of them, it creates file size with 55MB but on rest 40% of servers I have, it creates file size of 5 MB. weird thing is that one that generates 5MB text file, it takes me about 2 minutes in order to finish generating file. but when it generates 55MB file, it takes less than 10 seconds.

all of servers are running windows 2003 standard version on a RAID 1 confuguration. I checked if any of windows updates, windows services, or drivers are causing this issue but I can't find any so that I'm hoping if anyone can give me some advice for what's causing different result.

Here's the code

for (int i = 0; i < 100000 ; i++) { FILE *pFile = fopen( "text.txt" , "at"); fprintf( pFile, "1234567890feoiv jeoifjonfrjobejrgojbomerjgobogorgmborjbo\n" ); fclose(pFile); } FILE *pFile = fopen( "text.txt" , "at"); fprintf( pFile, "VERSION:1:TIME:%d\n" , timeGetTime() - dwPre ); fclose( pFile );

Thank you

[Zaita]

Nothing in the code should attribute to the problem you're seeing.

This code simply opens and closes a file 100,000 times and each time adds a line of text too it. It then finishings by adding a string with the time at the bottom.

[PiMaster]

You could definitely optimize that code more:

//open the file ONCE
FILE* file=fopen("text.txt","at");
for(int i=0;i<100000;++i){//make ++ on the left side - doesn't make a temp obj
	fprintf(file,"1234567890feoiv jeoifjonfrjobejrgojbomerjgobogorgmborjbo\n");}
fprintf(file,"VERSION:1:TIME:%d\n",timeGetTime()-dwPre);
//close the file ONCE
fclose(file);

(BTW, use the [ code ] [ /code ] tags (without spaces) to format your code like this ^^^)

[raesoo80]

Thank you all so much for the advice.
I'll keep in mind to use next time
Thank you all again.

[Zaita]

@PiMaster: The idea of the code is to be not optimized on purpose. That's the point of the benchmark :)

[raesoo80]

Hello PiMaster,
would you mind if I ask you to create filewrite.exe file using the code you modified as above and somehow send it over to me?T_T

[raesoo80]

@ Zaita/ so there's nothing wrong with the code itself?

[Zaita]

@raesoo80: Nope. The code is specifically designed to be inefficient so it can test a higher load on the system. The opening/closing of files 100,000 times is part of the test. Having it do it only once isn't going to be a complete test.

[raesoo80]

@ Zaita/ the purpose of this code is absolutly same as what you said. But any idea what would cause
difference in file size?

[Zaita]

As to my original reply. I see nothing in the code that would allow me to think that. Unless you were running it multiple times. Even variances in block sizes/file systems wouldn't account for such a huge difference. Unless the file system was compressed?