[opus] [RFC V3 7/8] armv7, armv8: Optimize fixed point fft using NE10 library

[Viswanath Puttagunta]

Hello Timothy,

Great to hear from you!

Fired up my hardware today and this issue looks like a regression in
Ne10 library.

The commit in Ne10 [1] that I tested to be working successfully back in May
5b63074db45000f9688460990ee3f5e147d93782
which is the Patch Phil at ARM added to fix the overflow issue in nfft=60 case.

After git-bisect, looks like the culprit patch in Ne10 [1] is
cf33c0d51a445bbe6ad7e21c2af875acee07b838

Phil,
Please fix the regression at the earliest and let me know if you need any help.

Regards,
Vish

[1]: https://github.com/projectNe10/Ne10.git

On 5 October 2015 at 20:21, Timothy B. Terriberry
<tterriberry at mozilla.com> wrote:
>
> I'm trying to get these cleaned up and landed, but I'm running into some trouble with this patch. Using commit a08b29d88e3c (July 21) of Ne10, I'm seeing test failures for 60-point FFTs:
>
> nfft=60 inverse=0,snr = -3.312408
> ** poor snr: -3.312408 **
> nfft=60 inverse=1,snr = -16.079597
> ** poor snr: -16.079597 **
>
> All other sizes tested appear to work fine (84 to 140 dB of SNR). This doesn't match the results you reported on this list in May, where the FFTs appeared to pass tests, but the MDCTs did not (the latter of which I presume was fixed by the May 12 commits to Ne10). I don't see any commits since then that I would expect to have broken this. Any ideas?