Benchmark Results:

The benchmark is to run TooB NAM with an A1 model to generate 100s of audio. Lower times are better.

TooB NAM has a compile-time switch that uses either the NeuralAudio NAM implementation or
the Neural Modeler Core NAM implementation.

Results

amd64 (AMD Ryzen 7 7735HS, Ubuntu 24.04 amd64)

    NeuralAudio x64:      5.416s
    Neural Modeler Core:  7.965s

    Compiler options /Ofast, MMX, SSE, SSE2, no AVX.

aarch64 (Raspberry PI 4, ARM Cortex A72, Raspberry PI OS Bookworm)
     Neural Modeler Core: 92.677s
     Neural Modeler Core: 31.589s

     Compiler options: -Ofast -march=-mcpu=cortex-a72 -mtune=cortex-a72  -DEIGEN_DEFAULT_L1_CACHE_SIZE=32768

aarch64 (Raspberry PI 5, ARM Cortex A76, Rasperry PI OS Trixie)

     Compiler options: -mcpu=cortex-a76 -mtune=cortex-a76 -DEIGEN_DEFAULT_L1_CACHE_SIZE=65536

The Model: 

Package: Fender Deluxe Reverb %2765 Reissue %7c Clean %7c SM57 %2b Royer R-121 %2b Room
Model: Fender DRRI %7c Clean %7c Room Only %7c Full Rig
Downloaded from Tone3000.

The model file is attached.


Compilation flags:

    




summaries for Mono audio adapaters.
Check PWA on windows, edge and chrome.
Make sure Pi 4 default audio device is still hidden when selecting devices!

A final solution for rate-limiting on Tone3000.

A way to force input to mono?


Tooltip on File Property Select: add to release notes.
sAFEfILEnAMES: RENAME, &C. Safe bank names? 
NEW BANK
WINDOW SIZE CALCUALTIONS ARE WRONG ON LINUX HIDEF DISPLAYS.

Carla Project icons.
check reload after change of LV2 plugins.

Exactly one underrun per seek in Toob Player


json nan/inf is fatal.

- pipewire aux in?



pcm.pipedal_aux_in {
    type file
    file "/tmp/aux_input_fifo"
    infile "/tmp/aux_input_fifo"
    format "raw"
    slave {
        pcm null
        format S16_LE
        rate 48000
        channels 2
    }
}