>>> There are some fairly severe performance hits in engine support unless the 
>>> engine includes all the submodes as well.
>>> That includes things you are just starting to play with now, like the 
>>> combined 
>>> AES+SHA1 on x86.
>> ??? Here is output for 'speed -engine intel-accel -evp
>> aes-128-cbc-hmac-sha1' for 1.0.0d, i.e. through engine.
> 
> It depends what the engine is and how the device backing it works.  If
> anything has much latency -- extra trips through the interface can be
> a killer.
> 
> Engines using unprivileged instructions -- effectively, special software
> engines -- are really just a few extra layers of function call indirection.
> It's not surprising extra trips are cheap.  But engines with real hardware,
> even where evil hacks are used to map the hardware into userspace, don't
> tolerate this well.

Argument is specifically about unprivileged instructions. Peter agues
that engine should be avoided and I argue that additional [small till
proven otherwise] overhead is reasonable price to pay for easier
maintenance and faster adoption.
______________________________________________________________________
OpenSSL Project                                 http://www.openssl.org
Development Mailing List                       [email protected]
Automated List Manager                           [email protected]

Reply via email to