Re: [Freetel-codec2] Arch specific vector optimizations for LPCNet?

Hi Richard, As far as I've read, AVX/AVX2 is in all Intel's i3, i5 and i7 processors in 99% of all laptops and desktops today. AMD also. As of last year, we have "aarch64" in affordable small board computers eg. Raspberry Pi 4B. It has even more SIMD features, perhaps even better than AVX2.

Re: [Freetel-codec2] Arch specific vector optimizations for LPCNet?

On Wed, Apr 22, 2020 at 12:54 AM Al Beard wrote: > *Hi Richard,* > > *Is this useful?* > > * > https://indico-jsc.fz-juelich.de/event/105/contribution/5/material/slides/1.pdf > * > It was very interesting but I'm

Re: [Freetel-codec2] Arch specific vector optimizations for LPCNet?

Hi Richard, Is this useful? https://indico-jsc.fz-juelich.de/event/105/contribution/5/material/slides/1.pdf Alan VK2ZIW On Mon, 20 Apr 2020 12:52:25 -0500, Richard Shaw wrote > I ended up fixing it using __aarch64__ instead. > >

Re: [Freetel-codec2] Arch specific vector optimizations for LPCNet?

I ended up fixing it using __aarch64__ instead. https://github.com/drowe67/LPCNet/commit/3c04282bef6df20f93da6295efae3630d1b171c1 Thanks for the tip though! Richard > ___ Freetel-codec2 mailing list Freetel-codec2@lists.sourceforge.net

Re: [Freetel-codec2] Arch specific vector optimizations for LPCNet?

On Mon, Apr 20, 2020 at 11:23 AM Jeroen Vreeken wrote: > On Aarch64 you do not need to check for NEON since it is a mandatory > part. (The 'asimd' flag is always present) > Because of that a number of defines in gcc are no longer present. > I had the same issue with the LPCNet code on an 64bit

Re: [Freetel-codec2] Arch specific vector optimizations for LPCNet?

On Aarch64 you do not need to check for NEON since it is a mandatory part. (The 'asimd' flag is always present) Because of that a number of defines in gcc are no longer present. I had the same issue with the LPCNet code on an 64bit raspberry pi. You need to compile with '-mtune=native

[Freetel-codec2] Arch specific vector optimizations for LPCNet?

So I'm working on packaging LPCNet for Fedora now and I'm having an interesting time figuring out the base optimizations each arch can provide and I currently have them forced as shown below since we can't assume the host system is the same as the build system. x86_64 -> AVX2 i686 -> AVX armv7hl