Topic: Optimization flags drawbacks

Hello, we noticed that by enabling the following flags “--enable-sp --enable-sp-asm” in wolfSSL library, we do see significant improvement in system bootup time and CPU usage.

However, since those flags do not come enabled by default, we would like to understand what are the potential drawbacks. Specifically, what are the cases where enabling those flags is not recommended?



Re: Optimization flags drawbacks

Hi David,
Thanks for reaching out!! The meaning of these flags are well documented here: … able-spopt . Please have a look.

Note that `--enable-sp-asm` is enabled by default for the following architectures: x86_64/aarch64/amd64.

Here are some other Single-Precision flags that you might be interested in:

  --enable-sp             Enable Single Precision maths implementation
                          (default: disabled)
  --enable-sp-math-all    Enable Single Precision math implementation for full
                          algorithm suite (default: enabled)
  --enable-sp-math        Enable Single Precision math implementation with
                          restricted algorithm suite (default: disabled)
  --enable-sp-asm         Enable Single Precision assembly implementation
                          (default: enabled on x86_64/aarch64/amd64)

All of them are documented in the document I linked to above.
I hope this helps!
Warm regards, Anthony