the different results got from openssl and ipp

[Chen_Chen]

Hello~
Today, I used openssl and ipp to encryptthe same plaintext by AES256 algorithm with the same key in ECB mode, but returned two different results...
I am puzzled with it...
Is there someone can solve it??
Thanks~

[Vladimir_Dudnik]

Chen,

I'd expect the result may depend on how do you call IPP functions, what parameters you passed to it and how you call OpenSLL.

I do not think there are people on the forum smart enough to give you an answer based on information you provided.

Regards,
Vladimir

[PaulF_IntelCorp]

Hello Chen,

Are you saying that you are comparing the output of OpenSSL with IPP enabled and without IPP? In other words, did you run a test using the standard distribution of OpenSSL and then do it again using the IPP-patched version of OpenSSL?

Paul

[Chen_Chen]

Thank you,paul~
I run a test using the same key to encrypt the same plaintext using the encrypt functions in OpenSSL andusing the encrypt functions in IPP separately, then the results are different.

I do not using the IPP-patched version of OpenSSL, I only want to judge their performanceseparately!

[PaulF_IntelCorp]

Hello Chen,

Since there are so many variables associated with the sort of comparison you are making, it is difficult to know if your comparison is valid.

I suggest that you compare OpenSSL against OpenSSL. That is, make your comparisons using OpenSSL as you get it from the OpenSSL distro site and then again patched to use IPP. If it is faster when using IPP then you have a confirmation that IPP can provide a faster solution.

This article, Boosting OpenSSL AES Encryption with Intel IPP, may help you with that evaluation process. Note that the article measures the performance boost when using a processor that supports the Intel AES-NI instructions. Even if you don't have a processor with those new instructions, you can still use the process they describe to perform a valid comparison on your platform.

Paul