• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 80 subscribers
  • Views 1462 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

AES-256 Benchmark

steve williams
Prodigy 50 points
Other Parts Discussed in Thread: AES-128

I am trying to figure out if I can use an MSP-430 as an AES-256 encryption device and need some benchmark information that allows me figure out the bytes/sec throughput possible as a function of clock speed and crypto operating mode. And, with that speed information benchmark, the amount and type of memory needed. Can anyone point me to this kind of data?

  • 0
    Intellectual 965 points

    Instead of using the MSP430 MCU itself, you may want to consider the MSP432 which has the AES-256 on board as a peripheral. Granted the product is not in full production but the MSP432 LaunchPad (using preproduction silicon) is currently available. From what I can recall of the throughput it takes a couple hundred cycles or so to do encryption or decryption, but I am not sure if it is clocked at the full 48Mhz of the MSP432.

  • 0
    Prodigy 50 points
    I found a paper here:

    web.cs.ucdavis.edu/.../gl2012.pdf

    that says for 128 bit code block, in worst case mode (GCM), they got 429 clocks/block using accelerator, and 699 clocks/block without. If these numbers are a good representation than this part is not right for my application (need could orders of magnitude faster).
  • 0 in reply to steve williams
    TI__Guru 56685 points

    Hello Steve,

    Our MSP430CC430 and MSP430F6xxx family of devices have an AES-128 accelerator on chip. Also our MSP430FR59xx/69xx and MSP432 family of devices have an AES-256 accelerator. these accelerators are byte implementations. this means for encryption it requires the following cycles:  (# of rounds) *16 +16. For AES-128 this means 176 and for AES-256 this means 240. For decryption, you also have to compute the key first and then add the corresponding formula above.

    This of course does not account for different chaining modes, and unfortunately I do not have data on that currently. The paper you link to seems like a good source of independent testing though. The "worst-case" you mention is just one particular chaining mode. If your end application does not need this mode, then that figure will not apply.

    Regards,

    JH

  • 0 in reply to Jace H
    TI__Guru 56685 points
    All,

    I was able to find some additional information about our AES-256 accelerator. Please see the link below on pages 368-9.

    www.ti.com/.../slau367g.pdf

    Regards,
    JH

**Attention** This is a public forum