The anomaly of hmac_sha1 performance testing on Android

[Winter]

1down votefavorite

I am testing the performance of cryptographic libs on Android 4.03. But the result shows an anomay like the figure blow:  I gradually increase the input size, the execution time increases in a linear way. But when the size reaches a point (197 * 160B in this figure). The performance suddenly becomes better. The same thing happens in testing AES_CBC. So why it happens in this way?

[Naresh Mehta]

On Tue, Nov 22, 2016 at 4:01 AM, Winter <wang...@gmail.com> wrote:

I gradually increase the input size, the execution time increases in a linear way. But when the size reaches a point (197 * 160B in this figure). The performance suddenly becomes better. The same thing happens in testing AES_CBC. So why it happens in this way?

Can you also do the core load measurements also? My guess is that this is the time when the other core kicks in and some more threads are spun which 

reduces the execution time. Quite interesting observation though.

--
--
unsubscribe: android-kernel+unsubscribe@googlegroups.com
website: http://groups.google.com/group/android-kernel
---
You received this message because you are subscribed to the Google Groups "Android Linux Kernel Development" group.
To unsubscribe from this group and stop receiving emails from it, send an email to android-kernel+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

BR; Naresh

visit me at:
http://www.naresh.se/