[Libre-soc-dev] Inverse DCT

Luke Kenneth Casson Leighton lkcl at lkcl.net
Sat Jul 31 17:39:57 BST 2021


---
crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68

On Sat, Jul 31, 2021 at 2:34 PM Luke Kenneth Casson Leighton
<lkcl at lkcl.net> wrote:
>
> On Fri, Jul 30, 2021 at 3:44 PM Luke Kenneth Casson Leighton
> <lkcl at lkcl.net> wrote:
>
> > next step is to link it into instructions and write a simulator unit test
>
> first couple of unit tests successful, for i-DCT inner and outer butterfly.
> next up is the COS table and then putting it all together.
> the COS table has to be created in reverse-butterfly order from DCT.

nope, before that i had to fix the errors in the previous unit tests:
      vec[jh] += vec[jl]  # i-DCT
not
     vec[jl] += vec[jh]   # DCT

and also use FFT twin MUL-ADD-SUB not the specialised DCT MUL-ADD-SUB

now those are corrected, using both of the inner and outer butterfly
schedules produces the expected i-DCT results.

https://git.libre-soc.org/?p=openpower-isa.git;a=blob;f=src/openpower/decoder/isa/test_caller_svp64_dct.py;h=ccdae2367cf536ef061c51ce957543c2606f2eda;hb=623bdfe4a2c3d511a1fb99555001abfc2ab6ddd7

l.



More information about the Libre-soc-dev mailing list