Hashing a file with different hashes in one go

[Arokh]

Hi,

 

I'm currently writing a application which calculates different hashes.

Every hash I "have" has inheritet the HashAlgorithm class and now

I want to use them to calculate those hashes.

 

And I want it to be calculated as fast as possible.

At first I thought it wouldn't be so hard to do so,

but I guess I was wrong.

 

Since I have 4 cores I wanted to make 4 threads.

Each thread would be responsible for calculating a hash.

 

Then I have a main loop which reads from the file,

then "passes" the data onto the threads,

and tell it to calculate the next part of the hash.

 

But I keep on having something like synchronisation problems.

And everytime I try to rewrite it I end up with something that doesn't work,

it is starting to frustrate me :-\ .

 

Is there a basic "guideline" I can use?

 

I'm currently looking into a circular buffer.

If I fail again I'll post the code of it here,

so you can give some advice if you want.

[Arokh]

I think I almost got it with the Circular Buffer idea based on:

http://en.wikipedia.org/wiki/Circular_buffer#Absolute_indices

 

But somehow everytime it appends data from previous streamreads,

and if I try to fix it I get an endless loop.

 

Here is the code I'm using so you know what I'm talking about:

http://pastebin.com/f690fd7b2 (CSharp code tag seems to be a little broken)

 

To start it, copy paste of the code should do.

 

The testcode which shows my problem (Program class),

creates a testfile ("Test.txt") filled with a pattern.

Then 4 readers which read from the circular buffer put everything they get into a new file "Test{ReaderIndex}.txt".

 

Now the resulting files should be exactly the same, but they are 32kb bigger.

(2 Blocks in the Circular Buffer)

 

Has anybody an idea what I'm doing wrong?

[Arokh]

I think I've got it to work.

Hopefully there aren't any bugs left which only happen by chance.

 

If anyones interested:

http://pastebin.com/f5fb69518