Scientific-SNAP branch. Provides means of leveraging Xilinx ap_fixed …

[fmdepaul]

Scientific-SNAP fork. Provides means for leveraging Xilinx ap_fixed data-type for both sw and hw actions. Added hls_double_mult app based on said data-type .

Signed-off-by: Flavio M. de Paula [email protected]

[fmdepaul]

Found some issues and addressed on on SHA 32c1f8a

[bmesnet]

@fmdepaulf Hi Flavio, You may have missed the information that from the hls_double_mult, I have created a clean release named hls_decimal_mult (and removed the old one). This release takes care of the simple and double floating point and is more parameterized . It may be easier to maintain one unique release. Are your changes applicable to this already registered action? Thx

[fmdepaul]

@bmesnet Hi Bruno, I uploaded this ap_fixed version of hls_double_mult so that folks have available to them two different approaches to deal with single and double floating-point precision numbers. I didn't intend to overwrite hls_decimal_mult. However, given that I have hls_decimal_mult with ap_fixed in my sandbox. I could either upload it with an alternative name or overwrite the existing one. Let me know what you prefer. I would then delete hls_double_mult. To reiterate my point, I think it would be a good idea to have those two approaches as examples since, in fact, they handle float/doubles in different ways. Please, let me know what you think.